Collapse Watch №5 — One Wallet Pulled 7.7 Million LBC in Three Blocks. 6.2 Million of It Came Off Flagged Videos That Stayed Up.

7,710,000 LBC left Odysee's stake ledger in three consecutive blocks. On October 26, 2026 (LBRY blocks 2209636–2209638; block timestamps are not strictly ordered and run from 02:32:37 to 02:45:39 UTC), 42 stakes carrying a single channel signature — @Monolith — were withdrawn in one burst. Thirty-five of them, 6,210,000 LBC, sat on videos in our flagged corpus. Those videos were not removed. They still resolve live on Odysee; only the money left. This issue traces where the wallet behind that signature has been, where the coins went, and what the signature does to our own earlier reporting — which requires two corrections, set out in sections IV and V.

I. Three blocks

Of the 42 withdrawn stakes, 41 were supports and one was an abandoned repost; together they total 7,710,000 LBC. Thirty-five sat on videos in our flagged corpus (6,210,000 LBC). We traced the funding of 24 of them, exactly 5,010,000 LBC; those 24 break down as follows:

The other 11 stakes in our corpus (1,200,000 LBC) carry the same signature. Nothing was taken down. The channel pages and videos above still load. What changed is the number attached to them — and, for any viewer or ranking system that treats stake as a signal, what that number appears to say.

II. What @Monolith is

@Monolith is a channel (claim id f7053d0e7d76cdd4c9b447935372485e43e558e9) created April 27, 2021 at block 952451 and last updated January 16, 2022. It has no title, no description, no email and no website. It has 81 uploads (May 2021 to August 2025) and 247 reposts (January 2022 to June 2026), led by four channels (107 of the 247): @Qwinten (36), @CosmicEvent (30), @TheSearch4Truth (28) and @Fingerbob (13). Its own stake on the channel is about 101 LBC.

The 24 traced supports were placed between January 2 and June 26, 2026: a series of placements over six months, with individual transactions a few days apart, funded by recycling earlier bulk withdrawals. Shared inputs and change chains alone link the 24 into six groups; self-support and withdrawal links — a support can be withdrawn only by whoever holds the key — merge those six into one wallet cluster. The same cluster funded @Monolith's reposts, whose signatures Odysee's hub validates, so the wallet holds @Monolith's key. That is the whole of what the chain shows. Nothing on chain, and no self-published statement we found, says who runs @Monolith, and we do not name anyone: this issue refers to "the @Monolith wallet" and "one wallet." @Monolith's signature is also on the top stakes of @OzFlor, @StopTheCrime and @louismarschalko.

III. Twenty-one withdrawal events, 67 million LBC — recycled, not new

October 26 was not a one-off. Our chain scan finds 21 withdrawal events of @Monolith-signed stakes between December 2024 and October 2026: 377 withdrawals, 67.0 million LBC cumulative. Eighteen of the events were bulk (two or more withdrawals at once); three were single withdrawals, one of them in December 2024. That figure is a running total of the same coins being withdrawn and placed again, mostly the same coins each time. It is not 67 million LBC of new money. The largest prior events:

The scan covers one-input, one-output withdrawal shapes only, and covers 2025–2026 fully; it is a floor on the activity, not a ceiling. Earlier years were not scanned.

IV. Where the money went after October 26

As of block ~2222368 (November 4), 39 of the 42 withdrawn outputs — 5.51 million LBC — have not moved. Three (1,000,000, 1,000,000 and 200,000 LBC) were spent straight back into new @Monolith-signed supports, the remainder of each returned as unspent change:

• A 500,000 LBC tip to the @amaterasusolar channel's own address, 81 seconds after the withdrawal (block 2209649).
• 500,000 LBC staked on a claim titled "Saturn-cult-world-order" (@BRTaylorMataphysics), plus a 100,000 LBC tip to the same claim.

