In the twelve months to 9 August 2026, Minsk pardoned 425 prisoners in three deals brokered by Washington. The groups included foreign citizens as well as people on Viasna’s list of political prisoners. In the same twelve months Viasna recognised 538 new ones. We reported both numbers in “The Offer in the Same Speech” and left them side by side. A book published last winter gives a way to read them together.

The book is Making a Career in Dictatorship by Christian Gläßel and Adam Scharpf (Oxford University Press, 2025). Belarus is not one of its cases: the main evidence is the personnel files of the Argentine army, with shorter studies of the SS, Stalin’s NKVD and the Gambian army. We use the book as a peg. The material below is our own counted series, the test they allow is a limited one, and we say where it stops.

The argument, briefly

Security organisations are pyramids in which many officers enter and few reach the top, and those who fall behind usually know it early. The authors call this career pressure, and they describe two extreme ways out of it. An officer can join a coup, or he can volunteer for the work that others avoid and hope that loyalty will be rewarded.

In Argentina the second route ran through Battalion 601, the intelligence unit at the centre of the “Dirty War”. Officers with poor grades from the military academy were more likely to join it, and those who joined stayed in the service longer than their peers.

The second half of the book turns to the ruler, because if dirty work keeps weak officers loyal, then the amount of dirty work matters to him. The authors test this on all authoritarian regimes since 1945. Coup risk is lowest when repression is stable. It rises after large increases and after large cuts. A ruler who understands this, they write, will “invent threats, expand repressive institutions, and delegate violent tasks” so that his officers stay busy.

A stock and a flow

Read this way, the prisoner figures hold two different quantities. One is a stock, the number of people held on a given day: Viasna counted 1,135 on 23 December 2025, then 911 at the end of March after 250 people were freed, and 879 on 9 August. The other is a flow, the number of new cases. Viasna publishes it every month, and over these twelve months it averaged about 45. It never stopped: the lowest month was 18, in February 2026, and the highest was 78, in October 2025. Viasna’s review for December 2025 draws the same contrast for the whole calendar year. Up to 360 political prisoners were pardoned in 2025, and at least 509 people were newly recognised.

The stock is what Minsk trades, and each release had a price that was named in public. On 11 September 2025, when 52 people were pardoned, Washington lifted its sanctions on Belavia. On 13 December, with 123 pardons, it began to lift sanctions on potash. On 19 March 2026, the day 250 people were pardoned, the US envoy John Coale announced that Belaruskali and the Belarusian Potash Company were leaving all US sanctions lists, together with two banks and the Finance Ministry. In a year the stock fell by a little more than a fifth. Pardons were one of three ways off the list. People also finish their sentences, and in April 2026 Viasna reviewed the list and moved about 100 people to the status of former political prisoner.

Coup risk is lowest when repression is stable. It rises after large increases and after large cuts.

The flow is a different kind of number. No published term of any deal covers it, and it entered the talks only at the third deal. After the March release Coale said this was the first time he had raised new arrests with Lukashenko. “We free ten people and they arrest ten more,” he said, and he hoped Lukashenko had understood that this was unacceptable, Zerkalo reported. In the same remarks he said that everyone would be freed by the end of the year. The flow belongs to the people who produce it: investigators, GUBOPiK officers, prosecutors and judges. The book’s claim is that a ruler cuts this number at his own risk. The Belarusian series shows a ruler who has reduced the stock and left the flow alone.

The two highest months came straight after the first deal. Fifty-two people were pardoned on 11 September 2025. Viasna added 77 people to its list in September and 78 in October. Recognition follows arrest with a delay, so this does not prove a response. It does show that the first release did not slow the intake.

After the March release, and after Coale’s remark, the monthly figure ran lower: 36, 32, 43 and 39 from April to July. That is 150 people in four months, and a mean of 37.5 against 49.1 in the eight months before. Viasna warns against reading that as relief. Its review for May 2026 says that fewer recognitions do not mean less repression, because information on new cases has become harder to obtain. The stock points the same way. It stood at 837 at the start of May, at 873 a month later and at 880 at the end of July. A list can grow only if more people enter it than leave.

Output that does not follow the threat

The usual account of repression starts from a threat: people protest and the state answers. On that account the Belarusian numbers should have fallen by now. The last large street marches were in the autumn of 2020. Registered public associations went from 3,021 in January 2021 to 1,337 this year, as we showed in “The Cat Without the Grin”. Independent newsrooms work from abroad.

The Ministry of Information’s register of extremist materials moved the other way. Courts made 19 designations in 2020, then 427 in 2021, 639 in 2022 and 888 in 2023. The figure was 1,459 in 2024 and 1,381 in 2025. By 3 September this year it was 1,043, which is a pace of about 1,550 for the full year.

The number of courts matters more than the number of rulings. Seven courts issued designations in 2020. In 2021 there were 85, and in every year since there have been between 90 and 112. Sixty-two courts made their first designation in 2021 and another 34 in 2022. For each of them this was a new line of work, and it came with a public count attached. Viasna describes the process in similar terms. Its December review calls the designation of materials mechanical and conveyor-like, and says that new entries now mainly keep up the statistics of the “fight against extremism”.

The stock is what Minsk trades. No published term of any deal covers the flow.

The cases themselves often come from material the state already holds. In the second quarter of 2026, Human Constanta counted 297 people added to the “extremist” lists. Of these, 227 came from one case, the Belaruski Hajun chat bot, which closed in February 2025 after it was hacked. A seized database can supply a quarter’s work, and no new protest is needed for that.

The bench at twenty times the median

The book’s chapter on the NKVD in 1937 describes regional offices with wide freedom over whom to arrest, and numerical targets set by the centre. The authors call the result a perverse incentive system, one that officers with weak records used in order to stand out.

We have no evidence of quotas in Belarus and we do not claim any. What the register shows is how unevenly the work is spread. All 118 raions have at least one designation, and the median raion has 12. The single court in Maladechna has 236, the highest figure per bench in the country, as we found in “The Busiest Bench”. Neither population nor administrative rank explains a bench that works at twenty times the median.

In August we could only describe that gap, and the book turns it into a question about people. Who sits on that bench, and where were their careers in 2020? Belarusian judges are appointed and reappointed by Lukashenko’s decree, so the pressure the authors describe has an obvious channel. We have not yet looked.

For 96 courts this was a new line of work, and it came with a public count attached.

Where the fit stops

Belarus has had no coup attempt from inside the security services, so half of the book’s mechanism cannot be seen here. In 2020 the officers who disagreed left the service, and some of them formed BYPOL. The authors advise outside actors to make that kind of exit easy, and Minsk made it costly. On 4 May 2021 Lukashenko signed decree No. 174, which stripped more than 80 former officers of their ranks. Outlets that counted the published list found 86 or 87 names. The Interior Ministry listed BYPOL as an extremist formation on 8 November 2021, and the Supreme Court declared it a terrorist organisation in August 2022. The group split in 2023, and most of its staff now work under the name BELPOL.

A larger limit concerns the evidence, because our series fit the ordinary explanation too. The state may simply see an exile Instagram account as a threat and designate it for that reason. A steady policy would produce a steady flow, and so would an apparatus that needs work. The two readings separate only when the stock moves sharply, and that is happening now. If new cases fall for two quarters after a large release, the ordinary reading is the better one.

The book rests on personnel files and this piece rests on output counts, which is weaker evidence. Part two will use the book’s own kind of data. The EU listed 84 Belarusian officials between October and December 2020, and most of them served in law enforcement or the courts. We are coding where each of them stood in August 2020 and where each stands now. We will also count how many have ever been removed from the list. The authors argue that sanctions change behaviour only when they offer a way out.

If new cases fall for two quarters after a large release, the ordinary reading is the better one.

What to take from this

• Minsk reduced the number of political prisoners by about a fifth in a year. Viasna added new ones every month: 45 on average, never fewer than 18.

• The two highest months, 77 and 78, followed the first US-brokered release. Since May 2026 the total has been rising again.

• The extremist register grew from 19 designations in 2020 to 1,459 in 2024, while organised protest inside the country disappeared.

• In 2021 and 2022, 96 courts began this work for the first time. Between 90 and 112 courts have done it every year since.

• The book’s explanation is possible, and our data cannot yet confirm it. The confirming evidence would be careers, and that is part two.

Watch items

9 August 2027. Viasna’s count of new political prisoners for the twelve months before that date. If the book’s reading holds, the figure stays above 450 even if another release of 200 or more takes place. A figure below 300 counts against it.

31 December 2026. Coale’s own deadline: “by the end of the year we will free everyone.” Viasna’s list stood at 880 at the end of July. We expect more than 600 names on it on 1 January 2027. If the list is emptied, the test moves to the intake: at 45 a month it would hold about 270 names again by mid-2027.

31 March 2027. The stock, read together with the monthly figure. Viasna says its count of new cases is losing coverage, so a low monthly figure alone proves little. The stock is the check: sentences end every month, and the list shrinks unless new people enter it. If the stock is under 800 on this date and no new release deal has taken place, the intake has really fallen and the ordinary reading gains. If it is at 880 or above, the intake has held.

15 January 2027. The register’s total for 2026. We expect between 1,450 and 1,650 designations. A total under 1,200 would mean the pace broke in the autumn.

The next release deal. Whether any published term of it concerns new cases. So far none has.

Method and limits

The release dates, the sanctions steps, decree No. 174, the two BYPOL designations and the three EU listing rounds of 2020 (40, 15 and 29 people) were checked against dated reports in our corpus before publication. Coale’s remarks are given as Zerkalo and Radio Svaboda carried them on 20 March 2026, in Russian and Belarusian. The English here is our translation of those versions, and we have not found the original wording.

Prisoner counts are Viasna’s, read from its Telegram channel by a parser and checked by hand against named posts. Viasna’s list is a recognition list: it lags behind arrests, and people leave it by finishing their sentences as well as by pardon. Monthly figures are Viasna’s own. They come from two recurring posts on its Telegram channel, the list of people recognised in the month and the monthly human rights review, and from the same reviews on spring96.org. Where the two posts differ (February 2025: 33 and 39, April 2025: 33 and 41) we use the review, which matches Viasna’s half-year total of 183. December 2025, 50, comes from the review on spring96.org. Two months are derived. For January 2025 we use 26, which is the half-year total less the other five months; the Telegram post of 28 January said 18. For August 2025 we use 40, which is Viasna’s total of 509 for 2025 less the other eleven months; the post of 27 August said 37 so far. The twelve calendar months from August 2025 to July 2026 add up to 543. The published figure of 538 covers 10 August to 9 August, so the two agree to within one per cent.

We first tried to build this series by adding up single recognition statements from all outlets. The result was 1,948 against a known 538, because recaps, pardons and lists of former prisoners were counted as new cases. We discarded it.

Register figures come from our crawl of the Ministry of Information’s list: 6,018 positions by 159 courts, with ruling dates from 4 September 2008 to 3 September 2026. Counts are by ruling date and leave out the six annulled positions, which is why 2024 stands at 1,459 here and at 1,462 in “Written in Pencil”. “Courts” means courts that appear in the register in a given year. It does not measure judges or hearings.

The U-shaped curve is drawn after Figure 7.5 of the book and carries no data. The book’s cross-national test concerns coups, and this piece makes no forecast about one. The sentence on judicial appointments describes the formal procedure. We hold no career data on any judge and name none.

The lists of “extremist formations” and of persons kept by the Interior Ministry are separate instruments from the register of materials. The Human Constanta figure refers to those lists.

The Centre for New Ideas has published Anatomy of Belarusian Power, a network study of 271 members of the ruling elite by Yury Drakakhrust and Anton Radniankou. It is the first map of this size built with network methods, and it is careful about its limits. Its strongest finding is simple: the people who connect the system are often not the people at the top of the official list. The findings most likely to be quoted are weaker. The claim that the system is hard to break, and the claim that Russian influence is narrow, both lean on choices in the method that the report does not test. This brief separates the two.

How the map was drawn

The authors collected public biographies of officials from deputy minister level upward, heads of regions, security chiefs, managers of large state firms and a few people close to Lukashenko without a formal post. Lukashenko himself is left out on purpose. The study looks at ties between members of the elite and sets access to him aside.

Two people are linked if they worked in the same institution in overlapping years, or studied at the same university at the same time. Each shared year adds weight to the link. The result is 3,457 ties among 271 people. An algorithm then sorts them into groups, and three standard measures describe each person’s place: how many ties they have, how close they are to everyone else, and how often they sit on the shortest path between two other people. The last one, called betweenness, is the measure of a broker.

The people who connect the system are often not the people at the top of the official list.

The report finds a large administrative machine of about 200 people and a smaller security apparatus of 59. The security side splits in two. The army and the Interior Ministry form one closed world, while the KGB sends more of its people into civilian posts. That picture is plausible and well argued.

Where rank and position part ways

The most useful part of the study is a list of 17 “super-brokers”. They are 6% of the elite, but 63% of all shortest paths between other members run through at least one of them. The list does not match the table of ranks.

KGB chairman Ivan Tertel has only 38 ties, 67th in the network, yet he is the third-strongest broker. His few ties connect parts of the system that are otherwise apart. Prime Minister Alexander Turchin is the opposite case. He is sixth by number of ties and only 33rd as a broker, because most of his contacts already know each other. Two very senior people are missing from the list altogether. Natalya Kochanova chairs the upper house and would take over if the post of head of state fell vacant. She ranks 85th as a broker. Security Council secretary Alexander Volfovich ranks 42nd. Instead the list includes two deputy heads of the Minsk regional executive committee and a deputy foreign minister.

This does not mean Kochanova has less power than a regional deputy. The authors say so themselves. What the list shows is who has shared a career with many separate groups. For anyone planning contacts inside the system, that is a better guide than job titles.

