(this is a long read, but it has to be, in the interest of remaining fair to the accused)I was asked by a couple forum members to look into this account connection. The evidence is not as clear as the ILuckyGuyI farm, but it still strongly suggests the accounts in question are owned by the same person.
A few of these users popped up on my sigban radar over the last couple months as the quality of their posts is not very compelling, and has been recognized as such by some members of the community. For instance:
Hmm... the only thing that surprises me is why no one has tagged you for spam yet. I'll do it.
I looked at your posts, they are completely incoherent, I don't know what word generator you use, but it doesn't look like AI because even AI understands the meaning of terms like Clásico, but you don't.
Try to ignore my posts and not reply to them, it’s not exactly a huge problem, but your spam is so incoherent that it’s annoying. If I were you, I’d think twice about what you’re doing, because right now you deserve a ban.
Part I. StylometryThe farm was
first identified on the Italian local board, where gbianchi performed a
stylometric analysis of the board's most active accounts, which is a way of capturing the "linguistic fingerprint" of each poster. Even though I don't know Italian, I read the translated version of the entire thread and his methods seemed to make sense.
What he found was 2 weak clusters of 2-3 accounts and 1 strong cluster of 8 accounts who are far more similar to each other in terms of writing style than the other users. The cluster can be seen in the bottom right corner in this heatmap comparison of user stylometry scores:
Qui si vede bene l'enorme somiglianza di tutto il gruppo cluster 3
The strongest overlaps of stylometry scores are found in that bottom-right cluster, and most of the other overlaps are less significant (<50% overlap between most other accounts).
Among the measurements taken into consideration when analyzing sigban candidates are
unique word ratio (a measure of how often a user is likely to repeat the same words) and
semantic repetition (how often the same ideas or concepts are being expressed regardless of the words used to do so). Together they account for roughly 16% of a user's total "post health score".
Finally there is
ngram repetitiveness, which is a measure of the predictability of a user's words in a sequence. For example, a poster that writes "I agree with you completely" in half their posts will likely have a lower ngram score than someone who varies their sentence structure more often. Campaign managers, LoyceV and DdmrDdmr all have artificially low ngram scores because of their regular posting of template content. Other posters have low ngram scores because they write the same thing over and over, which often includes low-quality content posted for the sake of filling a campaign quota. This metric comprises about 6% of a user's post health score.
These 3 metrics can also be used as a form of linguistical fingerprint, unique to each person, as they hardly budge across any measurable length of a user's post history.
The first Bitcointalk user to identify two alt accounts using stylometric analysis was
tspacepilot, who used it as evidence that
Panthers52 was the alt account of Quickseller.
We're going to download the corpus of posts of Quickseller, ACCTSeller (his outed alt), Panthers52 (his accused alt), hilariousandco, dooglus, and me. We'll then build language models using maximum likelihood parameter estimation for all of the ngrams in each corpus up to n=3. For those who don't know, 1-grams are all of the single word tokens in the corpus, 2-grams (called bigrams) are all of the word pairs, 3-grams are all of the word triples, etc... For this project, a 3 gram model seemed appropriate...
...
Well, as you can see, qs' model predicts my corpus with a perplexity of 382, predicts hillarious with 263, predicts dooglus with 361. But crucially, predicts the posts of ACCTSeller and Panthers52 at 105 and 118!!!!
What this means is that QS's posting style, when measured quantitatively shows through his attempts to hide what he was doing.
This finding was later confirmed by Quickseller.
I performed a stylometric analysis of 6 of the accounts mentioned above and compared it to an analysis of 6 random, highly-active Italian board accounts (note that this is an analysis of English-language posts only), and this was the result:
"Cluster 3" AccountsRandom Italian Board Accounts| username* | ngram rep | sem rep | UWR |
| ---- | 93.5 | 80 | 79.4 |
| ---- | 64 | 95.2 | 100 |
| ---- | 93.4 | 70.8 | 94 |
| ---- | 88 | 73.1 | 57.6 |
| ---- | 93.1 | 83.2 | 91.1 |
| ---- | 91.3 | 61.4 | 83.9 |
| |
| st. dev | 11.6 | 11.6 | 15.0 |
What we can see here is that the Cluster 3 account ngram scores are remarkably similar to each other compared to the random group, while semantic repetitiveness and unique word ratio scores are also highly similar compared to the random group. I'm certain these observations would persist if we added the other two Cluster 3 accounts, or expanded the random group to include more users.