Five further @Monolith-signed placements were funded from other coins: a 90,000 LBC tip to the same @amaterasusolar address in block 2209635 — before the burst, funded from the change of an earlier repost transaction — and 50,000 LBC tips each to @Cahlen, @LadmisPan and @CosmicEvent, plus a 10,000 LBC tip to another claim. About 590,000 LBC is now held at the @amaterasusolar address, unspent. Across all eight placements about 1.35 million LBC was re-staked, of which about 850,000 LBC was tips. We found no exchange deposit. @amaterasusolar was created December 6, 2020 and has 867 claims; its own spending pattern differs from the cluster, and it is a recipient — we have not shown it to be the same wallet.

The other 18 stakes in the burst carry the same @Monolith signature. They sit on claims of other channels — @OzFlor (200,000 LBC), @pabloemege (1,000,000 LBC), and 100,000 LBC each on @FreeMax, @III, @LadmisPan, @StopTheCrime, @TruthOnlyMedia, @anonyme, @louismarschalko (two stakes), @palurintano, @Qwinten, @BRTaylorMataphysics (two), @TheRedneckPreppy, @FlowerPower3 and @uniónhispanoamericana — plus a 0.01 LBC abandoned repost. Eleven of them are among the 35 stakes in our corpus; together with the 24 traced stakes they make the 6,210,000 LBC item-level drop our scanner recorded across the corpus in that interval. @pabloemege's claim had been abandoned in April 2026 and @BRTaylorMataphysics has no items in our flagged corpus, so neither is among the 35. Not every channel named here is on our self-stake lists, and none is shown to own the stake on its claims.

V. What this means for our own reporting: a correction

We checked the signature on each listed channel's top stake. Of the 54 channels on our October 5 list, 13 had their top stake placed by @Monolith. Those 13 channels hold about 70% of that list's 5,165,302 LBC pool (their whole pools); @Monolith's verified top stakes alone are about 62%. On the current 37-channel list, 10 did: those 10 channels hold about 81% of the peak pool (their top stakes alone, about 72%) and 46% of the live pool. A further 8 (October list) and 6 (current list) channels had their top stake signed by another channel. Twenty (October) and 11 (current) were genuine self-stakes — signed by the channel itself or funded from the claim's own address. The rest were unsigned (funder unknown) or the claim's own bid.

The thirteen entries whose top stake was signed by @Monolith (@Unknown is an anonymous claim, not a channel), all of which our articles named as self-stakers: @Unknown, @ChristinatheAstonishing, @OzFlor, @Qwinten, @thisworldworks, @Plazma, @OurFreeSociety, @StopTheCrime, @AussieFighter, @ceylon, @i_malcontent, @louismarschalko and @thecrowhouse. @CosmicEvent (top stake 1,000,000 LBC), @amaterasusolar and @yellowgenius carry the same signature but are not on the formal lists. Two further channels our articles named, @TonyHeller and @woeih, had their top stakes signed by different channels (@Hantulio and @Syncretism respectively).

Some self-stakes held up. @neohumaneve's 495,000 LBC is its own wallet's: the same 495,000 LBC has moved across its own videos since August 2025, and on October 9 it was placed again under another channel that @neohumaneve controls, @MagentickA. @RedPillPhilosophy, @LogicDecoded and @jeremysITlab are genuine self-stakes as well; @TheNorthStateBiker's figure is its own claim bid. @TheBornGenesis's 3,009,483 LBC is claim bids and is unchanged and correct as published.

Those figures describe the lists we published. After we restored 73,679 items to the dataset on November 4 (section VI below), the analyser's list grew to 55 channels, and our new automated attribution classifies 24 of them as confirmed self-staked (19 signed by or paid from the channel or claim, 5 the claim's own bid), 22 as staked by another channel and 9 as unsigned, mixed or with the funder unknown. A signer counts only if it controls at least half of the stake. Those labels now appear on each channel's page and in /data/stake_concentration.json after the next scan.

Two caveats bind these numbers. We checked each channel's top stake only, and for unsigned stakes we traced the funding one hop. And we found no on-chain link between the @Monolith wallet and the named channels' own wallets, so those stakes cannot be attributed to those channels' operators. Each affected article now carries a correction note. The full method is in the methodology.