Volfovich shows the limit from the other side. On 23 July, according to BIPART’s July–August digest, Lukashenko gave the State Secretariat of the Security Council, which Volfovich heads, oversight of KGB organisational and personnel matters. His network rank stayed where it was. His authority over the third-strongest broker grew. Supervision of this kind runs from one post to another by decision. A map of shared careers does not record it.

Sturdy partly by design

The report’s policy advice rests on one simulation. The authors remove the strongest broker, recalculate, remove the next one, and so on. The network holds together for a long time. Removing 20 people shrinks the largest connected part by only 9%. Up to a fifth of the network, it hardly matters whether people are removed in order of importance or at random. The same share stays in one piece either way. Only the paths get longer.

Part of this sturdiness comes from how the ties were built. When everyone who passed through one institution in the same years is linked to everyone else, each institution becomes a tight knot. The Academy of Public Administration alone produces 1,043 ties, 30% of the total. The Presidential Administration produces another 932. A tight knot has many spare paths by definition, so a network built this way will resist the removal of single people almost regardless of what happens in real politics. The authors show that removing both institutions still leaves 246 of 260 people connected. That is a good test, but it removes whole institutions and keeps the knots from all the others, such as the Belarusian State University with 278 ties.

The Academy case matters most. It runs many courses and retraining programmes for different levels of officials. Two people enrolled there in the same year may never have met. The usual fix in network research is to give ties from large institutions less weight, so that a shared ministry of 40 senior people counts for more than a shared university. The report does not try this, and we would expect the “super-broker” list to be more stable under that change than the robustness curve is.

A tight knot has many spare paths by definition, so a network built this way will resist the removal of single people.

One more result needs care. The report describes how people who were almost invisible become key brokers once others are removed. Yury Tertel, head of the state wildlife inspection, rose from 155th place to a betweenness of 0.138, higher than the original leader’s best value. But betweenness is measured as a share of paths in the network that exists at that moment. After 65 removals the network is smaller and already breaking, so the same number means something different. The idea that the broker role passes to unknown people is reasonable. This comparison does not prove it.

The Russian trace, counted again

The report counts 55 people, one in five, with a Russian trace: study in Russia, work there, or birth there. It finds that almost none of the ties between them were formed in Russia. People with a Russian past met each other at home, in Belarusian institutions. This is a clever use of the data, because each tie carries the place where it was formed.

The report then makes two further claims. The trace is twice as common at the top, 7 of the 17 brokers against 20% overall. And it is only average in the dense core of 65 people, which the authors read as proof that Russia is not built into the structure.

Both numbers change when they are read against group membership. The trace is common in the security groups, 38 to 39%, and rare in the government group, 9%. The 17 brokers include five people from the security groups. If each broker is given the rate of their own group, we would expect about four with a Russian trace, not seven. Seven is more than that, but with 17 people the gap is close to chance: a simple exact test gives p = 0.05. The core runs the other way from the report’s reading. It is 95% civilian, so the expected share is about 16%, and the observed 20% is above it. By the report’s own logic, the trace is slightly over-represented in the core too.

There is also a small problem of consistency. The text says the 214 ties inside the Russian-trace group formed 171 in Belarusian, 41 in Soviet and 2 in Russian institutions. The chart on the same page shows 170, 38 and 6. Other numbers differ between text and tables too. The share explained by the main factor is 71.1% in one place, 72.7% in another and 73% in a third. None of these changes the story. Together they suggest that parts of the report come from different runs of a random algorithm, and the dataset is not published to check.

The 17 brokers include five people from the security groups, and that alone explains much of the Russian share at the top.

What the graph cannot hold

The method sees shared jobs and shared schools. It cannot see ownership, contracts, family or friendship. So some findings follow from the method rather than from Belarus. Big business owners such as Nikolai Vorobei or Alexander Shakutin land at the edge of the network because their ties to the state run through ownership and contracts, which the method does not record. Lukashenko’s sons are spread across groups for a similar reason. The report says business and family “do not form their own group”, but the data could not show such a group even if it existed.

The same is true of the weak seam between civilian and security officials. The authors write that biographies of security officials are thinner, especially for training in Russian military academies. A seam that looks narrow may partly be a gap in the files. Fourteen members of the security apparatus have no ties at all outside their own group. Some of them may simply have short public biographies.

The report’s own best example of a hidden broker shows the same problem. Yury Tertel, who climbs from 155th place in the removal test, heads the State Inspection for the Protection of Flora and Fauna. The report asks readers not to confuse him with KGB chairman Ivan Tertel. The BIPART digest describes the two men as brothers. It also reports a plan this summer to change whom the inspectorate answers to, which the Security Council’s secretariat did not back. The method sees Yury Tertel’s career in the border guard. It cannot see the family.

Confidence

Watch items

The report predicts that the core survives personnel changes. If it is right, at least 15 of the 17 brokers will still hold a post inside the study’s scope on 30 June 2027.

It also predicts that departures are absorbed from inside the system. If any of the 17 leaves before 30 June 2027, the successor should come from the dense core or the second tier. A newcomer from outside the 271 would count against the claim.

The data could not show a business group even if one existed.

If CNI publishes its list of people and institutions, a recount that gives large institutions less weight should keep at least 12 of the 17 brokers. If fewer than ten survive, the list describes the Academy’s enrolment years more than it describes power.

Method and limits

This brief uses only the numbers published in the report. We did not have the underlying data and did not rebuild the network. The expected Russian-trace figures multiply each person’s group by that group’s published rate, using the group counts in tables B2.4, B3.1 and B4.1. The exact test compares 7 of 17 with 48 of the other 254 people. Our earlier reviews of CNI work, on transit monitoring and on the Telegram propaganda map, used the same approach of recounting from the authors’ own tables.

September 9, 2026 a Belarusian court sentenced Dmitry Laieuski, once the lawyer for Viktar Babaryka, to three years for giving interviews to independent media. He had lost the right to practise years ago — not by a ministerial order, but by a vote of his own bar association’s disciplinary body. That route accounts for a third of all the lawyers pushed out of the profession since 2020. The ministry itself withdrew only nineteen licences.

Three doors, one outcome

The Right to Defence project keeps a named register of persecuted Belarusian lawyers, updated on 24 June 2026. It separates them by the mechanism used, and the separation is the story.

Eighty lawyers, 53% of the total, were found unfit at an assessment by the qualification commission attached to the Ministry of Justice. Fifty-one, 34%, were expelled by the disciplinary bodies of their own regional bar associations. Nineteen, 13%, had their licences withdrawn directly by the ministry for “conduct incompatible with the title of advocate”.

Together that is 150 people who can no longer work as lawyers. A further sixteen face criminal prosecution, and the register counts 155 individuals in all.

The attestation route deserves a note. It used to be a routine check every five years. The rules now allow the commission to summon any lawyer at any time, outside the schedule. A procedure designed to confirm competence became the main instrument for removing people.

The ministry signed nineteen orders. The other 131 left by other doors.

What the overlaps show

Eleven names appear on two of the register’s lists, and every one of those overlaps is the same pair: a lawyer facing criminal prosecution who was also expelled by the disciplinary body of their own bar.

Of the sixteen lawyers under criminal prosecution, eleven were expelled by colleagues. None of the sixteen lost a licence by ministerial order.

That pattern is worth stating carefully. It does not show that bar associations acted freely. Belarusian law has steadily placed the collegiums and their disciplinary bodies under the direct supervision of the Ministry of Justice, and refusing to expel a colleague has itself cost lawyers their own status: one was removed after declining to vote out a colleague who took political cases.

What it does show is where the decisions were formally taken. On paper, the profession removed its own.

On paper, the profession removed its own.

The scale, against the last time

After the 2010 election, human rights monitors linked six licence withdrawals to political persecution. The current count is 150. Beyond them, Right to Defence estimates that roughly 300 more lawyers left the profession themselves under threat of prosecution, which is a figure of a different kind and not part of the named register.

The shape is familiar from our other work this month. More than 2,000 organisations closed, most of them lately by their own decision. Seven manor houses struck from a heritage register rather than demolished. Sentences extended inside colonies without a new crime. In each case the state avoids the act that would need a signature and a defendant.

A procedure meant to confirm competence became the main way of removing people.

One more thing about the register

The website that holds this register has itself been declared extremist material in Belarus. Its front page carries instructions on how to visit it safely.

So the document recording how lawyers were removed from the profession is illegal to read in the country where it happened. Anyone inside Belarus who opens it to check a name is committing an offence, and the lawyer they might have consulted about that is on the list.

The record of how lawyers were removed is illegal to read where it happened.

Confidence

Watch

• The register passes 160 named lawyers before the end of the year. It stood at 155 in June.

• Any lawyer removed by attestation is restored after an appeal. We know of none.

• A regional bar association publicly refuses an expulsion demanded of it.

• Laieuski’s sentence is reduced or overturned on appeal, or he joins a future release.

Method and limits

All counts come from the Right to Defence register as it stood on 24 June 2026: sixteen under criminal prosecution, nineteen stripped of a licence by the ministry, eighty removed through attestation, fifty-one expelled by their bar. We counted the unique names ourselves and found 155 individuals across the four lists, with eleven appearing twice.

The register is a project’s compilation from open sources and submissions, and it says so. It is a floor, not a census: its own page asks readers to write in about anyone missed.

The figure of roughly 300 lawyers who left the profession without a formal removal is the project’s estimate and is not a named list. We keep it separate from the 150 throughout.

We do not know what any individual disciplinary body was told or threatened with before it voted. The register records the formal route, and this piece follows it. That is a statement about procedure, not about the people who sat on those commissions.

Last week a headline said Belarusians had handed in almost 1.5 billion dollars of foreign currency in a single month. The real figure is 274 million. The 1.5 billion is the gross amount households sold before subtracting what they bought, and the correct number sat two paragraphs below the headline in the same article. The difference is more than fivefold, and it points the story the wrong way.

Three numbers, one release

The National Bank publishes household currency operations every month. For August 2026, as reported by BelTA from the bank’s press service, households bought 1.183 billion dollars and sold 1.457 billion. The difference — what they gave up on balance — is 274 million.

All three figures appear in the release. The headline took the second one.

This matters because gross and net answer different questions. Gross turnover tells you how busy the exchange windows were. Net tells you whether households are giving up currency or accumulating it. A country where people change money often can have a large gross figure and no net movement at all.

The headline took the gross figure. The right one was two paragraphs down.

What the correct figure shows

Households have been net sellers all year. Over January to August the running total is 1.61 billion dollars sold more than bought.

We wrote about this run on 13 August, when the July figure was 337.5 million and the seven-month total 1.3 billion. August adds 274 million, and the two totals reconcile.

What we cannot say is whether August was weak. The monthly series over the past two years runs from about 45 million to over 600 million, a spread of more than ten times. Against that, a move from 337 to 274 is inside the ordinary variation, and calling it a slowdown would repeat the original error in the opposite direction.

Who is on the other side

The more interesting part of the release is the comparison the headlines skipped.

Over the same eight months, resident companies were net buyers of 895 million dollars. A year earlier, over the same period, they bought 1.439 billion on balance. Corporate demand for foreign currency is running about 38% below last year.

So the picture is households selling steadily while the companies that would normally absorb that currency need less of it than before. Total market turnover, meanwhile, reached 101.6 billion dollars over eight months, some 21% above the same period of 2025.

The National Bank’s own explanation for the household side is rates: deposits in Belarusian rubles pay enough that people close currency accounts and move to the national currency. That is a reasonable account of a decision, and it does not require anyone to be in difficulty.

Households sell steadily. The companies that would absorb it need less than last year.

Calling August a slowdown would repeat the same error in the other direction.

Why the mistake is worth naming

A fivefold error in a widely shared figure would matter on its own. It matters more here because the two readings support opposite stories.

At 1.5 billion a month, households are dumping currency at a pace that suggests something urgent. At 274 million, they are doing what they have done every month for over a year, at the smaller end of the usual range. The first reading invites alarm; the second invites a question about deposit rates.

Neither reading is available from the headline figure alone, which is the argument for checking gross against net before repeating a number. The release contained everything needed. Nobody had to find a source.

The release contained everything needed. Nobody had to find a source.

Confidence

Watch

• The September release, due in early October, puts the monthly net figure inside the 45 to 640 million range that has held since 2024.

• Corporate net purchases for the full year come in below 2025’s level. Through August they are 38% lower.

• Household net sales for 2026 exceed the 1.61 billion already recorded by the end of August.

• The National Bank publishes deposit-rate data alongside the currency release, which would let its own explanation be tested.

Method and limits

The August figures are the National Bank’s, as reported by BelTA from its press service on 8 September. We could not open the bank’s own statistical tables: its site refuses automated access, and so does the financial portal we tried. So the primary document here is a press-service account, not the table it summarises.

Our monthly range of 45 to 640 million is reconstructed from monthly reports in our own corpus, not from the bank’s series. Those reports carry three different figures in the same paragraph — household net, company net, and the running annual total — and separating them takes hand-checking, which we have not completed. The range is therefore indicative, and the piece uses it only to say that August is unremarkable, not to place it precisely.

On 14 September 2026 Defence24 published We Can’t Afford to Lose Belarus, a 15-page report based on a closed panel held under the Chatham House rule on 6 May. More than twenty experts on Belarus took part. Their shared conclusion is that the European Union needs its own long-term strategy towards Minsk, and that doing nothing has a cost.

The report ends with a short update on events after the panel. It says no real normalisation followed, and migration pressure moved from the Polish border to the Latvian one. It also says Washington reduced its activity on Belarus to focus on Iran and the Middle East. On 15 September, the day after publication, the US envoy John Coale was in Minsk. It was his first visit in six months, and it was no surprise. By 30 August he had said in public that he planned to travel to Belarus in mid-September to free more prisoners. Belarusian media had the exact date by 11 September. On the day the report appeared, Coale was in Vilnius, meeting Lithuania’s foreign minister.

