--- title: Evidence: how Lorg Check was tested description: How we test Lorg Check, the results with their dates and sample sizes, the service's own live count, and the mistakes we made and fixed. url: https://lorglabs.com/evidence.html --- # Evidence a sceptic can check How we test, what we found, on which dates and on how many tokens. And where we were wrong. ## How we test - **Live**: New tokens are watched as they launch. - **Confirmed first**: Each suspected trap is confirmed by a plain wallet buying and selling. Under half back makes it a trap. - **Same seconds**: Every checker is asked about the same token in the same seconds. ## 39 live traps. Lorg Check blocked 38. Three sessions on Base, 4–5 October 2026. Each trap was confirmed first with a real wallet that bought, sold, and got nothing back. | 39 live traps | Called it a trap | No verdict | Couldn't check | | --- | --- | --- | --- | | Lorg Check | 38 | 1 | 0 | | GoPlus | 0 | 39 | 0 | | Honeypot.is | 1 | 0 | 38 | A small sample: 38 of the 39 are one scam family, a trap hidden in the pool instead of the token · pools of $129 to $1,592 · asked 3 to 27 minutes after each pool appeared ## A second kind of trap, live with $69,000 in its pool 9 October 2026, outside the sessions above. Our weekly scan of new pools turned up a trap built from a different template. A plain wallet that bought it and sold it a block later got 0.1% of its money back. All three services were asked in the same minute. | One live trap | Answer | | --- | --- | | Lorg Check | **Block** Selling back lost 99.9%. | | GoPlus | Not a honeypot. | | Honeypot.is | Couldn't check it: pair not found. | One token: an example, not a rate · it was 15 hours old when asked · two earlier copies of the same template are blocked too ## Look-alike wallets, planted to catch the next payment 14 October 2026, on Base. A wallet paid another one dollar of real USDC: a test before a bigger payment. Two minutes later a fake "USDC" token wrote two false entries into their histories, from and to wallets made to look like theirs: the same first and last four characters. A fake "ETH" gave one of the copies a made-up past, and the next night a fake "USDT", its name spelled with letters from other alphabets, planted a third. Whoever copies an address from their history next pays a stranger. | The three planted look-alikes | Answer | | --- | --- | | Lorg Check | **Block** Names the real address each one imitates. The three fake tokens are blocked too: "This is not USD Coin". | | GoPlus | No warning on any of the three. | | Webacy | No warning (one of them checked). | Said plainly: until that day our own check called them fine too. We built the rule the same day, and tried it on live traffic before it changed a single answer: in one slice of Base activity, 19 of the 27 never-used wallets around fake money were planted look-alikes of real addresses, and the honest wallets beside them were left alone. One case, and one sample: an example, not a rate · the same trick cost one trader about $50 million in December 2025 ## 27 new launches, and what became of them One session on Base, 5 October 2026: every launch with $1,000 or more in its pool half an hour in. | 27 launches | Block | Caution | Called fine | No verdict or couldn't check | | --- | --- | --- | --- | --- | | Lorg Check | 5 | 18 | 4 | 0 | | GoPlus | 0 | 0 | 25 | 2 | | Honeypot.is | 0 | 0 | 8 | 19 | Within two hours, 15 of the 27 pools had been emptied. We had blocked 5 of them and cautioned the other 10. GoPlus had called all 15 fine. The 4 tokens we allowed all still held their pools. The two-hour figures are our own reading of the pools on the day ## 103 launches followed for a week Every new token on Base from 18 to 25 September 2026 with $1,000 or more in its pool, checked as it stood 30 minutes in and then followed for a week. **77 were scams.** We blocked 51, said caution on 26 and allowed none. Every block was a scam. ## The sets we re-run before every change **58 of 58**confirmed honeypots blocked **27 of 27**well-known tokens allowed **124 of 124**sanctioned addresses flagged **0 of 50**normal wallets blocked ## New launches we check without being asked Every half hour the service picks a token just launched on Base, checks it exactly as it would for a customer, signs the answer and keeps it. A day later it looks again: was the pool emptied, pulled, or turned into a trap? Every answer is public, and each page of them is dated in Bitcoin, so nobody, us included, can change what we said or when we said it. **–**new launches checked unasked **–**answered "block" **– of –**blocked tokens gone a day later **–**pages of answers dated in Bitcoin Read from the service when this page opens What it doesn't show, said plainly: these are launches we chose (at most 48 a day: the ones with the most money in them, and one in three a copy of a known trap), not everything that launched. And "gone" means the pool was emptied, pulled, forged or turned into a trap; an honest token's pool can be emptied too. [How it works, and how to check it yourself.](https://lorglabs.com/docs-track-record.md) ## The service's own count, as it stands now These figures are read from the running service each time this page opens, and anyone can read them at `GET /v1/stats`. Before the first customers, nearly every check here is our own, and the count says whose they were. **–**checks run **–**answered "block" **–**different addresses checked **–**of them blocked Read from the service when this page opens ## We publish our misses. On 4 October we allowed a token whose pool was then emptied by the one wallet that owned its money. We traced why the same day, fixed it the same day, and since then we never allow a new token unless we have seen who can withdraw its pool. A day later, a trap we were asked about three minutes after it appeared got "unknown" instead of "block"; we now pick up new pools every minute.