\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F716c5b04f39695bfd1a3fe324734aa73%2F1.png?generation=1750829178293377&alt=media\")
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F3aa25c4d76ee3da6dabd532b42b2eb1b%2F2.png?generation=1750829193976281&alt=media\")
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F316d41dba444839e1d236ca3020b687a%2F3.png?generation=1750829205058827&alt=media\")
The Kaggle team just finished the final data update and issued medals, but I really hope that the Kaggle team will take care of the cheating in the competition. It is so speechless and infuriating.
\nJust a friendly (but serious) reminder to the organizers: please donāt overlook possible cheating in this competition. Time series challenges are based on real-world data, and rankings often get locked in the final round ā which means a single unchecked submission can throw the whole leaderboard off.
\nItās a bit suspicious when brand-new accounts with no visible activity suddenly climb the top ranks. Maybe theyāre just geniusesā¦ or maybe not. Either way, worth a closer look.
\nGo ahead, hit me with those downvotes š
\nThis is an example of a suspicious area.
\n
\n
\n
Checking the activity records of some accounts, we can see that:
\nSome accounts have no active records for a long time, but suddenly come online and submit predictions during the competition;
\nThese accounts have consistent active cycles and contribution graphs.
\nThese signs are highly consistent with the strategy of \"controlling multiple accounts to participate in the competition together and trying to submit multiple results to test the leaderboard.\"
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F2bb81356baa2a3cf93eac37120982ceb%2F1.png?generation=1752512736617604&alt=media\")
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F4829ba4a31af1f09512f110023187f43%2F2.png?generation=1752512748777631&alt=media\")
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2Fe7f1752bce9948b51ef9c74170287fe7%2F3.png?generation=1752512758211731&alt=media\")
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17484946%2F75d736d77efd6caef80c50b4d6c09039%2F4.png?generation=1752512769047195&alt=media\")
It might be worth investigating whether some of the top submissions are based on forked versions of the same open-source notebook ā and whether they made identical or minimal changes that coincidentally boosted their scores.
\nAnother strong recommendation: track whether the email address, account activity, or ownership details change after medals are awarded. If someone suddenly disappears or hands off their account, thatās probably not a coincidence.
\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward
", "messages": [ { "id": 3248483, "postDate": "2025-07-14T17:11:17.787Z", "content": "The Kaggle team just finished the final data update and issued medals, but I really hope that the Kaggle team will take care of the cheating in the competition. It is so speechless and infuriating.
\nJust a friendly (but serious) reminder to the organizers: please donāt overlook possible cheating in this competition. Time series challenges are based on real-world data, and rankings often get locked in the final round ā which means a single unchecked submission can throw the whole leaderboard off.
\nItās a bit suspicious when brand-new accounts with no visible activity suddenly climb the top ranks. Maybe theyāre just geniusesā¦ or maybe not. Either way, worth a closer look.
\nGo ahead, hit me with those downvotes š
\nThis is an example of a suspicious area.
\n
\n
\n
Checking the activity records of some accounts, we can see that:
\nSome accounts have no active records for a long time, but suddenly come online and submit predictions during the competition;
\nThese accounts have consistent active cycles and contribution graphs.
\nThese signs are highly consistent with the strategy of \"controlling multiple accounts to participate in the competition together and trying to submit multiple results to test the leaderboard.\"
\n
\n
\n
\n
It might be worth investigating whether some of the top submissions are based on forked versions of the same open-source notebook ā and whether they made identical or minimal changes that coincidentally boosted their scores.
\nAnother strong recommendation: track whether the email address, account activity, or ownership details change after medals are awarded. If someone suddenly disappears or hands off their account, thatās probably not a coincidence.
\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward
", "rawMarkdown": "The Kaggle team just finished the final data update and issued medals, but I really hope that the Kaggle team will take care of the cheating in the competition. It is so speechless and infuriating.\n\nJust a friendly (but serious) reminder to the organizers: please donāt overlook possible cheating in this competition. Time series challenges are based on real-world data, and rankings often get locked in the final round ā which means a single unchecked submission can throw the whole leaderboard off.\n\nItās a bit suspicious when brand-new accounts with no visible activity suddenly climb the top ranks. Maybe theyāre just geniusesā¦ or maybe not. Either way, worth a closer look.\n\nGo ahead, hit me with those downvotes š\n\nThis is an example of a suspicious area.\n\n\n\n\nChecking the activity records of some accounts, we can see that:\n\nSome accounts have no active records for a long time, but suddenly come online and submit predictions during the competition;\n\nThese accounts have consistent active cycles and contribution graphs.\n\nThese signs are highly consistent with the strategy of \"controlling multiple accounts to participate in the competition together and trying to submit multiple results to test the leaderboard.\"\n\n\n\n \n\nIt might be worth investigating whether some of the top submissions are based on forked versions of the same open-source notebook ā and whether they made identical or minimal changes that coincidentally boosted their scores.\n\nAnother strong recommendation: track whether the email address, account activity, or ownership details change after medals are awarded. If someone suddenly disappears or hands off their account, thatās probably not a coincidence.\n\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward ", "votes": 8 }, { "id": 3248488, "postDate": "2025-07-14T17:18:09.963Z", "content": "It seems weāve had an unusually high number of āquant prodigiesā emerge overnight ā all with perfect solutions and minimal public activity. Impressiveā¦ perhaps too impressive.