The talks in Minsk ended on 16 September with what Coale called an interim agreement. As a goodwill gesture, Belarus will free 25 people, and the US will lift sanctions on two state companies, the paint maker Lakokraska and the forestry concern Bellesbumprom. Coale added that more releases and more delistings would follow once the remaining issues are agreed. Both statements first appeared on Pul Pervogo, a Telegram channel close to Lukashenko.

A gap of 131 days between a discussion and its written record is normal for this kind of paper. The update paragraph is the weak part. It describes a quieter calendar than the real one. The report also expects the US to press Poland to change course. A version of that was already visible next door: in August, Lawfare described US pressure on Lithuania to reopen potash transit.

Credit first

The report separates co-aggression from co-belligerence. Belarus gave its territory for the invasion of Ukraine, but its own army has not fought there. The authors argue this gives the EU more tools for Minsk than for Moscow. Their advice that Brussels should not wait for signals from Washington is sound. So is the point that UN mechanisms keep political prisoners on the agenda, whatever happens in Washington or Brussels. The report also rejects a full lifting of EU sanctions, which keeps its call for engagement within limits. Its forecast that any normalisation would be limited and gradual has held up well. The 16 September deal freed a tenth of the March number, far from the hundreds Coale had hoped for in August.

Its most useful idea is a warning. Panellists asked Europe to separate real signals of change from political noise, and to beware of a “false dawn”. The sixth recommendation repeats this. Yet the report never says what a real signal would look like, or which number would show one.

Where the envoy and the counters disagree

The prison is the obvious place to test the warning, because that is where the trade happens.

Between releases, the stock has twice turned back up.

Coale has claimed credit for 700 to 800 releases since mid-2025. Over the same period, Viasna’s count of political prisoners fell from 1,185 at the start of June 2025 to 911 on 16 September 2026. That is a net fall of 274. It is not yet clear whether that reading already excludes the 25 people freed the same day.

The two figures may not cover exactly the same people. Still, the curve has a clear pattern. Each release pulls the stock down. Between releases, the stock has twice turned back up: in January, and again since June. It was 860 on 14 June, 863 at the end of July, 889 on 31 August and 911 in mid-September.

Part of the spring fall was bookkeeping. In April Viasna reviewed its list, and about 100 people were moved to the status of former political prisoner. So the net fall of 274 overstates what the releases achieved.

Viasna also keeps a second counter: everyone recognised as a political prisoner since the 2020 campaign. It stood at 4,655 on 14 June and 4,815 on 16 September. That is 160 new political prisoners in 94 days, about one every 14 hours. In the same period the stock grew by 51, so roughly 110 people left the list, through completed sentences and pardons.

At the meeting on 15 September, Lukashenko described the objection himself. Émigrés, he said, had persuaded the Americans that Minsk releases some people and detains twice as many. He called this an exaggeration. Twice as many is indeed too high. Over the summer the ratio was closer to three in for every two out. In the twelve months to 9 August, Viasna recorded 538 new political prisoners. The three largest US-brokered releases of that period freed 425 people. We first set out this arithmetic in August, in The Offer in the Same Speech.

A released political prisoner who went home on a Belarusian passport has no Russian equivalent.

Viasna lawyer Pavel Sapelka calls the system a revolving door. The report makes the same point in general words. It says releases belong to a transactional policy and prove nothing about a new course. What it does not add is the one series that would tell a reader, month by month, whether that is still true.

Two registers that kept moving

A false dawn can also be checked outside the prisons. On 1 January 2021 the Ministry of Justice listed 3,021 registered public associations. On 1 January 2026 it listed 1,337. The fall was steepest during 2023 and 2024. In 2025, the year of the first US deals, it slowed to 60 but did not turn.

The register of extremist materials moves the same way. In 2024, its busiest full year, courts added 1,462 entries, about four a day. In the first 225 days of 2026 they added 977, a slightly faster pace. From July 2027, people who found a new non-profit will [face KGB checks](LINK: Who May Start One). None of this is in the report. Each item is dated and counted, which is exactly what the sixth recommendation needs.

Poland, and a test with a date

The report calls Poland’s position “very limited and rather rigid”. Its Polish panellists said the regime does not see Warsaw as a political partner. Both points may be true of public diplomacy. But eight days before the panel, Polish intelligence and the KGB completed an exchange at the border. It freed Andrzej Poczobut and four others, with US help, and Donald Tusk met Poczobut at the crossing. The report mentions Poczobut only as a gesture that should not be overestimated. It leaves out that the Polish state had just negotiated directly with Belarusian security services.

Twice as many is too high. The count still rose at every reading since July.

Poczobut then gave the report its best test case. In June he said he had a passport and assurances that he could go home. On 8 September he crossed back into Belarus and went to Hrodna. Reuters described the trip as a visit, while he has spoken of going home. Tsikhanouskaya praised his courage and said his return was no sign of a thaw. The same week, a Catholic news agency reported that a 17th Polish-born priest had been refused permission to keep serving in Belarus this year.

Here the report touches a real difference between Minsk and Moscow, but gets the example wrong. It says Russia, unlike Belarus, has not released its own political prisoners. That is incorrect as written. In August 2024 Moscow freed 16 people in an exchange with Western states, among them the opposition politicians Vladimir Kara-Murza and Ilya Yashin. Yashin later said an FSB officer had warned him against coming back. A released political prisoner who went home on a Belarusian passport has no Russian equivalent. Whether the 25 people freed on 16 September stay in Belarus is the next check of the same kind. Minsk does not offer every prisoner a way home: of 886 political prisoners pardoned since 2024, 205 were expelled. Whether Poczobut stays free in Hrodna through the autumn is a dated, observable test of the report’s own warning.

Gaps in the picture

The panel’s second theme was Minsk’s relations with Moscow and Kyiv. Kyiv hardly appears in the text. There is no mention of Ukraine’s sanctions against Lukashenko on 18 February, which cited relay stations for Russian attack drones on Belarusian territory. The public programme lists 22 speakers. Fourteen have an affiliation shown, and none of those is Ukrainian or Latvian.

Of 886 political prisoners pardoned since 2024, 205 were expelled.

The report says part of Belarusian youth is turning pro-Russian. It gives no source. Chatham House’s urban survey put support for an alliance with Russia at 44% in November 2025, the highest since September 2020. That figure covers the whole sample, and the report does not say whether young people drive it.

One point concerns method. The speaker list is public, and the report labels some views as “Lithuanian” or “Finnish”. When a public list shows only one or two institutions from a country, a national label can point to a person. The Chatham House rule protects who said what, and labels like these weaken that protection.

Watch items

• Viasna’s count will be higher on 31 December 2026 than at its first reading after the 25 people freed on 16 September are removed from the list.

• No single release of 100 or more people will be announced before 31 December 2026.

• By 15 November 2026, Poczobut faces no new criminal case and can still cross the Polish-Belarusian border in both directions.

• The Union of Poles in Belarus holds the congress it postponed from 13 September by 5 March 2027, the end of the six months Poczobut named.

• At the February 2027 renewal, the EU extends its Belarus sanctions without removing the potash measures.

Method and limits

The calendar uses public reports only and leaves out events we could not date. Viasna’s counts come from its own statements and from agencies quoting it. Readings between June and December 2025 and between April and June 2026 are not shown. The April 2026 list review moved about 100 people out of the stock without any release. The flow figures for June to September are the difference between Viasna’s two counters, so they carry no names. The envoy’s 700 to 800 and Viasna’s net change measure different things. His figure may include people who were never on Viasna’s list, and people also leave the list when their sentences end. Association figures are the Ministry of Justice’s 1 January line as quoted by Lawtrend and state media.

On 14 September 2026 YouTube terminated another group of Belarusian state channels. CTV went down during a live talk show, and three radio channels of the state broadcaster followed: Radyo Stalica, Radio Kultura BY and the First National Channel of Belarusian Radio. State agencies listed those four.

They did not list a fifth one. The channel BELARUS-RTR | Belarus News, which belongs to the CTV group, stopped working on the same day. We had been collecting it for eight months, so there is a record of what it did before it went.

That record is the reason for this piece. The channel is not just another item on the list: it is the outlet that took over after the last removal wave, and its takeover is visible week by week.

April, and the week after

On 3 April YouTube removed BelTA, CTV and ONT, which together held more than six million subscribers. A backup CTV channel appeared at once and lasted five days.

Belarus-RTR did not need to be created, because it already existed and simply grew. In March it published about thirty videos a week. From the week of 6 April the pace rose to between fifty and seventy-three videos a week, and it stayed there until the end of summer.

The backup channel lasted five days. The heir lasted five months.

The audience followed. In January the channel collected 470,000 views, and in July it collected 39.5 million. Split the data at the April date: the first three months gave 388 videos and 7.4 million views, the next five gave 1,273 videos and 99.8 million. Subscribers went from 48,400 in January to 134,000 in August.

One more detail sits in the raw data: in the May snapshot the channel changes its name. Until April it was simply Belarus News, and from May it became BELARUS-RTR | Belarus News, with the television brand added in front.

The rule this runs into

YouTube’s help pages are direct about successors. After a channel is terminated, its owner may not use, own or create any other channel. Owners of terminated channels may not use anyone else’s channel to get around it. The ban covers new channels, bought channels, and channels where the terminated party appears often or prominently.

Circumvention is a violation in its own right, and the penalty is termination of every linked channel.

A successor channel is not a hiding place. It is a signature.

Our data describe that shape almost exactly. A channel of the same group doubled its output in the first week after the group’s main channel was terminated. It absorbed the audience, then put the broadcaster’s name on the door.

The termination notice points the same way. We checked both channels from a European exit and from a Turkish one, and both show the same line: the account was terminated for a violation of the Terms of Service. That is the wording ONT quoted in April, while terminations for accumulated strikes usually name the Community Guidelines and a specific policy instead. The mapping is not documented anywhere, so we hold this part loosely.

What the content does not explain

We classified all 3,593 videos of the two channels by title and description, and the result does not fit a content story.

The radio channel does carry a large war layer. It published 358 videos about the war around Iran, which drew 39.1% of its views. But that layer peaked in March and April. By August it was down to 23 videos a month, and monthly views had fallen from 10.35 million to 2.03 million. The topic had been fading for four months when the channel went.

Belarus-RTR is almost the opposite, with Iran content at 1.7% of views. Security-service material, the kind that carried names and personal data on CTV in the spring, is 1.3% of the corpus and half a percent of views on both channels together. Most of it is ordinary crime reporting.

One channel ran on a Middle Eastern war. The other ran on tractors. They were removed on the same day.

The two channels are very different in what they publish, and identical in who owns them.

What was actually lost

The loss is real, but it is not the loss of a political audience. Videos mentioning Lukashenko account for 80.3% of all views on the two channels, and on Belarus-RTR the figure is 90.5%, most of it from 453 short clips.

The best performing video of the year shows Lukashenko being presented with a tractor, and it has 13.8 million views. The same clip was also the radio channel’s top video in May, which tells you the material comes from one central pool.

Long-form video is a different story. Its median is 2,470 views on Belarus-RTR and 676 on the radio channel, so the programmes themselves had almost no audience here.

What to watch

By 15 November 2026. Whether the CTV group runs another YouTube channel that passes 50,000 subscribers. If a new heir appears and survives, the circumvention reading weakens.

By 31 December 2026. Whether the three radio channels return under new names. A quiet return would suggest the decision was narrower than an owner-level one.

By 31 March 2027. Whether any of the removed outlets publishes a strike notice naming specific videos. That would move the third row of the table above from Medium to Low.

Method and limits

Data come from the YouTube Data API v3, collected in monthly snapshots from January to August 2026, and each snapshot holds videos published that month. After removing duplicates we have 3,593 videos from the two channels. View, like and comment counts are snapshots taken at collection time, so newer videos are undercounted.

Categories were built with regular expressions over titles and descriptions. Descriptions on these channels are often news digests that glue unrelated stories together. Every category was therefore counted twice, with and without descriptions, and both numbers were checked before use. Shorts are separated from long-form at 180 seconds.

The corpus ends on 31 August, so it says nothing about September, and channel status was checked on 15 September from two exits. We have no correspondence between the platform and the newsrooms, and no formal statement of grounds. The mechanism described here rests on YouTube’s published rules and on measured behaviour, not on a document.

The full analysis, with the category tables and the code, is published on FactCheck.LT.

A widely shared post last week reported that the number of sole traders in Belarus fell 11.8% over the year to 1 August 2026, and that the worst-hit activity was information and communication, down 35.1%. Both figures are correct. They are also the last point of a line that has been falling for five years, and the fall is accelerating. We pulled the series from fifty of National Statistical Committee of the Republic of Belarus own monthly bulletins.

The five-year line

Belstat publishes a table in every monthly bulletin comparing the number of sole traders with the same date a year earlier. Read in sequence, the bulletins give a rolling picture.

In 2021 both the total and the information sector sat near 95, which is a fall of about five per cent. In 2022 the total held at 94.7 to 95.2 while information dropped to between 84.2 and 88.8. In 2023 both recovered slightly. Then the lines part. Through 2024 the information index fell from 87.8 to 72.5 inside a single year, while the total went from 94 to 88.2. In the most recent bulletins the information index reads 64.3 to 67.9, and the total 85.9 to 88.2.

So the sector has not just shrunk. It has shrunk faster than the whole every year since 2022, and the gap has widened from about ten points to more than twenty.

Not a bad year. A five-year slope that is getting steeper.

What separates the categories

The latest reading sorts the activities cleanly, and the sorting is not about how profitable they are.

At the bottom: information and communication at 64.9, professional, scientific and technical work at 77.9, other services at 85.6. In the middle: vehicle repair, industry, construction, real estate, all between 86 and 93. At the top, growing: health at 102.4, transport at 104, and arts, sport and leisure at 116.4.

The line runs between work that can be done from anywhere and work that cannot. A programmer, a translator, a designer or a consultant can keep the same clients from Warsaw, Vilnius or Tbilisi. A minibus driver, a dentist or a gym instructor cannot.

Transport is the control case. It sat between 96 and 100 for four years and has now turned positive. Whatever is emptying the information sector is not emptying the roads.