VI. A second correction: our retention cliff

The scanner's 180-day retention rule removed items from the public dataset when the keyword search had not re-found them in 180 days. That is not removal from Odysee. Between October 17 and November 2 it moved 74,453 items — the April 2026 deep-sweep cohort, all but 29 of them between October 17 and 22 — out of the public set. On November 4 we re-checked them: 74,448 resolved, of which 73,679 are still live and have been restored and 769 no longer resolve; the other 5 did not resolve either way. The 769 are held in our archive as not found at restore, and are not yet counted among the 199 two-strike confirmed removals below, which require two consecutive misses in the public set. We have not classified why those 769 stopped resolving; in earlier issues, items that vanished had been abandoned by their own uploaders rather than moderated.

The public set now holds 105,820 items. Of those, 99,391 are confirmed live at last check, 199 are confirmed removed (two consecutive misses) and 6,230 have not yet been re-checked. The retention rule now keeps any item confirmed live by a direct check in the last 45 days, and stale items are re-checked on rotation. The homepage no longer claims 0% or 100%.

The same rule explains a second number. Our self-stake count fell from 53 to 37 between October 16 and 22 because fewer channels were being analysed — 3,586 down to 1,368 — not because channels stopped.

VII. The platform in the window, October 11 – November 4

Traffic recovered. Odysee's Tranco rank improved from 9,930 on October 10 to 9,126 on October 22 — the best reading in the roughly 37-day window we hold — and has been flat since (9,208 on November 3). It has been inside the top 10,000 since October 10 (9,930 that day, 9,864 on October 11). Over the same period Odysee rose about 800 places, Rumble about 120 (2,699 to 2,582) and BitChute fell about 330 (7,870 to 8,203). We reported the opposite direction in №4; this is the other half of that record.

Uploads are down. On-chain stream claims per day averaged 7,397 in late September (September 19–30), 5,389 on October 1–10, and 4,876 from October 11 to November 3 — 34% below late September. October 18–26 averaged 4,019 a day, with lows of 3,377 (October 24) and 3,438 (October 18). A one-off spike of 8,853 and 8,669 on October 27 and 28 came from bulk publishers. Distinct publishers per day stayed flat, at 1,085–1,392 against 1,065–1,272 in September, and new channels run at roughly 176–313 a day, unchanged. Our inference, not a measurement: fewer uploads from bulk or automated uploaders, not fewer uploaders.

YouTube sync is broken. Channels that relied on it stopped updating. Their newest uploads: @SaltyCracker September 19, @TheBabylonBee September 18, @timcast September 18–19, @PaulJosephWatson August 30. EEVblog, thequartering and Veritasium were indexed in a single catch-up batch on October 27–28, then nothing. Three users say so in Odysee's own roundup threads (UTC, verbatim):

• @gordonp57, Nov 3, 20:33: “Channels are still not updating”
• @SuperHacker, Nov 3, 22:11: “YouTube Sync is broken yet again. BitChute is able to keep it working for years now.” (The comment goes on to other complaints.)
• @stevethefishdotnet, Oct 11, 13:48: “You're right. I used to watch Vara Dark, Salty Cracker, GTV Japan, Babylon Bee, Paul Joseph Watson often on here.That all ended about 22 days ago.”

Other complaints in the same threads concern payments and content quality. @KevCalner, Nov 3, 22:31: “I agree, I’d like for Odysee to at least enable regular cash options.” @InkTeeVee, Oct 11, 00:34: “Odysee should do us a massive favour and ban the low effort spam AI slop permanently we don't want it to clutter the LBRY network”

Official silence. “Channel Roundup: October 2026” was posted November 3 at 20:00 UTC. By 23:39 UTC it had 35 comments and no official reply. The September roundup thread has 292 comments and no official reply.