*
I removed the usernames of this batch of accounts for their privacy, but would be more than happy to share all data pertaining to how these scores were calculated upon request, to a neutral party.Part II. Posting ScheduleAnother interesting thing is the posting schedule of these accounts, which follows a format we've come to be familiar with. These are heatmaps for posts made between June 1st and Sept 1st of this year:
m4r1o
changaa
banana33
giorgione
cande86
verdinio
So we can see the accounts post in a somewhat interchangeable sequence, with m4rlo usually posting earliest in the day and cande86/verdinio tending to close it out. Interestingly, these accounts almost never post on Tuesdays -- a unique habit to share among 6 accounts in the same local board.
Part III. TimestampsI fed Grok the post timestamp history for all 5 accounts and asked it to perform an analysis, and it concluded that "the overall pattern is one of synchronized, overlapping sessions rather than independent browsing." Using the prompt, "Of all the dates I provided you, which ones are the most exemplary of multi-accounting patterns?", I got the following result (I verified these examples as accurate):
Top-tier examples (classic rigid pattern)
These show all 5 accounts active, high volume, continuous 8–15 minute interleaving, and clear sequential cycling through the same small set of core threads:
2026-03-05
The purest early example of the “round-robin cycling” pattern. Accounts hand off posts in the same threads (5176196, 5271157, 1159855, 5545447, etc.) with very regular short gaps throughout the day.
2026-05-21 & 2026-05-22
Excellent late-period examples of the original style still in full force — dense multi-account sessions with tight turn-taking and heavy reuse of the core thread set.
2026-01-07, 2026-01-08, 2026-01-14, 2026-01-15, 2026-01-21, 2026-01-22, 2026-01-28, 2026-01-29, 2026-01-30
Extremely strong January cluster. Many of these days feature near-continuous activity with all five accounts participating in the same threads minutes apart.
2026-04-01, 2026-04-02, 2026-04-09, 2026-04-15, 2026-04-16, 2026-04-23, 2026-04-24
Classic high-intensity April days showing the same mechanical interleaving and thread cycling.
To give you an idea of what its talking about, here is the timestamp history for a few of the dates mentioned above:
=== 2026-03-05 (5 accounts active) ===
09:52:44 user 3624576 thread 1159855
10:05:24 user 3626674 thread 1159912 (+12.7 min)
10:18:48 user 3648901 thread 5176196 (+13.4 min)
10:46:11 user 3626679 thread 5176196 (+27.4 min)
12:09:10 user 3626674 thread 5271157 (+83.0 min)
12:21:08 user 3648901 thread 5271157 (+12.0 min)
12:33:15 user 3648901 thread 1159855 (+12.1 min)
14:00:18 user 3624576 thread 5545447 (+87.0 min)
14:26:45 user 3626679 thread 5176196 (+26.4 min)
14:41:31 user 3626679 thread 5271157 (+14.8 min)
14:53:10 user 3624576 thread 5176196 (+11.7 min)
15:19:28 user 3657771 thread 5176196 (+26.3 min)
15:31:18 user 3657771 thread 5271157 (+11.8 min)
15:54:10 user 3648901 thread 5176196 (+22.9 min)
16:22:31 user 3626679 thread 1159855 (+28.4 min)
16:48:52 user 3657771 thread 1159855 (+26.4 min)
17:03:28 user 3626679 thread 5271157 (+14.6 min)
17:26:49 user 3648901 thread 1159855 (+23.4 min)
17:40:38 user 3626674 thread 5545447 (+13.8 min)
17:42:44 user 3657771 thread 5576470 (+2.1 min)
17:52:49 user 3626674 thread 1159855 (+10.1 min)
18:07:22 user 3657771 thread 5545447 (+14.6 min)
18:19:42 user 3657771 thread 5575988 (+12.3 min)
18:34:51 user 3626674 thread 5431360 (+15.2 min)
21:45:56 user 3624576 thread 5176196 (+191.1 min)
21:58:40 user 3624576 thread 5267086 (+12.7 min)
22:12:59 user 3624576 thread 5431360 (+14.3 min)