A consultant can move. A minibus cannot.

Whatever is emptying the information sector is not emptying the roads.

What this is not

It is not a straightforward story about emigration, and we cannot make it one.

The status of a sole trader has been under legislative pressure since 2023. Many were required to re-register as legal entities or move to self-employed status, and hiring staff was barred. So part of the total fall is people changing their legal form without going anywhere, and the total index would have fallen with or without anyone leaving.

But that pressure applies to every activity equally, and the outcomes are not equal. A rule that hits all sole traders cannot explain why information and communication loses three times what transport does. Something is separating the categories, and the most obvious candidate is whether the work can be carried across a border.

We cannot prove that from this table. What the table proves is that the divergence is real, sustained and widening, which is more than the single figure showed.

A rule that hits everyone equally cannot explain unequal outcomes.

Confidence

Watch

• The information index stops falling in the next quarterly readings. It has fallen in every year since 2022.

• Transport stays above 100. It crossed this year for the first time in the series.

• National Statistical Committee of the Republic of Belarus publishes absolute counts by activity rather than only the index, which would allow the fall to be read in people.

• Any official explanation of the divergence between activities appears. So far there has been none.

Method and limits

The series comes from the table on business entities that National Statistical Committee of the Republic of Belarus prints in each monthly bulletin, pages 14 to 21 depending on the issue. Each issue carries two blocks with identical activity labels, one for legal entities and one for sole traders; we separated them by value, since the information index sits near 100 for companies and far below it for sole traders.

The figures are indices against the same date a year earlier, not counts. A reading of 64.9 means the number is about a third lower than a year before; it does not say how many people that is.

We could not date the individual bulletins. Belstat’s archive in our corpus has no publication date field, and the period labels in the extracted data are unreliable. So the chart shows the range across all bulletins of each year rather than a monthly line, and nothing here should be read as saying what happened in a particular month.

The 11.8% and 35.1% in the original post match our reading of the latest bulletin exactly, which is the main check we have on the extraction.

Every week our digest sorts Belarusian coverage into sixteen topics and records how much of each comes from state outlets and how much from independent ones. Fifteen of the sixteen sit where they always sit, or wobble. Culture has moved one way for five weeks running. In early August the state held 23% of it. In the week of 31 August to 6 September it held 65%.

The series

In the first half of August, state and pro-regime sources produced 15 documents that the digest tagged as culture, against 51 from independent outlets. The next readings ran 22 against 40, then 31 against 45, then 30 against 38. In the week just past: 52 against 28.

No other topic behaves like this. Russian integration and the Global South stay with the state every week. Democratic forces and Western dialogue stay with independent media. Military and education coverage swings back and forth without direction. Culture went one way, five weeks in a row, and crossed over.

The crossing falls in a particular week. The XXXIII Day of Belarusian Writing opened in Kostyukovichy on 5 September with about 600 performers and a greeting from Lukashenka, who called it a festival of distinctive national culture. Two days earlier SB, the government daily, presented him as the author of a concept of cultural sovereignty resting on care for the purity of the native tongue.

Twenty-three per cent in early August. Sixty-five in the first week of September.

What was underneath

The volume moved one way. The substance moved the other.

In the same week, independent outlets reported that the Hrodna drama theatre is preparing a play about Denis Davydov, a Russian hussar and poet of the Napoleonic wars. The Republican Theatre of Belarusian Drama opened its season with Pushkin’s Little Tragedies, in Russian. The Kupalauski theatre announced a laboratory where Russian directors will stage sketches drawn from Belarusian drama.

Also in the same week, Leanid Barščeŭski published the first complete translation of the Odyssey into Belarusian.

So the festival of the native tongue took place in a week when a state theatre was rehearsing a Russian poet, the national drama theatre had opened in Russian, and the largest Belarusian-language literary event was a translation published by a man who lives in exile.

A festival of the native tongue, in a week when the national drama theatre opened in Russian.

The largest Belarusian-language literary event of the week was published from exile.

What the numbers do and do not mean

They do not mean the state is producing more culture. They mean the state produced more coverage tagged as culture, in a week when it was holding its own cultural festival. Festivals generate press releases, and a state with a large press apparatus can dominate any topic on the day it chooses to.

What is harder to explain away is the four weeks before. The share moved from 23% to 35% to 41% to 44% before the festival week, without a single event of that size. Something was rising steadily and the festival finished the job.

And there is one reading the numbers cannot settle. A topic held by the state may mean the state is speaking loudly, or it may mean the other side has less left to report on. Over the same period we counted more than 2,000 organisations closed, seven manor houses struck from the heritage register, and a rule that will put the security service at the entrance to any new one. Those are not cultural stories in the tagging, but they remove the people who make cultural stories.

They remove the people who make cultural stories.

Confidence

Watch

• The state share of culture falls back below half in the next two weekly readings. If it does not, the festival was not the cause.

• Another of the sixteen topics shows a five-week monotonic move. So far none has.

• A Belarusian state theatre announces a premiere in Belarusian for this season.

• The Day of Belarusian Writing generates a published figure for attendance or cost, which would allow the coverage to be set against the event.

Method and limits

The weekly balance comes from our own digest pipeline, which assigns each document to one of sixteen topics using a language model. We are therefore measuring what that pipeline calls culture, not culture. A different classifier would draw a different line, particularly between culture, education and ideology, which overlap.

The five readings are not identical windows. The first covers 1 to 16 August and the rest are single weeks, so its counts are not directly comparable to the others; the share it gives is, which is why the second chart uses shares.

Document counts reflect what our collectors hold, and collection is not equally complete across sources. The direction of the series is the claim here, not the exact percentages.

On 3 September Sviatlana Tsikhanouskaya told the Lithuanian portal 15min that Russia can use Belarusian territory to escalate and provoke NATO neighbours, and that she receives sensitive information from Belarus about new military infrastructure going up there: military antennas, near the Ukrainian border and near the EU border alike. The next day the Lithuanian president’s chief national security adviser, Deividas Matulionis, said intelligence cannot confirm that infrastructure is growing, and asked where she gets such information. Both deserve to be taken seriously, and the disagreement is narrower than it looks.

What was said

Her words were specific. Sensitive information from Belarus. New military infrastructure that a country not preparing to escalate would not need. Various military antennas. Not only at the Ukrainian border, but at the border with EU states too.

Matulionis was equally specific, and his phrasing matters. Intelligence cannot confirm her claims about growing military infrastructure. He did not call the claims false. He said his service cannot stand behind them, and he wondered aloud about the source.

That question is the interesting one, because a good deal of what she describes needs no source at all.

What is already on the record

Take the antennas first, since they are her own example. Relay masts that extend the range of Russian drones over Ukraine became public in January 2026. Researchers had located four by March, and on 18 February Ukraine imposed sanctions on Lukashenka over them. None of this required a confidential channel; we set it out on 21 August while checking someone else’s chronology.

The rest of the Ukrainian-border picture is equally open. On 2 August the commander of Belarus’s special operations forces announced a new air-assault brigade near Homel, through the government daily. A rail spur toward that border shows in cadastral maps from late 2023, and a named state institute did the design work. The Russian defence ministry confirmed the Oreshnik deployment on 30 December 2025.

So for one half of her claim, the evidence is published, dated and attributable. The Belarusian state published some of it.

The strongest part of the case needs no confidential source. Minsk announced it.

Some of the evidence was published by the Belarusian state itself.

Where the two accounts actually part

The other half is the EU border, and there the record is thin.

We hold nothing comparable: no announced unit, no cadastral trail, no imagery, no sanctions listing. Our own count of reported exercises across fifteen months puts activity in the western regions, but exercises are training, not new construction, and the two should not be conflated.

That is exactly where an intelligence service should know more than we do, and exactly where it says it cannot confirm. The disagreement is not about Belarus arming. It is about one specific claim, on one specific border, sourced in a way nobody outside can test.

He did not say she is wrong. He said his service cannot stand behind it.

Why this matters for the argument

Nothing here suggests the claim is untrue. Intelligence services decline to confirm things for many reasons, including that a thing is true but not yet shareable.

The point is about what makes an argument hold when it is contested. Nobody can examine a claim that rests on sensitive information, so it stands or falls on the audience’s trust in the speaker. Anyone can check a claim that rests on a cadastral map, a sanctions listing and a government newspaper — including people who do not trust the speaker at all.

For the Ukrainian border, the second kind of evidence exists in quantity. Leading with the first kind invites the question Matulionis asked, and nobody can answer it.

This is not a small matter of presentation. Belarus’s democratic movement is asking European governments to act on its reading of what is happening inside the country. That reading persuades more, not less, when it points at what those governments can verify for themselves.

A cadastral map persuades people who do not trust you. A sensitive source does not.

Confidence

Watch

• Any public evidence of new military infrastructure on the Belarusian side of the EU border appears: imagery, procurement, cadastral records or a sanctions listing.

• Lithuanian intelligence revises its position in either direction after the CSTO exercises of 11–16 October.

• Tsikhanouskaya’s office sets out which of its claims rest on open sources and which do not.

• A second NATO government states a position on the same question. So far only Lithuania has.

Method and limits

The quotations come from LRT’s report of the 15min interview, published 3 September, and from Delfi’s report of Matulionis on the same subject. We have not seen the 15min interview in full, and a longer version may contain qualifications that the reports omit.

“Cannot confirm” is not “denies”, and the difference carries the argument. We use the phrasing Delfi published and do not strengthen it.

We compiled the open-source items listed here in earlier work, and date each to the report that established it. The relay masts, the brigade, the rail spur and the Oreshnik statement are separate matters that happen to sit on the same border; nothing here treats them as a single programme.

Our exercise counts come from our own corpus and measure reported exercises, not exercises. We use them above only for what they can support: training activity is not the same as new construction.

From 1 July 2027 the security service will check the people who found a Belarusian non-commercial organisation before it can register. The check falls on the founders, not on the organisation. The rule comes from an act on the registration and liquidation of legal entities adopted on 1 June 2026, and covers private institutions, associations and unions. Four independent outlets reported it. The state press did not.

The exit was closed first

We counted the other end of this on 28 August. By the end of June 2026, Lawtrend’s register held 2,044 closed organisations: 1,244 that the state shut by order and 800 that resolved to dissolve themselves. In the quarter to June, for the first time, more closed themselves than the state closed.

Against that stands the sector as it is now: about 1,943 non-commercial organisations of the three main legal forms remain on the register.

So the shape of the last six years is a sector reduced by roughly half its own size again, mostly through orders, lately through exhaustion. The new rule addresses what is left of the other direction. Nobody needs to refuse a registration if the security service checks the applicants first.

Two thousand and forty-four closed. Now the entrance.

What the rule does

The reporting is thin but consistent. DW, citing Viasna, wrote on 25 August that a check of the founders by the KGB becomes a compulsory stage of registration from 1 July 2027, for natural and legal persons alike, and for associations and unions. Euroradio reported the same day that new civic initiatives will be cleared by the security service from 2027. CityDog explained the procedure on 29 August. Reform named the act on 6 September, inside Viasna’s monthly review.

Two features are worth separating.

The check is of people. A registration refusal normally rests on the documents: the charter, the name, the address, the founding meeting. A check of founders rests on who they are, and it comes before anyone reads the documents.

And the lead time is long. The act dates from 1 June 2026 and takes effect thirteen months later. None of the reporting explains why, and we will not guess.

Four outlets carried it. The state press did not mention it once.

The case for the other reading

There is an argument that this changes less than it appears, and it deserves stating.

Registration in Belarus has not been a neutral procedure for years. Refusals have been common, grounds have been elastic, and applicants have long assumed that the security services had a view. On that reading, the 2027 rule writes down what already happened, and writing it down is more honest than the alternative.

That argument is reasonable, and it points at the real change. An informal practice can be denied, varied or quietly dropped. A stage in a procedure cannot. After July 2027 nothing remains to appeal against and nobody remains to blame: the refusal will be lawful, procedural and final.

An informal practice can be dropped. A stage in a procedure cannot.

The check is of people, not of papers.

Nobody told the public

Our corpus for June to September holds four independent outlets on this rule and nothing at all from state or pro-regime sources.

This changes who may found an organisation in Belarus. The state press covers registration rules, charity campaigns and volunteer awards in detail, and it has not mentioned this at all. As with the transfers of farms we wrote about on 3 September, the silence is not inattention to the subject. It is silence about one part of it.

Confidence

Watch

• Someone publishes or quotes the act in full, with the list of legal forms and the time limit for the check.

• A state outlet describes the rule before it takes effect in July 2027.

• The number of registered non-commercial organisations falls below 1,900 in the next official count.

• Any organisation registers successfully after 1 July 2027 and says publicly how long the check took.

Method and limits

We have not seen the act. Everything here about its content comes from four reports and from Viasna’s monthly review, which is their common source. The two dates, 1 June 2026 and 1 July 2027, are as those reports give them.

The closure figures are Lawtrend’s, which states them as minima. We published them here on 28 August. The count of registered organisations is from Lawtrend’s December 2025 monitoring and covers the three main legal forms.

The Centre for New Ideas published the third wave of its Telegram mapping on 8 September (report page, PDF, RU). Alexander Stukanau presented it at a webinar the next day, and Euroradio wrote it up the same evening under the headline that pro-regime channels are losing influence. The report covers 170,140 posts from 33 pro-regime channels between January 2025 and May 2026, plus 7,395 reposts between them. It is the largest published count of this segment. It is also the third team in a year to find the same channel at the top.

What the count shows

Two thirds of all reposts inside the network come from one channel: Pul Pervogo, the presidential press service account. BelTA makes 37% of all reposts and takes content from only three sources. ZhS Premium makes 16% and takes from sixteen. Pul Pervogo itself made 14 reposts in seventeen months. The report calls this a broadcasting tree rather than a network. One producer, two relays, and a periphery that repeats the flow without talking to itself. Reciprocity is low and clustering is close to zero.

