![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10356799%2Fd163ed2b0c7ede151d682709f77b2007%2F2026-01-15%2017.24.53.jpg?generation=1768494365054215&alt=media\")
Dear Organizers and Committee Members, @qiupenggeng
\nFYI Kaggle Team @addisonhoward
\nFirst of all, we would like to express our immense gratitude for organizing this competition and for the opportunity to work with such rare data. We also sincerely congratulate the winners.
\nHowever, we wish to publicly express our confusion regarding the fact that the majority of participants seemingly misunderstood the assigned task. When a massive number of teams interpret the conditions differently than the organizers intended, it may indicate that the problem statement was not sufficiently clear.
\nAs an example: all major AI models (GPT, Claude, Gemini, DeepSeek), after reading the competition rules, evaluation criteria, and reviewing the files, unequivocally concluded that GROUPS, not just PAIRS, were expected in the submission.
\nOur team, likely along with other participants, spent significant time on our solution and was genuinely interested in achieving a strong result. Those who were seriously tackling this competition would have certainly reacted to clarifications from the organizers during the contest, had any been provided. Moreover, questions regarding the submission format were indeed raised on the forum several times during the competition.
\nFor instance, even in its current form, our solution contains valuable data. If one analyzes the proposed Groups, with minimal matching, it is possible to identify at least 8 correct pairs (from the reference list) that align with the provided ground truth. Our output format was oriented towards groups, as the template appeared to require.
\nHere is some group examples from our solution with more obvious parts:\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10356799%2Fb0ed247c7605ef599a3147804e3dd8dc%2F2026-01-15%2017.25.04.jpg?generation=1768494731767743&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10356799%2F76916d8fda846cf27b017aa47f05169b%2F2026-01-15%2017.25.15.jpg?generation=1768494763831085&alt=media\")
We are not insisting on a revision of your decisions or the list of winners. Nevertheless, we are discouraged by the results and the effective disqualification of the majority of teams. Perhaps finding 2 joined fragments (the result of the 2nd place winner) is a major achievement from an archaeological perspective, but it seems to us that the outcome of a competition hosted by such a globally respected university in the field of archaeology could have been more meaningful and valuable.
\nPerhaps 3rd place still has a deserving winner waiting to be recognized.
\n\"It’s not Gods who burn pots\" team // @samvelkoch, @ambisinistra, @sofiamayzel
", "messages": [ { "id": 3391774, "postDate": "2026-01-15T16:43:26.237Z", "content": "Dear Organizers and Committee Members, @qiupenggeng
\nFYI Kaggle Team @addisonhoward
\nFirst of all, we would like to express our immense gratitude for organizing this competition and for the opportunity to work with such rare data. We also sincerely congratulate the winners.
\nHowever, we wish to publicly express our confusion regarding the fact that the majority of participants seemingly misunderstood the assigned task. When a massive number of teams interpret the conditions differently than the organizers intended, it may indicate that the problem statement was not sufficiently clear.
\nAs an example: all major AI models (GPT, Claude, Gemini, DeepSeek), after reading the competition rules, evaluation criteria, and reviewing the files, unequivocally concluded that GROUPS, not just PAIRS, were expected in the submission.
\nOur team, likely along with other participants, spent significant time on our solution and was genuinely interested in achieving a strong result. Those who were seriously tackling this competition would have certainly reacted to clarifications from the organizers during the contest, had any been provided. Moreover, questions regarding the submission format were indeed raised on the forum several times during the competition.
\nFor instance, even in its current form, our solution contains valuable data. If one analyzes the proposed Groups, with minimal matching, it is possible to identify at least 8 correct pairs (from the reference list) that align with the provided ground truth. Our output format was oriented towards groups, as the template appeared to require.
\nHere is some group examples from our solution with more obvious parts:\n
We are not insisting on a revision of your decisions or the list of winners. Nevertheless, we are discouraged by the results and the effective disqualification of the majority of teams. Perhaps finding 2 joined fragments (the result of the 2nd place winner) is a major achievement from an archaeological perspective, but it seems to us that the outcome of a competition hosted by such a globally respected university in the field of archaeology could have been more meaningful and valuable.
\nPerhaps 3rd place still has a deserving winner waiting to be recognized.