=== 2026-05-21 (3 accounts active) ===
09:54:07 user 803429 thread 5248531
10:59:15 user 803429 thread 5347904 (+65.1 min)
11:26:33 user 808310 thread 5583369 (+27.3 min)
21:28:07 user 816893 thread 5582888 (+601.6 min)
21:41:47 user 808310 thread 5583622 (+13.7 min)
22:03:49 user 803429 thread 5347904 (+22.0 min)
22:20:04 user 816893 thread 5516918 (+16.2 min)
22:29:29 user 803429 thread 5583157 (+9.4 min)
22:43:08 user 808310 thread 5583585 (+13.7 min)
22:57:51 user 803429 thread 5091185 (+14.7 min)
23:16:42 user 803429 thread 5347904 (+18.9 min)
23:22:35 user 816893 thread 5582146 (+5.9 min)
23:38:11 user 803429 thread 5248531 (+15.6 min)
23:45:53 user 808310 thread 5581901 (+7.7 min)
23:56:20 user 803429 thread 5091185 (+10.4 min)
=== 2026-04-16 (5 accounts active) ===
08:51:07 user 3626679 thread 5271157
09:01:57 user 3657771 thread 5545447 (+10.8 min)
09:13:34 user 3626679 thread 5545447 (+11.6 min)
09:25:59 user 3626679 thread 5176196 (+12.4 min)
09:47:49 user 3626679 thread 1159912 (+21.8 min)
09:58:56 user 3626679 thread 1159855 (+11.1 min)
10:11:13 user 3626679 thread 5176196 (+12.3 min)
10:22:56 user 3624576 thread 5271157 (+11.7 min)
10:34:31 user 3624576 thread 1877785 (+11.6 min)
11:03:28 user 3626679 thread 5545447 (+28.9 min)
11:39:11 user 3624576 thread 5267086 (+35.7 min)
11:50:32 user 3624576 thread 5176196 (+11.3 min)
12:01:50 user 3624576 thread 1159912 (+11.3 min)
12:12:44 user 3624576 thread 5545447 (+10.9 min)
12:24:36 user 3657771 thread 5271157 (+11.9 min)
12:43:51 user 3648901 thread 5545447 (+19.2 min)
13:03:40 user 3626674 thread 5271157 (+19.8 min)
13:14:47 user 3648901 thread 5176196 (+11.1 min)
13:25:19 user 3657771 thread 5267086 (+10.5 min)
13:35:02 user 3648901 thread 5271157 (+9.7 min)
13:45:07 user 3657771 thread 5549160 (+10.1 min)
14:09:17 user 3657771 thread 5176196 (+24.2 min)
14:19:51 user 3657771 thread 5545447 (+10.6 min)
14:50:21 user 3648901 thread 5492359 (+30.5 min)
15:01:32 user 3626674 thread 5176196 (+11.2 min)
15:13:53 user 3626674 thread 5267086 (+12.3 min)
15:34:22 user 3626674 thread 1159912 (+20.5 min)
15:35:58 user 3648901 thread 1159912 (+1.6 min)
15:44:27 user 3626674 thread 5545447 (+8.5 min)
15:49:08 user 3648901 thread 5545447 (+4.7 min)
You can see there is a noticeable even-ness in post spacing and threads are sometimes shared by 4 or 5 accounts. While the accounts don't always post in rote sequence (as in the ILuckyGuyI case), the spacing between timestamps is unmistakably similar from day to day.
Part IV. SeclogFinally, we have recent
seclog activity which shows that 4 of the 6 accounts changed their password within 53 minutes on October 5th, while a 5th account changed their password the next day:
| Account_ID | Username | Password Changed |
| 3624576 | giorgione | October 5, 2026 – 09:54:53 AM |
| 3626674 | cande86 | October 5, 2026 – 09:59:44 AM |
| 3648901 | banana33 | October 5, 2026 – 10:02:32 AM |
| 3657771 | changaa | October 5, 2026 – 10:05:43 AM |
| 3626679 | m4r1o | October 6, 2026 – 09:12:34 AM |
Not sure much more needs to be said here.
Part V. ConclusionI believe these pieces of evidence paint a strong picture of multi-accounting when taken together as a whole. I also believe that by reading through these accounts' post histories, one could independently come to the conclusion that they are operated by the same person -- given their similar grammar, word choice, post structure and subject material.
After a final review of their post history, I decided to issue a 360-day sigban to all 6 accounts.
Since the multi-accounting is not being proven with blockchain evidence, I will leave these accounts a neutral tag only for now.