FORESIGHT reached the same apex in July from a different corpus. We took five personal channels (Azaronak, Tur, Dzermant, Shpakouski, Gigin) and counted every forward each of them made. Pul Pervogo was the most forwarded source for all five, 9,900 forwards in total. The Digital Freedoms Lab reached it a third way in August. In its eleven-channel sample it found 1,066 shared posts against four independent ones, with Nevolfovich and ZhS Premium as the upstream pair (Office Hours). Three teams, three samples, one apex. That part is settled.

Three teams, three samples, one apex.

Where the frame ends

CNI’s graph is closed. An edge exists only when both ends are among the 33 channels. Anything forwarded from outside the roster does not appear. Our forward count is open. It records every source a channel forwards, whether we collect that source or not. Placed side by side, the two pictures agree on the apex and disagree about what stands next to it.

In the FORESIGHT count, the second and fourth most forwarded sources are the Telegram channels of two state television companies, STV News (4,756 forwards) and ONT News (3,831). NEWS.BY and SB.BY follow. None of these four is in CNI’s roster, so in CNI’s tree they do not exist. Further down come Russian nodes: Solovyov (248), Dyukov (174), Sputnik Belarus (126), Korotchenko (98), RT (59). All five personal channels forwarded every one of them. The report’s second wave concluded that the segment did not show a significant dependence on the Russian segment. That conclusion follows from the roster. A graph with no Russian node cannot show dependence on one.

A graph with no Russian node cannot show dependence on one.

There is a second gap, and the report’s own coding exposes it. It finds that 52% of posts are copy-paste or simple aggregation, and that in 34% of posts the original source cannot be identified. A forward graph sees none of that. The 7,395 reposts are 4.3% of the 170,140 posts. The tree is drawn from that 4.3%. The other half of the low-originality content moves along routes the graph cannot see. The reviewer at the webinar, Iryna Sidorskaya, made the same point in different words: the division of functions is proven, the “one editorial office” is not. Proving an editorial office needs timing, not forwards. Same-minute publication across channels is a measurement we have in an earlier FORESIGHT layer, and it belongs in a separate piece.

The number that depends on the clock

The report’s line for the press is that views fell 32%, from about 85 million a month in the first quarter of 2025 to 58 million in March to May 2026, while posting stayed flat (10,789 posts a month at the start, 10,195 at the end). Belarusian Silovik is excluded from that series because it holds 66% of all views on its own and its per-post counts barely vary, which the report reads as administered reach.

Telegram view counters are cumulative. A post from January 2025 had seventeen months to collect views before the network was crawled in mid-June 2026. A post from May 2026 had weeks. Most views arrive in the first days, but not all of them, and forwards and search keep adding for months. The report does not say when views were captured, so a reader cannot separate audience decline from capture lag. Our own YouTube layer has the same defect, and we wrote it into the Stage 1 design: without a capture timestamp, the indexing date is the only proxy. The direction of the CNI finding is plausible. Its size is not measured.

The direction is plausible. The size is not measured.

Subscriber counts are cleaner, because they are dated: 1 January 2025, 1 January 2026, May 2026. Nineteen of the 33 channels lost subscribers during 2025. Belarusian Silovik gained 64,531 and Pul Pervogo 15,231, and together the two took 74% of all growth. Two regional channels, PuLS Grodno and Pul 4 Region, lost 27 to 33 thousand each. The report reads that as a reset or rename rather than an exodus, and it is right to hedge.

What the third wave changed

The method moved to LLM coding, the roster grew from 16 channels to 33, and the window from 12 months to 17. The report then compares tone with the second wave across that change: neutral tone dominated then, three near-equal shares now. No validation sample is reported, no agreement figure, no model named. A four-point fall in hate speech, from 11% to 7% of posts, cannot be told from coder drift without one. The shift from domestic to foreign topics (domestic politics 25% to 17%) has the same problem in a different form. Four institutional channels were added to the roster, and the report itself says part of the shift may come from their protocol posts.

The tree is drawn from the 4.3% of posts that were forwards.

None of this makes the tree wrong. Three independent counts put the same channel at the top, and the functional split between positive institutional channels and negative personal ones is the report’s best-supported result. The frame is the issue: the tree has branches that run to state television and to Russian channels, and the frame was drawn before those branches.

Watch items

• By 30 September 2026: FORESIGHT re-runs the five-channel forward count restricted to January 2025 to May 2026 and reports the share of forwards going to sources outside CNI’s 33. Expected above 40%. Below 20% and the “closed frame” argument here is overstated.

• By 31 May 2027, if wave four runs on the same cadence: Pul Pervogo’s share of intra-network reposts below 55%, continuing the 77% to 59% slope. Above 65% and the report’s “slow diversification” reading reverses.

• Any re-crawl of the January to May 2025 posts: views within 5% of CNI’s figure and the capture-lag caveat here falls; a gap above 15% confirms it.

Method and limits. Report figures are from the CNI PDF (September 2026), tables 1 to 5 and Appendix A; the interactive network map is linked from the report by QR code. FORESIGHT figures are from the July 2026 forward-graph run over five personal channels plus NEWS.BY and Pul Pervogo (about 267,000 posts); they are lifetime totals from forwarded_from metadata and are not restricted to CNI’s window, which is why the first watch item exists. DFL figures are from the Human Constanta / Digital Freedoms Lab report covered in “Office Hours”. CNI’s shares are shares of labels for multi-label variables (legitimation, enemy identity, source type) and shares of posts for the rest. Subscriber counts are CNI’s, from external channel statistics that the report does not name.

Between 26 August and 2 September, Russian Houses in Belarus handed out 1,500 backpacks to first-year pupils in at least five regions. The action is called “Let’s go! For knowledge”, the backpacks carry a print called “Space friendship”, and independent outlets report the wider brand as “Russia is with you”. The Russian Houses handed them out inside a national campaign that the Belarusian Red Cross runs. The International Federation suspended that society in December 2023 and has not readmitted it.

The rollout

The launch was in a Mahiliou school on 26 August. Homel region and Mozyr followed on the 27th, Hrodna region on the 30th and 31st, Brest on the 31st. On 1 and 2 September the Hrodna Russian House visited a special school, and Minsk schoolchildren went to a city cinema event.

TASS reported on 4 September that more than a thousand pupils had received the backpacks. Reform and Belarusian News put the planned figure at 1,500. Sputnik had announced the handover on 26 August.

This is not a one-off. In May Rossotrudnichestvo gathered all its humanitarian work with children in Belarus into one named project, “Together for a happy childhood”. The August action was its first republic-wide event.

What did not change

It would be easy to describe this as a surge. Our corpus does not support that.

The four Russian House channels in Belarus published 186 documents about children and schooling in the second half of 2024, 415 in 2025 and 297 in the first eight months of 2026. That works out at about 35 a month in each of the three years. The rate is flat.

What changed is the shape. Before, the work came as many separate events: film clubs, book donations, trips to Artek, chess tournaments, literary quizzes. Now it has one name, one national campaign, and one delivery partner.

The volume is the same. The structure is new.

One name, one national campaign, one delivery partner.

The partner

The Belarusian Red Cross has not been a full member of its own movement for nearly three years.

In October 2023 the International Federation’s investigation found that its secretary general, Dmitry Shevtsov, had breached the movement’s Fundamental Principles, misused the emblem and breached the IFRC integrity policy. The findings followed his statements about nuclear weapons and about moving children to Belarus, and his visit to Luhansk and Donetsk. The Federation asked for his dismissal by 30 November 2023.

The society re-elected him instead. On 1 December 2023 the Federation suspended its membership. The society lost its member rights, new IFRC funding stopped, and it can no longer vote or stand at statutory meetings. In the previous year it had received about 1.7 million Swiss francs from the Federation.

The same investigation is worth quoting for what it cleared. It found that another organisation, not the Belarusian Red Cross, led the movement of children from Luhansk and Donetsk, and that IFRC money did not pay for it. The society’s work with children is inside Belarus.

Shevtsov is still secretary general. In December 2025 he said the society has no official contacts with the international Red Cross. In August 2025 state enterprises in Homel region required their workers to fund it.

A society suspended for breaching neutrality is now the channel for a foreign state’s programme.

What this shows and what it does not

The chain is documented at every link: a foreign state agency, a named children’s programme, a national charitable campaign, and schools.

It shows nothing about the children. A backpack is a backpack. Some 87,500 children started school in Belarus this month, so 1,500 backpacks reach under two per cent of them. Reaching a child is not persuading one. The seminar record we wrote about on 1 September made the same point about the state’s own military training: we can measure what is installed, not what is absorbed.

What the chain does show is access. A foreign government reaches Belarusian schools through an organisation that sits outside the oversight of its own movement. Nobody but the state needs to approve that route.

The chain shows access. It shows nothing about the children.

Confidence

Watch

• The 1,500 figure is confirmed or corrected by the organisers with a regional breakdown. So far only the total has been published.

• “Together for a happy childhood” runs a second national action before the end of the year.

• The Belarusian Red Cross is readmitted to the IFRC, or the Federation restates the suspension.

• Any Belarusian school or education authority declines to host the initiative, which none has publicly done.

Method and limits

Counts come from our corpus for 1 July 2024 to 9 September 2026. They match documents about schooling, first-year pupils, backpacks, Russian Houses, Rossotrudnichestvo or the Red Cross, with a Belarusian context in the same passage. Russian House channels here means the four regional Telegram channels in Minsk, Homel, Hrodna and Brest. No source was excluded from this run.

The 1,500 figure is the organisers’ own. TASS reported more than a thousand actually handed over. We have not seen a breakdown by school or region, so our count of regions is a floor, drawn from what the channels themselves reported.

The IFRC findings and the suspension come from the Federation’s own statements of October and December 2023. Whether the arrangement described here breaches any rule of the movement is a question for the movement, and we do not answer it.

Belarusian consumer confidence is now close to zero: −0.8% on the Russian method and −0.2% on the European one, against an EU average of −14.3%. On paper, urban Belarusians feel better about their economy than Poles, Greeks or Russians do. The same survey shows that a third of those whose income fell are behind on their bills, and that job losses have almost quadrupled in two years. The figures come from the quarterly survey of the Monitoring of the Belarusian Economy project, run by MacroBY.org, which has polled about a thousand urban residents eighteen times since December 2021.

What people say

The index sits where it does for a reason. Negative views of the past year fell in the third quarter: 28% say the country’s economy got worse, down six points, and 27% say their family’s position got worse, down five. Expectations are the strongest part. Only 13% expect their household finances to worsen, while 31% expect them to improve.

Compared with Europe this looks strong. It is also almost unchanged: the Belarusian index rose by 0.9 points over the quarter while the EU average rose by 5.1. Belarus is not improving faster than Europe. Europe moved and Belarus did not.

What people do

A quarter of respondents say their income fell over the past month. Take out those who blame only the exchange rate and it is 22%; take out rising prices as well and 18% still report a fall for other reasons.

Ask that quarter how they coped, and the answer has changed. In the third quarter of 2024, 6% of them had fallen behind on obligatory payments such as utilities and loan instalments. In every wave since, the figure has run near 30%, and stands at 28% now. Delaying a large purchase used to be the common response. Missing a payment has replaced it.

Job loss has moved too. At the end of 2024, 1.4% of respondents said they had lost a job. In the second and third quarters of 2026 it is 5.4% and 5.5%. Almost four times as many.

Spending tells the same story from the other side. Half of respondents, 51%, say they have spent more over the past six months, and 77% of them attribute this to prices. Ten per cent attribute it to higher wages. That last figure is worth holding next to the official one: the Eurasian Development Bank reported real wages in July up 9.1% year on year. One in ten households noticed.

Six per cent were behind on payments two years ago. Around thirty per cent have been ever since.

Real wages rose 9.1%. One household in ten noticed.

The gap is not a contradiction

Both halves can be true, and the survey shows why.

The index is built mostly from expectations, and expectations in Belarus are held up by one component: how households think their own finances will go. That reading is 9.6%, the most positive part of the whole index. Views of the past year are negative, at −3% for the economy and −3% for family finances. The best time for a large purchase scores −9.5%.

So people are not saying that things are good. They are saying they expect their own situation to hold. Meanwhile 63% still expect prices to outrun their income, and only 6% expect the reverse. Confidence here means the absence of expected deterioration, not the presence of improvement.

Confidence here means no expected deterioration, not expected improvement.

Where the distrust went

The survey repeated its trust questions from July 2024, and the two-year comparison is the most careful part of the bulletin.

Distrust fell almost everywhere. Fewer people now say they distrust the courts, down ten points, the government, down ten, parliament, down ten, non-state media, down eleven, state media, down eight. That reads like a state regaining legitimacy.

It did not become trust. It became “don’t know”. Uncertainty about parliament rose ten points, about political parties eight, about state media eight, and the bulletin notes that distrust of state media redistributed into that answer entirely. Its authors state the two possible readings plainly: either distrust really has fallen, or people have become more afraid to answer a question about trusting institutions.

Something similar happened between people. Two years ago 23% agreed that most people can be trusted; now 15% do. Trust in one’s own family fell from 94% to 86%, in personal acquaintances from 78% to 69%, in neighbours from 50% to 42%. Eight points at every distance, including the closest one. The share who found the question hard to answer rose from 17% to 30%.

Distrust did not turn into trust. It turned into “hard to say”.

Confidence

Watch

• The arrears figure falls back toward its 2024 level of 6% in the next wave. It has not in eight waves.

• Job losses stay above 5% in the October wave, which would make two quarters a trend rather than a season.

• The share answering “don’t know” on institutional trust keeps rising in the next repeat of those questions.

• Real wage growth and the share of households noticing it converge. In July the gap was 9.1% against one in ten.

Method and limits

All figures come from the Monitoring of the Belarusian Economy project run by MacroBY.org: eighteen online waves of about a thousand urban residents aged 18 to 64 since December 2021, weighted to the urban population by sex, region and age.