\n\"It’s not Gods who burn pots\" team // @samvelkoch, @ambisinistra, @sofiamayzel
", "rawMarkdown": "Dear Organizers and Committee Members, @qiupenggeng\n\nFYI Kaggle Team @addisonhoward \n\nFirst of all, we would like to express our immense gratitude for organizing this competition and for the opportunity to work with such rare data. We also sincerely congratulate the winners.\n\nHowever, we wish to publicly express our confusion regarding the fact that the majority of participants seemingly misunderstood the assigned task. When a massive number of teams interpret the conditions differently than the organizers intended, it may indicate that the problem statement was not sufficiently clear.\n\nAs an example: all major AI models (GPT, Claude, Gemini, DeepSeek), after reading the competition rules, evaluation criteria, and reviewing the files, unequivocally concluded that **GROUPS**, not just **PAIRS**, were expected in the submission.\n\nOur team, likely along with other participants, spent significant time on our solution and was genuinely interested in achieving a strong result. Those who were seriously tackling this competition would have certainly reacted to clarifications from the organizers during the contest, had any been provided. Moreover, questions regarding the submission format were indeed raised on the forum several times during the competition.\n\nFor instance, even in its current form, our solution contains valuable data. If one analyzes the proposed Groups, with minimal matching, it is possible to identify at least 8 correct pairs (from the reference list) that align with the provided ground truth. Our output format was oriented towards groups, as the template appeared to require.\n\nHere is some group examples from our solution with more obvious parts:\n\n\n---\n\n\n\n---\n\n\n\nWe are not insisting on a revision of your decisions or the list of winners. Nevertheless, we are discouraged by the results and the effective disqualification of the majority of teams. Perhaps finding 2 joined fragments (the result of the 2nd place winner) is a major achievement from an archaeological perspective, but it seems to us that the outcome of a competition hosted by such a globally respected university in the field of archaeology could have been more meaningful and valuable.\n\nPerhaps 3rd place still has a deserving winner waiting to be recognized. \n\n\"It’s not Gods who burn pots\" team // @samvelkoch, @ambisinistra, @sofiamayzel\n\n", "votes": 6 } ], "comments": [], "raw_markdown_by_id": { "3391774": "Dear Organizers and Committee Members, @qiupenggeng\n\nFYI Kaggle Team @addisonhoward \n\nFirst of all, we would like to express our immense gratitude for organizing this competition and for the opportunity to work with such rare data. We also sincerely congratulate the winners.\n\nHowever, we wish to publicly express our confusion regarding the fact that the majority of participants seemingly misunderstood the assigned task. When a massive number of teams interpret the conditions differently than the organizers intended, it may indicate that the problem statement was not sufficiently clear.\n\nAs an example: all major AI models (GPT, Claude, Gemini, DeepSeek), after reading the competition rules, evaluation criteria, and reviewing the files, unequivocally concluded that **GROUPS**, not just **PAIRS**, were expected in the submission.\n\nOur team, likely along with other participants, spent significant time on our solution and was genuinely interested in achieving a strong result. Those who were seriously tackling this competition would have certainly reacted to clarifications from the organizers during the contest, had any been provided. Moreover, questions regarding the submission format were indeed raised on the forum several times during the competition.\n\nFor instance, even in its current form, our solution contains valuable data. If one analyzes the proposed Groups, with minimal matching, it is possible to identify at least 8 correct pairs (from the reference list) that align with the provided ground truth. Our output format was oriented towards groups, as the template appeared to require.\n\nHere is some group examples from our solution with more obvious parts:\n\n\n---\n\n\n\n---\n\n\n\nWe are not insisting on a revision of your decisions or the list of winners. Nevertheless, we are discouraged by the results and the effective disqualification of the majority of teams. Perhaps finding 2 joined fragments (the result of the 2nd place winner) is a major achievement from an archaeological perspective, but it seems to us that the outcome of a competition hosted by such a globally respected university in the field of archaeology could have been more meaningful and valuable.\n\nPerhaps 3rd place still has a deserving winner waiting to be recognized. \n\n\"It’s not Gods who burn pots\" team // @samvelkoch, @ambisinistra, @sofiamayzel\n\n" } }