The apps. The Google Play listing still reads “Updated on Aug 5, 2026,” at 2.73 stars across 6,840 ratings. Nathaniel Mitchell, 1 star, November 1: “Absolutely abysmal app. Can't watch a single video. Every video just endlessly buffers.” Baldy Grey, 1 star, October 26: “Many videos don't play at all, and this extends to the web version too, so I'm not sure if this is particular to the app but it doesn't seem like it.” The App Store build is still v0.8.2 (September 9) with one new review in the window — an Australian 2-star on October 12 saying AI content can't be hidden.

Developers. odysee-frontend commits resumed on October 12 ("Restore merging of server-side watch history," #3584) and October 14 (a publish NaN-deposit fix, #3586) after silence since September 21; odysee-api had four commits on October 16. On November 3, lbryio/lbry.com was rebuilt as a static site, removing an obsolete Bittrex trading FAQ.

What we did not find. No evidence of a hack or breach: nothing on HaveIBeenPwned, no GitHub advisories, no change to domain registration or nameservers, no unexpected certificates or subdomains. odysee.com is registered through October 11, 2027. Press coverage remains zero. The parent ecosystem's AR token fell from about $4.09 on October 10 to $3.17 on November 3, roughly 22.5%.

The Texas case. Odysee is not a party to MDL 4:26-md-03176 (E.D. Tex., Chief Judge Mazzant). A case-management conference was held October 15; an Initial Case Management Order followed on October 21, and the amended schedule sets a preliminary-injunction hearing for January 11, 2027 (a January 13 date in an October 15 order was withdrawn the next day). Through October 21, in the filings we could retrieve, none mentions Odysee or LBRY; docket entries after about October 21 were not retrieved. A motion to strike (Dkt. 89) was taken under advisement at the October 15 conference, and we found no ruling on it in what we retrieved.

VIII. What should happen

Odysee should disclose how stake and stake withdrawal feed ranking and recommendations; label content whose stake was placed by a third party rather than the channel; and publish a moderation contact that answers. As the record above shows, a wallet can withdraw millions of LBC from videos within a few blocks, and nothing a viewer sees distinguishes a creator's own stake from a stranger's. Two comment threads on the company's own posts hold 327 comments between them and no official reply.

Regulators should note that, as we set out in our search-ranking analysis, DSA Article 27 requires platforms to disclose the main parameters of their recommender systems, and that no Article 15 transparency report has been filed. If stake is a ranking input, its mechanics belong in that disclosure.

Methodology & verification notes

On-chain figures are from the LBRY blockchain: block heights, channel signatures and funding transactions as cited above, with the unspent-output check made at block ~2222368 on November 4. Wallet clustering uses shared inputs, change chains and self-support plus withdrawal; it establishes a common key holder, not an identity. Channel signatures are read from the claim and support scripts on chain; we did not cryptographically re-verify them ourselves. Upload counts are on-chain stream claims per day by creation date. Tranco ranks are from its public API and cover roughly a 37-day window. Quoted comments are verbatim, with UTC times, from the official roundup threads. Labelled inference: the reading of the upload decline as fewer bulk uploads. Labelled limits: top-stake-only attribution checks; funding traced one hop for unsigned stakes; withdrawal scan limited to one-input, one-output shapes; signatures read from scripts, not re-verified. The datasets are published at /data/flagged.json and /data/stake_concentration.json; the full method is in our methodology.

Related coverage on OdyseeWatchdog

• Collapse Watch №4 — The Domain Is Paid Through 2027. The Site Was “Unusable All Week.”
• Collapse Watch №3 — After 33 Days of Silence, Odysee Spoke. It Recommended Some Gaming Channels.
• Collapse Watch №2 — Day 48: The Flatline Holds, and the Only Stake Still Moving Is One Recycled Pot
• Collapse Watch №1 — Six Weeks After “It's Over Now,” Odysee Is Running on Autopilot
• Self-Stake Seventh Reading — The Counter Didn't Move. A Million LBC Did.

Published November 4, 2026 · OdyseeWatchdog Investigative Team