Two limits are the project’s own. The sample is urban, so rural households are not represented. And the responses are self-reported at a time when a rising share of people decline to answer questions about trust, which the bulletin itself flags as a possible sign of fear rather than of changed opinion.

The comparison with EU countries uses the European method, which drops the question about the current state of the economy. The bulletin explains why: for EU countries that version predicts behaviour better. It is still a comparison between a survey conducted in Belarus and surveys conducted in states where answering carries no risk.

Human Constanta’s Digital Freedoms Lab has published a forty-page map of how the Belarusian state reaches its critics abroad through screens. Its one new dataset is small: 102,200 Telegram posts from eleven pro-regime channels. Those posts say the channels are staffed, fed from one place and cheap to run. They say nothing about how many people the channels reach.

The document

The report is called “Transnational digital repression in Belarus: from doxxing to deepfakes” (Digital Freedoms Lab, Human Constanta). It covers August 2020 to July 2026. Most of it is a case typology: fabricated films, fake bots and link traps, recruitment through relatives, DDoS attacks, published addresses, financial “audits” of exile NGOs built from Lithuanian company registers. We counted 62 dated cases in chapters two and three. Almost every one of them has appeared in the independent press before. The typology is a filing system for known material, and a good one.

The new material sits in chapter one. The authors pulled every post from eleven channels through the Telegram client API, from January 2023 to 12 July 2026, and asked a simple question. Are these channels a network, or eleven desks with the same taste?

What a shared post proves

The report tests the network idea two ways. Identical texts are merged into one record with a list of the channels that carried it. Forwards are traced through the fwd_from field, which keeps the numeric id of the source channel even after a rename.

The identical-text count looks strong at first. Kniga GU BAZA and Pervyi Rubezh share 1,066 posts. ZhS Premium and Puls Grodno share 740. Then the authors check each pair at message level. Of the 1,066, only four were posted by both channels on their own. Of the 740, none. The overlap is forwarding, not co-authorship.

Of 1,066 posts shared by two channels, four were written twice. The rest were forwarded once.

One pair breaks the pattern. Nevolfovich and ZhS Premium share 221 posts, and 54 of them are not forwards. Two channels with no formal link publish the same text on the same day, each from its own account. The report reads this as a common upstream source. That reading matches what we found in our own larger sample: five personal propagandist channels plus the state news channel and the presidential press pool, about 267,000 posts, all forwarding the same thirty or so state and Russian nodes with the press pool at the top. The Human Constanta sample sits one level below ours. Its channels carry police material, ours carry the presidential line. The shape is the same.

The structure chart is the report’s best single figure. Pervyi Rubezh, the Brest regional channel, is 56% reposts. Puls Grodno is 31%. Everyone else is a producer. ZhS Premium alone wrote 38,817 original posts in three and a half years, about thirty a day, and forwarded almost nothing. It is the source, not the hub.

The clock

The second test is the calendar. Own posts from all eleven channels cluster between 9:00 and 18:00 Minsk time on weekdays. For five channels the share inside those hours is 75 to 89 per cent. Activity drops after 22:00 and bottoms out between 3:00 and 6:00. Saturday is quiet, Sunday quieter.

The authors are careful here. A working-hours profile rules out volunteers posting after their day job. It does not separate a ministry press office from a commercial newsroom. Two channels widen the window into the evening, ZhS Premium and Nevolfovich. Belorusskiy Front never really sleeps. And Ludazhory, the AI-comics channel run by an SB. Belarus Segodnya columnist, posts 31.8% of its material at weekends, the highest share in the sample. The columnist works from home.

Five channels post three quarters of their material between nine and six. Nobody runs a hobby on that schedule.

What the cases are made of

The typology has one finding that runs against its own title. The data behind these operations almost never comes from hacking. The ByHelp recipient list came from a volunteer’s computer seized in Minsk. The Viasna training videos of 2019–2020 had been in the prosecutors’ hands since the Bialiatski case. The Kabanchuk screenshots were lifted from a Pastebin dump by an unknown hacker a month before GUBOPiK claimed them. The “leak” of the New Belarus platform was Polish-language company data from open registers. Financial doxxing works from rekvizitai.vz.lt, a public Lithuanian register that exile NGOs are legally required to feed.

This is the report’s real argument for host states. Nothing in this system crosses the threshold at which ordinary protection starts. There is no intrusion, no attacker inside the jurisdiction, no event a police report can name. The report calls this a design feature, and we agree.

The case curve needs a caveat the report does not give it. Dated cases rise from five in 2022 to twenty-one in 2025. Part of that is the method. The Telegram collection begins in January 2023, and the 2022 cases are the ones that made it into major outlets. A rising line drawn on a rising number of sources is not evidence of a rising practice.

Gender, counted elsewhere

The chapter on AI content lists eleven women targeted by Ludazhory and states that erotic synthetic material is used against women only, while men get alcoholism, treason and prison slang. The claim rests on reading the channel, with no counts. Our own coding of the FORESIGHT corpus on technology-facilitated gender-based violence gives it a number: 2,671 relevant documents since 2018, and 99% of the documents that are themselves acts of violence come from the state or pro-regime segment. The two readings agree. Only one of them can be checked.

Erotic fakes for women, alcohol and prison for men. The report saw it. We counted it.

What is not in the report

The eleven channels are described by post count only. There are no subscriber figures, no views, no forwards outward. A channel with 39,677 posts and a channel with 116 posts sit in the same table, and nothing says which one anybody reads. TGStat and Telemetr are cited for channel descriptions, not for audience.

Two of the eleven are quiet cases of substitution rather than continuity. Obratnaya Storona stopped on 28 January 2026. OS Exclusive started on 3 February, six days later, with the same format and a near-identical name. The report notes the dates and declines to claim a single editorial team. That is the right call and a missed measurement: the fwd_from id of the first channel would have settled it.

The numbers do not fully agree with themselves. The text gives the top three pair counts as 1,066, 740 and 725. The chart gives 1,065, 724 and 716. The difference is small and the direction is unexplained.

Human Constanta is a target in its own report. Its staff appear in a 2022 SB. Belarus Segodnya interrogation video and in a 2025 financial “audit” by Shtab Onoshko. The report does not say so in the methods section. It should, once, and move on.

The same organisation wrote the report and appears in two of its cases. That is a note for the methods section, not a reason to doubt the count.

Takeaways

The eleven channels are not one desk. They are one feed. Shared posts are forwards; independent co-publication exists in one pair only, and points upstream.

Posting follows a state working day. That excludes the volunteer model and nothing else.

The material is cheap. Seized devices, public registers and old leaks supply most of it, and the system is built to stay under the threshold where host-state law reacts.

The case curve reflects collection, not incidence. Treat 2022 as under-documented, not quiet.

Reach is unmeasured. Until it is, “network” describes the supply side only.

Watch

By 31 December 2026: the Investigative Committee names a TikTok or Threads post in at least one public case announcement. Its June 2026 statement promised monitoring; a case would show it is real.

By 31 March 2027: the Neurocomics series in SB. Belarus Segodnya passes 200 issues. It stood above 150 in July 2026 at roughly one a week. Fewer than 180 would mean the project slowed.

By 31 December 2026: OS Exclusive’s forward metadata shows forwards from the Obratnaya Storona id, or the old channel resumes posting. Either outcome closes the substitution question.

Method and limits. Channel figures are the report’s own (Table 1, Fig. 2–5), read from the PDF. The 62-case count is ours, from the dated cases in chapters 2 and 3; undated episodes are excluded and two-date cases count twice. The 267,000-post comparison refers to our earlier analysis of the pro-regime Telegram pyramid; the two samples overlap in period but not in channels, so the shape comparison is qualitative. The TFGBV figures come from a FactCheck.LT study. No reach data for the eleven channels was available to us either.

Last week we put two Belarusian registers on Anton Jäger’s diagram and followed one point across it for sixteen years. This piece reads the diagram the other way. Hyperpolitics (Verso, 2026) ends with a chapter called “Escape Routes”. Its answer is short. Politics with no organisations has one way out: rebuild the organisations. Belarus is a country where that has been tried twice since 2020, once by the exile and once by the state. Both attempts can be counted.

Jäger’s Figure 3 has two axes, politicisation and institutionalisation, and four quadrants. Mass politics sits high on both. Post-politics sits low on both. Antipolitics and hyperpolitics share the quadrant of high politicisation and low institutionalisation. The fourth quadrant, high institutionalisation and low politicisation, he gives to the pacified second phase of mass politics. Those were the post-war decades of parties and unions that kept their members and stopped fighting. Western countries, in his account, move through these quadrants one at a time. Belarus has a population in each.

Two weeks in the mass-politics quadrant

For most of August 2020 Belarus was in the upper-right corner. Strike committees formed at MTZ, Belaruskali, Grodno Azot and the Minsk Wheel Tractor Plant. The Coordination Council was founded on 14 August with a presidium of seven. The Golos platform had 812,000 registered users by 5 August, before the vote. Honest People collected photographs of 1,310 precinct protocols out of 5,767. Those alone gave Tsikhanouskaya more than 80 per cent of what the Central Election Commission had credited to her nationwide. That is what high on both axes looks like: mobilisation with a count, a structure and an address.

It lasted as long as it took the state to find the addresses. Presidium members were arrested or pushed abroad within five weeks. Strike committee leaders were detained in September. The independent trade union federation was liquidated in 2022. The parties were re-registered down to four in 2023. The register of public associations lost 56 per cent of its entries between 2021 and 2026. The order is the point. The state worked down Jäger’s vertical axis first and left the horizontal one to the prison and pardon system.

The state worked down Jäger’s vertical axis first and left the horizontal one to the prison and pardon system.

The exile in the hyperpolitics quadrant

Jäger describes hyperpolitical membership as low-cost, low-commitment and easy to leave. The Belarusian exile has now measured that three times with the same instrument. Golos in August 2020: more than 800,000 registrations in two weeks. The Coordination Council election of May 2024, the first attempt to give the exile an elected body: 6,723 votes from twelve countries. The Coordination Council election of May 2026: 2,113 votes, after a DDoS attack on the platform and a three-day extension. Belsat set that against seven million eligible Belarusians and printed a turnout of 0.03 per cent. Against the diaspora the denominator is kinder and the picture is the same. Estimates of how many Belarusians have left since 2020 vary. PACE and Freedom House cite 200,000 to 500,000. The deputy interior minister put it at 350,000 by mid-2023. The materials submitted to the International Criminal Court in early 2026 say about 600,000. On the lowest of those, 2,113 votes is 0.6 per cent of the diaspora; on the highest, 0.35 per cent. In 2024 the same arithmetic gave about 1.3 per cent. The share fell while the diaspora grew.

Pavel Latushka, whose list won both elections, told Zerkalo that Belarusians he met in Warsaw had not known there was a vote. That is an accurate description of hyperpolitics. Attention is available; it is simply not attached to the institution. Our earlier counts point the same way. Coverage of the New Belarus passport went from 520 documents at its January 2025 peak to one in July 2026. Babariko’s return in August 2026 reached 95 per cent of its views on other people’s channels. The exile still has the grin. The Council is the cat, and fewer people each year notice it is there.

The exile still has the grin. The Council is the cat, and fewer people each year notice it is there.

The interior in the post-politics quadrant

Jäger makes a point that is easy to miss. Post-politics and hyperpolitics share most of their features: a weak civil society, hollow parties and a state insulated from pressure. What separates them is the noise. Inside Belarus the noise is illegal. The extremist-materials register added 977 items in the first 225 days of 2026. The association register stands at 1,337. In 2023 the ministry inspected more than a thousand of the survivors and warned 410 of them in writing. The result matches Jäger’s post-political corner on every indicator except one. His post-politics arrived when people stopped caring. This one arrived by sentence, and the people who cared are the 1,135 on Viasna’s list.

The state in the quadrant Jäger keeps for the past

The fourth quadrant is where the argument turns. Jäger reserves high institutionalisation with low politicisation for pacified mass politics, the world of large memberships and small stakes. Belarus’s official society sits there today, and its size is on the record. The Federation of Trade Unions puts its membership at more than four million, 94.8 per cent of the working population, a figure it has carried on its site since 2016. The Belaya Rus association reports 256,986 members in 8,908 primary organisations; the party of the same name, created in 2023, reported 84,000 members at its congress this year. Those 8,908 primary organisations are close to a fifth of the 47,812 organisational structures the ministry counted on 1 January 2026 for the whole country. The five organisations that the ministry lists as “subjects of civil society” hold most of the rest. That is 36 branches for every surviving association.

Lukashenko described this quadrant himself. Receiving the head of the Federation on 17 July 2025, he said the unions were already “probably the most powerful party organisation” in the country. In organisational terms, he added, they rated higher than any party that exists. He was not complaining. A union that outranks the parties and never strikes is the institution he wants, and it is Jäger’s pacified mass politics built to order.

What this does to the escape route

Jäger’s exit from hyperpolitics is reinstitutionalisation: bring back the mass organisation, the party with dues, the union with a strike fund. Belarus shows what the prescription produces when only one axis moves.

Move institutionalisation up without politicisation and you get the Federation of Trade Unions: four million members, 8,908 Belaya Rus cells beside them, and a leader who reports to the president. That is Jäger’s escape route, completed, in the hands of the regime. Move politicisation without a territory to institutionalise it in and you get the Coordination Council. It is an elected body whose electorate fell from 6,723 to 2,113 while the population it represents grew. Both are cats. Neither has a grin.

Both are cats. Neither has a grin.

The only moment Belarus held both axes at once was August 2020, and it was the institutional half that the state removed first. Chris Marker’s film, which gives Jäger his chapter title, is about the decade after 1968, when the mood survived the organisations. Belarus is the case where the organisations were taken away on purpose, and the mood was allowed to leave the country. Jäger’s diagram describes that well. His escape route does not, because it assumes the state is absent from the drawing. In Minsk the state is the one holding the pen.

