Hello, Codeforces!
The situation involving AhmetKaan has generated some discussion. I would like to summarize the information we have reviewed and explain the reasoning behind our final decision.
About evidence
First of all, it is important to note that we cannot have absolute proof in such cases. We cannot see private conversations with AI tools, chats with other people, or other external sources of help. The only reliable data available to us are the submissions. Because of this limitation, some decisions are based on some red flags we see rather than a single definitive proof. After the appeal, we reviewed the case and collected the following observations.
1. Codeforces Round 1085
- The submissions 365891594 and 365892799 look suspicious to me. The explanations provided so far did not fully resolve the concerns.
- More than that he solved E2 in only 4.5 minutes. Only a few legendary grandmasters achieved a similar speed. At the same time, there was a noticeable loss of time due to implementing essentially the same logic in a different form in the following submissions: 365890875, 365893579
2. Codeforces Round 1082
Reviewing earlier rounds, we identified additional cases that raised concerns. - 364093865 — solved only 28 minutes after the previous problem, again at a speed typical of very strong grandmaster-level participants.
- 364083892 — a relatively large implementation written in 24 minutes, while the participant usually writes similar logic without using structures.
3. Codeforces Round 1081
363899193 — a major change compared to the previous submission, completed in only 9 minutes. Both the code size and the underlying idea changed significantly.
Overall, these three consecutive contests stand out relative to the participant’s recent performance. Each of them individually raised some concerns, together they form a noticeable pattern.
4. Related accounts
We also found accounts that appear to be connected:
- Kastamonu — an account that seems to have been used by multiple users, including him.
- KKKonya — another account used by the same person. Comparing submissions from the same period suggests a strong similarity in coding style: 199987492, 192884638
5. Previous confirmed cheating cases
We also found previous instances of cheating:
Additionally, there are several more cheating cases from the past we just found:
- 164748868, 164767139 and 164767964 (he obviously just gave his friend solution)
- 168172651 and 168158838 (note that the second one was submitted later)
And some more cases with the same person.
Many community members would consider these systematic violations sufficient reason for a ban on their own.
Final note
The only question we tried to answer was whether the original ban decision could have been incorrect. After reviewing the case, we believe that the collected indicators support the original decision.