What to watch

The next Coordination Council election, due in 2028 on the current two-year cycle. A third fall, below 2,000 votes, would settle whether the 2026 result was the DDoS attack or the trend. A rise above 6,723 would refute this piece.

The Federation’s next reported membership. Its public figure has stood at “more than four million” since 2016; a dated update below that would mean the pacified quadrant is emptying too.

The Ministry of Justice line for 1 January 2027, due in February. If organisational structures pass 48,000 while associations fall again, the two lines are still moving apart.

Takeaways

• Belarus has a population in each of Jäger’s four quadrants at the same time. The state cut society along his vertical axis: strike committees and the Council in 2020, unions in 2022, parties in 2023, associations through 2024.

• Exile participation in its own institution fell from more than 800,000 Golos registrations in 2020 to 6,723 Council votes in 2024 and 2,113 in 2026. The number of Belarusians abroad rose over the same years.

• The state’s own organisations sit in the quadrant Jäger gives to pacified mass politics: 47,812 branches for 1,337 associations. The president calls the union the strongest party in the country.

• Jäger’s remedy of reinstitutionalisation has been carried out in Belarus by the regime; the exile’s version is losing its voters. Neither moves both axes at once, and August 2020 is the only point in the series that did.

Method and limits. Association and structure counts are the Ministry of Justice “as of 1 January” line as reconstructed last week. Golos registrations are the platform’s own counter as reported on 5 August 2020; the final figure was higher. Coordination Council votes are the election commission’s final results for 2024 and 2026 as reported by Nasha Niva, Reform and Zerkalo; the 2026 vote followed a DDoS attack and an extension, and the commission has not published registrations separately, so conversion from registered to voted cannot be checked.

The diaspora denominator is a range of published estimates (PACE, Freedom House, the deputy interior minister, Korshunau’s 500,000–600,000 for 2020–2024, and the ICC submission), as compiled by Belsat; none is a count.

Federation and Belaya Rus membership figures are the organisations’ own claims and have no independent check; the Federation’s page dates from 2016 and has not been updated. The Belaya Rus association and the Belaya Rus party are two registered bodies with separate memberships.

The 17 July 2025 quotation is from the president’s website as reprinted by pravo.by. Jäger’s diagram is used as he draws it; nothing here tests his account of Western Europe or the United States.

Anton Jäger’s Hyperpolitics (Verso, 2026) is built around one diagram. On the horizontal axis he puts politicisation: how much people mobilise, argue, protest and vote. On the vertical axis he puts institutionalisation: how far parties, unions and membership organisations channel that participation. Mass politics of the twentieth century sat high on both. The post-1989 decades sat low on both. The 2010s, in his account, split the two lines apart: politicisation rose while affiliation kept falling. The result is a politics of huge intensity and almost no organisation. His first chapter is titled after Chris Marker’s 1977 film about the decade after 1968, A Grin Without a Cat: the mood of a movement outliving the organisations that carried it.

Jäger draws that diagram by hand. He says so himself, calling it a synthesis of fragments from other people’s charts. Belarus does not appear in the book, and neither does any dictatorship. But Belarus is one of the few countries where both of his axes can be read from official registers, year by year, back to the 2000s. We did that.

What the two registers are

The institutionalisation axis comes from the Ministry of Justice. Every February it publishes one standing sentence: how many political parties, trade unions and public associations were registered on 1 January, together with their local organisational structures. We rebuilt the series for 2004–2026 from the ministry’s own page, from pravo.by and BelTA reprints, and from the annual reviews of the legal group Lawtrend, which quote the same sentence each year.

The politicisation axis is harder, because the state does not count mobilisation. We used the closest thing it does count: people the state has imprisoned for political reasons, as recognised by the human rights centre Viasna. The count on 1 January of each year comes from Viasna’s own Telegram channel, 4,473 posts read by a parser. For 2010–2019 the number is close to zero, and that is part of the finding.

Belarus is one of the few countries where both of Jäger’s axes can be read from official registers.

The institutionalisation line grew for ten years, then was cut

The number of registered public associations sat at about 2,200 from 2004 to 2010. Lawtrend describes those years as a frozen level. From 2011 it rose every single year: 2,325, 2,402, 2,477, 2,521, 2,596, 2,665, 2,731, 2,856, then 2,995 on 1 January 2020 and 3,021 on 1 January 2021. The peak comes after the protest year, when the protests were already over.

The fall starts a year later. On 1 January 2022 the ministry counted 2,978 associations. Then 2,544, then 1,973, then 1,397, then 1,337 on 1 January 2026. That is a loss of 56 per cent in five years. Trade unions went from 25 to 19 after the independent unions were liquidated in 2022. Parties went from 15 to 4 after the re-registration of 2023, and one of the four was created that same year.

None of this happened by people leaving. Lawtrend’s liquidation monitor counted at least 1,509 organisations closed by court or forced to close themselves between September 2020 and the end of 2023. In 2023 alone the ministry ran inspections of more than a thousand associations and issued 410 written warnings, one for every six organisations still on the list. The register also lags. Lawtrend’s own recount of live organisations at the end of 2023 found about 1,761 associations against the official 1,973, because organisations in liquidation stay on the list until they are struck from the company register.

The peak comes after the protest year, when the protests were already over.

The politicisation line rises and then leaks

Viasna’s count of political prisoners was 102 in October 2020. It was 169 at the turn of 2021, 969 a year later, 1,448 at the start of 2023 and 1,451 at the start of 2024. Then it began to fall: about 1,290 at the start of 2025, 1,135 at the start of 2026, and 909 by the end of March 2026 after 250 people were released in one US-brokered exchange. The number falls through pardons and deportations, and it rises through new sentences. Since 2024 the state has been running both valves at once.

Put the two lines on Jäger’s plane and Belarus traces a shape he does not describe. From 2010 to 2020 the point climbs the left edge: institutionalisation grows, politicisation in this measure stays at zero. In 2021 and 2022 it moves almost straight to the right, into the quadrant Jäger reserves for the mass politics of the past, high on both axes. It stays there for one year, because the state had not yet closed the register. From 2022 the point falls and drifts left at the same time. By 2026 it sits where Jäger places hyperpolitics: low institutionalisation, second axis still high. It is moving left. At the pace of the last two years, about 250 fewer prisoners a year, it reaches his post-political corner around 2030.

Jäger’s Western democracies reach that corner by drift. People stop joining, and after a while they stop caring. Belarus is being moved there by administration. One axis was cut by liquidation orders, the other is being lowered by pardon decrees and exchanges.

Where the state moved in

There is a second number in the same ministry sentence, and it runs the other way. The ministry also counts “organisational structures” of associations: local branches and cells. Those went from 35,634 in 2011 to 41,760 in 2018, 43,865 in 2021, 45,706 in 2023, and 47,812 on 1 January 2026. While the number of associations fell by more than half, the number of their branches rose by nine per cent.

The arithmetic is simple. The branches belong to the few organisations that survived and grew: the Federation of Trade Unions, Belaya Rus, the BRSM youth union, the veterans’ union and the Union of Women. In 2023 a law on the foundations of civil society gave a small set of organisations a special status, and the ministry’s list of those “subjects of civil society” is these five. Each surviving association now has on average 36 branches, against 15 in 2021. Lawtrend identifies the five as the main holders of those branches; the ministry does not publish the breakdown. Membership in Belarus moved into structures where joining is expected and politics is absent.

Membership moved into structures where joining is expected and politics is absent.

That is the fourth quadrant on Jäger’s diagram: high institutionalisation, low politicisation. He gives it to the second, pacified phase of mass politics, the post-war decades when parties and unions kept their members and lost their fights. Belarus’s official society sits in it now, a cat without any grin. The exile side of the same society sits in the opposite corner, where we have measured it before: attention without organisation, comebacks on other people’s channels, artefacts whose coverage fades within a year.

What the registers stopped saying

One more thing happened to the ministry’s sentence. Until 2021 it also gave the number of foundations and the number of new associations registered that year. From the 2023 edition both were gone, as Lawtrend noted at the time. The annual press release that used to carry the sentence, “On the registration of non-profit organisations in year X”, was last issued for 2020. Since then the numbers live only in a standing paragraph on the ministry’s website and in the president’s portal. The series still exists. The state just stopped announcing it in the year it turned down.

The series still exists. The state just stopped announcing it in the year it turned down.

What to watch

The 2027 edition of the ministry sentence, due in February 2027. If our reading is right, associations will be between 1,250 and 1,320 on 1 January 2027, and organisational structures will pass 48,000. If associations rise, the liquidation campaign has ended, as Lawtrend claimed in July 2025.

Viasna’s count on 1 January 2027. The last twelve months show a net fall of about 250 a year. Below 850 would confirm that the exit valve is now larger than the intake.

The foundation count. If the ministry restores it to the annual sentence, the 2023 omission was clerical. If it stays out for a third year, it was a choice.

Takeaways

• Registered associations peaked on 1 January 2021 at 3,021 and stood at 1,337 on 1 January 2026, a fall of 56 per cent through liquidation.

• Their local branches rose from 43,865 to 47,812 over the same years, held by five organisations that a 2023 law names as subjects of civil society.

• Viasna’s count of political prisoners went from 102 in October 2020 to 1,451 at the start of 2024 and 909 by the end of March 2026.

• On Jäger’s diagram Belarus crossed the mass-politics quadrant in 2021–2022 and now sits in his hyperpolitics quadrant, moving left. The quadrant he gives to pacified mass politics, high institutionalisation and low politicisation, is where the state’s own associations sit: 47,812 branches for 1,337 organisations.

Method and limits.

Institutionalisation figures are the Ministry of Justice “as of 1 January” line, reconstructed from the ministry page, pravo.by, BelTA, president.gov.by and Lawtrend’s annual and monthly reviews; 2004–2017 associations come from a table in Lawtrend’s 2016 review that reprints ministry data; 1 January 2019 is missing from our sources and the chart interpolates it.

The ministry counts organisations still in liquidation, so the official line runs above the number of live organisations. Political-prisoner counts are Viasna’s, read from its Telegram channel by a parser that keeps the value nearest to 1 January within ten days; where several posts disagree within a month, the modal value is shown.

Viasna is the only source for that axis in our corpus; Lawtrend’s monthly reviews, which repeat the number, are not in it. Both axes are administrative counts, one from the state and one from a group counting the state’s prisoners; neither measures mobilisation directly, and the politicisation axis is blind before 2020. The extremist-materials register (19 designations in 2020, 427 in 2021, 1,462 in 2024) supports the same reading and is kept out of the chart because it counts objects, not people.

Jäger’s book is used for its diagram, which he describes as intuitive; nothing here tests his argument about the West.

On 7 August Viktar Babariko published a video called “What does Viktar Babariko propose now?” on his own YouTube channel. The next day, at the sixth New Belarus conference, Pavel Latushka answered him in what Belsat headlined as a harsh reply. Within a week Belarusian YouTube had produced more videos mentioning Babariko than in June and July together.

This brief reads that comeback against seven months of counting. Since February, FactCheck.LT has filtered every video published by Belarusian YouTube channels for the names of four politicians: Sviatlana Tsikhanouskaya, Viktar Babariko, Maria Kolesnikova and Pavel Latushka. The monthly reports are on infopolicy.net, and the August numbers say two things at once: media attention to Babariko multiplied by eleven, while search interest from inside Belarus stayed at zero.

What the seven months look like

The monitoring covers between 12,000 and 15,400 videos a month, of which between 255 and 503 mention at least one of the four names, and views are then summed per politician.

Tsikhanouskaya has the highest peaks and the deepest falls. Her line runs from 4.0 million in February to 8.3 million in May, then down to 1.1 million in July and back up to 2.8 million in August. Latushka’s line moved inside a narrow band for six months, between 0.96 and 1.34 million, which in July made him the most-viewed of the four for the first time. In August he recorded his first decline, to 0.89 million.

Babariko’s line is a wave that gets weaker each cycle. After 2.2 million in February, each peak was lower and each trough deeper, until July marked the bottom with eight videos and 93,000 views. August is the third peak, at 1.03 million.

Each of Babariko’s previous peaks was lower than the one before. August broke that pattern.

Where the August views came from

Seventy-seven videos mentioned Babariko in August. They collected 1,026,000 views across 27 channels, and his own channel, “Viktar Babariko”, contributed four of them with 56,000 views, or 5.5 percent of his total.

The largest source was Euroradio with 19 videos and 223,000 views, followed by Nasha Niva with 166,000 views from only two videos. Belsat News, Zerkalo and Belsat Now added 306,000 between them. Nearly all of Babariko’s August audience watched him on channels he does not control.

The pattern inside his own channel is also clear. The programme statement of 7 August took 47,000 views, but three short clips posted between 18 and 20 August took 5,400, 1,600 and 2,500. The channel is older than August: it already appeared in the May sample with three videos, then went silent through June and July.

The channel was reactivated for one statement. Nothing suggests it was built to keep an audience.

The search test

Google Trends, restricted to Belarus, shows the same four names from 1 June to 1 September. The query is comparative, so all four lines share one scale, and the peak of the window is Latushka on 29 June, when a closed Putin–Lukashenko meeting at Valdai made him the most-searched of the four.

Babariko’s line is flat at zero for all 93 days. There is no bump on 7 August when the statement went out, none on 8 August when Latushka replied, and none on 19 August when Zerkalo’s Artyom Shraibman analysed the statement. The comparative scale hides small values, but it does not hide a return of this size. In May, Babariko’s Zerkalo interview registered 59 on the same scale.

Kolesnikova’s line is the only one with a daily signal in August. From the 19th, when Shraibman’s video called her a possible new leader, her index stayed between 6 and 14 every day. This is the fifth month she has shown that kind of baseline. Her YouTube footprint in August was 17 videos and 447,000 views.

One politician generated a million views and no searches. Another generated a fifth of that and was searched for every day.

Who is mentioned with whom

The filter also counts videos that mention two politicians together. In July no video mentioned both Babariko and Latushka, while in August 21 did. Tsikhanouskaya and Babariko appeared together in 34 videos, up from two. The four names had been drifting apart since May until the conference on 8 August pulled them back into a single conversation.

The Babariko–Latushka pair is different from earlier co-mentions. Previous overlaps came from shared events such as the Kyiv visit in May, whereas this one comes from a disagreement. Belsat’s headline of 8 August names the reply. Olga Tokarchuk’s channel framed it as “Babariko takes everyone apart, does not need Tsikhanouskaya”.

Two models that spent six months in separate lanes met in one room.

Confidence

What to watch

By 30 September. The “Viktar Babariko” channel has published at least eight videos in September. If it has, the August reactivation was the start of something; if it has not, it was a single statement.

By 30 September. A non-comparative Google Trends query for “бабарико”, geo=BY, shows any day in August above 20 on its own scale. If it does, the comparative flat line understated a real but small signal.

By 30 September. Videos mentioning both Babariko and Latushka number at least ten in the September sample. If fewer, the axis of disagreement did not hold.

By 31 October. Latushka’s own channel is back above 400,000 monthly views. In August it fell to 183,000 from 531,000 in July.

Method and limits

The YouTube filter matches all case forms of the four surnames in Russian and Belarusian, in video titles and descriptions. It does not read transcripts, and a video counts for a politician if the name appears anywhere in those two fields, so a news digest listing him in the description counts the same as an interview. The Tsikhanouskaya pattern also catches Siarhei Tsikhanouski, and August had an unusual share of videos about him.

Google Trends values come from a single comparative query. Each value is relative to the highest point among all four names in the window, so zero means the volume was too small to register against that peak rather than that nobody searched. The surname “Kolesnikova” is common in Russian, so part of her baseline may be namesakes. A topic-ID query would settle this and has not been run yet.

The politicians’ own channels are included in the counts, although FORESIGHT’s main corpus does not take them as primary sources. This series is a separate pipeline built on the YouTube API.

On 27 August Lukashenka told a government meeting that loss-making farms in Vitebsk region would go to stronger companies, including private ones. “Bluntly put, privatisation,” he said. He has refused that word for thirty years. State outlets quoted him, and then spent the same day explaining that this is not privatisation. What is being handed over is priced at one base unit, which today is 45 rubles, and what comes with it is debt.

What is changing hands

Ten farms in the Dubrovna district, together with the Vitebsky Bekon enterprise, go to the Servolux group. Each is priced at one base unit: 45 rubles, about 13 euros. The Orsha meat plant has already passed to the Dzerzhinsky agricultural combine, and farms in the Orsha district to Savushkin Produkt. Nobody published the terms of those two.

Uladzimir Kavalkin, who runs the Kosht Urada project, told Zerkalo what the price reflects. Many of these farms carry debts worth more than ten years of their own revenue. On his reading this is not a sale but a transfer of liabilities: the buyer is not getting an asset cheaply, the buyer is taking on the debt.

He also traces what came before. The same farms have been attached to the National Bank, handed to industrial enterprises, merged into holdings with profitable firms, and had their debts converted into local-authority bonds. Each attempt aimed to fix them. The debts kept growing.

Forty-five rubles buys a farm and the ten years of revenue it owes.

The word, said and unsaid

The vocabulary around the announcement is worth reading closely, because state outlets did not avoid the word. They quoted it and then took it back.

BelTA, pravo.by and Grodnonews all carried the sentence in full. ONT reported the same meeting and wrote that this is a transfer to management, or even a sale to private firms, but not privatisation. CTV, Mlyn and belarus-news each ran the same line: Lukashenka has always been wary of privatisation, and ordinary Belarusians have always backed him in that.

Then the word turns up again, approvingly, about other things. On 28 August SB reported that the National Agency for Investment and Privatisation is running road shows in China. On 29 August the same paper summed up the president’s week as one of fair privatisation.

So in four days the president said the word, the outlets that quoted him denied it, and the same outlets applied it approvingly to something else.

Spoken, denied, and then used approvingly about somebody else.

The buyer is not getting an asset cheaply. The buyer is taking on the debt.

Why the denial is not only spin

The denial has a case behind it, and it deserves a fair hearing.

Privatisation in the ordinary sense needs a price set by competition, a published procedure, and a buyer who wants the asset. None of that appears here. There was no auction. The price is a token. The recipients are three large processors that already buy from these farms, and the meeting’s own logic said they should now carry the farms that supply them.

But that argument cuts both ways. If this is not privatisation because there is no price discovery and no willing buyer, then it is a state directing private companies to absorb public debts. Kavalkin puts it plainly: there is nothing to privatise here except the debt, and doing it this way can only damage private business.

Both readings agree on the facts. They disagree about which word makes the state look better.

Both readings agree on the facts. They disagree about which word flatters the state.

Confidence

Watch

• Servolux, Savushkin Produkt or the Dzerzhinsky combine state publicly what they are taking on, and on what terms.

• The debts of the transferred farms are published, or written off, before the transfers complete.

• A second region is handled the same way. The meeting described Vitebsk as a model, and the president said the approach would be scaled to the whole country.

• State media use the word for a Belarusian transaction without a sentence explaining that it is not that.

On 15 July a Belarusian court declared two websites extremist: the EU4Belarus programmes and the Central Project Management Agency of Lithuania, the state body that administers them. On 28 July another did the same to the Polish Press Agency, Poland’s state news agency. On 3 August a third designated EUvsDisinfo, a project of the European Union’s own diplomatic service. Before this summer, the register had reached that class of target four times in four years. It reached it three times in three weeks.

A new kind of target

The register has listed foreign websites for years. Almost all of them were Belarusian: exile newsrooms, human rights groups, campaigns run by people who left. Bodies belonging to other states were rare.

There were four of them: Polish Radio in July 2022, a Lithuanian public broadcaster’s relocation site three months later, the website of Ukraine’s education ministry in March 2024, and the Council of Europe in December 2025.

Then came this summer, and the EU4Belarus programme site, the Polish Press Agency and EUvsDisinfo were all listed inside three weeks. Around them sits a wider shift. Foreign websites that are not Belarusian exile projects were designated five times in 2025 and fourteen times in the first eight months of 2026. Eight of those fourteen fall between 10 July and 6 August. They include Postimees of Estonia, Euronews, JAMnews, Ukraine’s Slovo i Dilo and the Polish ONW.

Four in four years. Three in three weeks.

What this does to the transnational reading

In August we argued that this register is the home end of transnational repression. It reaches nobody abroad. It makes contact with exile media a punishable act for people inside Belarus, so the tie is cut at the resident’s end, not the emigrant’s.

That still holds, and the new targets sharpen it.

Designating the Polish Press Agency does nothing to that agency, but it changes what a person in Homel or Brest may safely read. Designating EUvsDisinfo does not inconvenience the European External Action Service. It makes the EU’s own record of Russian and Belarusian disinformation an offence to share inside the country that record describes.

The EU4Belarus case is the clearest. That site is where Belarusians apply for EU-funded study programmes. The same ruling covered CPVA, the Lithuanian state agency that administers public investment and international development projects, along with its LinkedIn page and its account on X. Nobody in Vilnius is affected. What changes is that the page you use to apply for a scholarship now carries a criminal risk for the applicant.

The ban does not reach Warsaw or Brussels. It reaches the reader in Homel.

The tie is no longer only between those who left and those who stayed.

So the target class has moved from the diaspora to the institutions the diaspora deals with, while the effect stays exactly where it was. The tie being cut is no longer only between those who left and those who stayed. It is between residents and the public bodies of the countries next door.

There is a further oddity. These bodies appear to find out by other means. When Euronews was designated in August, its editorial director said the decision reached the channel without any notification. The register is published, but nobody tells the institution it now names.

A different bench does this work

The register holds 6,052 entries, of which 288 are foreign websites issued by 41 courts. That work is not spread the way the rest of the register is.

Minsk region issues 21.5% of all designations in the country and 2.1% of the foreign ones. Minsk city is the mirror image, with 17.3% of everything and 42.4% of foreign sites, while Mahiliou region sits at 9.6% and 16.0%. Foreign resources are a separate line of work, and only some courts do it.

In 2026 that line moved. Minsk city’s share of foreign designations fell from 42% across all years to 15% this year, while Mahiliou region’s rose from 16% to 37%. Minsk region issued none at all. Within Mahiliou the work sits in two courts: the Leninski district court with fifteen this year, and the Horki district court with ten.

This updates our own reading. In August we wrote that nothing was concentrated in Horki, which held eleven of the foreign designations and sixth place. Across the whole period that is still true, but for 2026 it is not. Horki began this work only in February 2025, and ten of its thirteen rulings fall in the current year. The concentration was not there when we looked, and it has formed since.

The concentration was not there when we looked. It has formed since.

Confidence

Watch

• Another state or intergovernmental body is designated before 31 October. Three in three weeks is a rate only if it continues.

• Mahiliou region keeps its lead in foreign designations through the fourth quarter, or Minsk city takes it back.

• Any designated foreign institution says it was told of its listing through official channels. When Euronews was designated in August, its editorial director said the decision came with no notification to the channel at all.

• The reasoning behind any of these rulings is published. It has not been in any case so far.

Method and limits

The register was parsed in full and holds 6,052 entries at the current pass. Foreign websites here means entries whose stored address sits outside the .by zone and is not a social platform account, with Belarusian exile projects on foreign domains separated by hand from the domain list.

The classification of what counts as a state or intergovernmental body is ours. Public broadcasters are included and commercial foreign media are not, so a different line would produce a different four, though not a different three.

Term starts today. Three days ago the expert club Nashe Mniennie published the record of a seminar on the militarisation of schooling in Belarus and Russia. The seminar ran under Chatham House rules, so no speaker is named, and I take part in the same series of discussions. For that reason this piece uses the report’s figures and does not judge between its participants. One of its conclusions deserves a wider audience: we can measure how far these practices have been installed, but not whether they have worked.

What can be counted

The installation is easy to count, and the numbers are large.

In 2011, higher education covered 98% of the relevant age group; in 2025 it covered 70%. About 60% of pupils in the ninth year now go on to vocational schools instead of academic ones, and the state has pushed them that way for several years. Of Belarusians studying abroad, more than 53% were in Russia in July 2026.

Field camps have moved the same way. In mid-August we counted how often state media mentioned them: 59 documents in 2024, 71 in 2025, and 280 in the first seven and a half months of 2026. That rise tracks the year the camps became compulsory for pupils in the tenth year and moved onto military bases. The education minister has now announced the next step: five-day camps become compulsory for college students too, and grow to ten days from the 2027/28 year. We set that as a watch item a fortnight ago, and it has already been answered.

Ninety-eight per cent to seventy. Fifty-nine documents to two hundred and eighty.

What cannot

None of this tells you what any of it has done to anyone.

Belarus has left the international school surveys that once allowed comparison with other countries. It has replaced them with a national test of its own, called NOKO. The second pilot ran in about 700 schools in November and December 2025, and Belarus now reports rising functional literacy, as Russia does.

Look at who does what there. The ministry decides what should be measured, builds the test, runs it, and announces the result. Nothing outside the system checks the answer.

The ministry decides what to measure, builds the test, runs it and announces the result.

The same gap runs through the rest. There is no representative Belarusian data on how pupils, parents and teachers actually receive military training, flag ceremonies or the change of language in class. The seminar was clear about this: a large apparatus is not the same thing as an effective one. Children’s military parades and photographs with weapons show that somebody organised them, not that anybody was convinced.

Why the difference matters

The two readings point to opposite futures. Only one of them has evidence behind it.

Suppose that installing the system means the system works. In that case a decade of this produces a generation shaped by it.

Now suppose the system is absorbed as ritual, with teachers reporting it, photographing it, filing it and privately discounting it. Then it produces something older and plainer: people who separate what they say from what they think. The late Soviet school produced exactly that.

The Belarusian participants leaned to the second reading, for a specific reason. Even the official line says this is not Belarus’s war. The militarist stories arriving from the east reach an audience that accepts them less readily than the Russian one. In those conditions the machine can be built to specification and still fail to do its job.

That is not a comfort. A system can fail to convince anyone and still succeed at everything else: it can sort children into tracks at fifteen, thin out the teaching profession, and remove the subjects where awkward questions get asked. None of that needs anyone to believe anything.

A system can fail to convince anyone and still succeed at everything else.

What would settle it

A survey would settle it. Not a state one, and not one paid for to confirm what its funders already believe. A properly sampled study of pupils, parents and teachers, broken down by region and type of school. It would ask about the war, the camps and the language of instruction, and it would publish its questions alongside its results.

It is the most useful thing anyone could commission about Belarusian schooling right now, and it is also the thing nobody has. Until it exists, everything written about how well indoctrination works in Belarus is a guess from the size of the apparatus. That includes what is written by people who agree with us.

Confidence

Watch

• The five-day college camps take place this term as announced, and state media report how many students attended.

• NOKO results for 2026 are published with the test, the sample and the scoring, and not only as a headline figure.

• A representative survey of pupils, parents or teachers on the military parts of schooling appears, from any source.

• The 2026 figure for higher education coverage is published and can be read against 70%.

Method and limits

The figures on coverage, vocational schooling, students abroad and the NOKO pilot come from the seminar report of 29 August. That seminar ran under Chatham House rules and names no participants, and neither do we. The field camp counts are our own, from our corpus, and we published them on 14 August.

I take part in the same series of discussions. This piece therefore reports the seminar’s figures and leaves its disagreements alone. The report records two of them: how deeply militarisation has been absorbed, and whether staying inside the system protects the profession or props the system up. We leave both as they stand.

Two figures need care. The fall in higher education coverage from 98% to 70% mixes three causes: a falling birth rate, the move of pupils into vocational schools, and emigration. No one of those causes explains it. And the 60% share of ninth-year pupils going to vocational schools is an expert estimate, which the report itself says needs checking.