\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward
\n\nIt seems weāve had an unusually high number of āquant prodigiesā emerge overnight
\n
Believe it or not you can be good at ML and not be on Kaggle for every competition..
", "rawMarkdown": ">It seems weāve had an unusually high number of āquant prodigiesā emerge overnight \n\nBelieve it or not you can be good at ML and not be on Kaggle for every competition.. ", "replies": [ { "id": 3248502, "postDate": "2025-07-14T17:51:40.240Z", "content": "@julianmukaj Think about what you say.
", "rawMarkdown": "@julianmukaj Think about what you say.", "votes": -3 }, { "id": 3248511, "postDate": "2025-07-14T18:06:33.560Z", "content": "@julianmukaj Brother, don't clean them up, it's not clean. You are a quantitative person, and these people are selling medals. This is a gray industry, not what you said about participating in something you are good at. Kaggle is a very low-cost platform for learning in various directions, such as ARC, solving mathematical Olympiads with large models, these are cutting-edge competitions, do you want Kaggle to become a place to sell medals? Finally, you can go and see that there are institutions playing for each game, and I am not targeting them. This is the link of the game that just ended, you can take a look
\nhttps://www.kaggle.com/competitions/waveform-inversion/discussion/587344
@julianmukaj First the cheaters win, then someone volunteers to be their PR manager. You canāt make this up. š
", "rawMarkdown": "@julianmukaj First the cheaters win, then someone volunteers to be their PR manager. You canāt make this up. š" } ] } ] }, { "id": 3248512, "postDate": "2025-07-14T18:09:22.420Z", "content": "Canāt believe after all this, someone actually showed up to defend the medal farming agencies š
", "rawMarkdown": "Canāt believe after all this, someone actually showed up to defend the medal farming agencies š" }, { "id": 3248506, "postDate": "2025-07-14T17:54:50.740Z", "content": "\n\ntbh, I've never seen anyone in industry take these medals seriously. winning prize money is something, 1st place is usually best. eotd, your resume still trumps it all.
\n
As kaggleqrdl said, no one regards medals as the only thing to prove themselves, but isn't Kaggle a place to learn and understand various subdivisions? I don't want Kaggle to become a place to sell medals.
", "rawMarkdown": "> tbh, I've never seen anyone in industry take these medals seriously. winning prize money is something, 1st place is usually best. eotd, your resume still trumps it all.\n\nAs kaggleqrdl said, no one regards medals as the only thing to prove themselves, but isn't Kaggle a place to learn and understand various subdivisions? I don't want Kaggle to become a place to sell medals." }, { "id": 3248497, "postDate": "2025-07-14T17:40:43.790Z", "rawMarkdown": "", "isDeleted": true, "replies": [ { "id": 3248501, "postDate": "2025-07-14T17:49:08.753Z", "content": "@kaggleqrdl Yes, but it's not just this competition. The earthquake wave competition a while ago was also like this. No one regards medals as the only thing. They are all for learning. But aren't you surprised by the number of cheaters? After all, Kaggle is a place to learn, and cheaters sell medals.
", "rawMarkdown": "@kaggleqrdl Yes, but it's not just this competition. The earthquake wave competition a while ago was also like this. No one regards medals as the only thing. They are all for learning. But aren't you surprised by the number of cheaters? After all, Kaggle is a place to learn, and cheaters sell medals." } ] } ], "comments": [ { "id": 3248488, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T17:18:09.963000", "content": "It seems weāve had an unusually high number of āquant prodigiesā emerge overnight ā all with perfect solutions and minimal public activity. Impressiveā¦ perhaps too impressive.
\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward
\n\nIt seems weāve had an unusually high number of āquant prodigiesā emerge overnight
\n
Believe it or not you can be good at ML and not be on Kaggle for every competition..
", "votes": 0, "replies": [ { "id": 3248502, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T17:51:40.240000", "content": "@julianmukaj Think about what you say.
", "votes": -3, "replies": [] }, { "id": 3248511, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T18:06:33.560000", "content": "@julianmukaj Brother, don't clean them up, it's not clean. You are a quantitative person, and these people are selling medals. This is a gray industry, not what you said about participating in something you are good at. Kaggle is a very low-cost platform for learning in various directions, such as ARC, solving mathematical Olympiads with large models, these are cutting-edge competitions, do you want Kaggle to become a place to sell medals? Finally, you can go and see that there are institutions playing for each game, and I am not targeting them. This is the link of the game that just ended, you can take a look
\nhttps://www.kaggle.com/competitions/waveform-inversion/discussion/587344
@julianmukaj First the cheaters win, then someone volunteers to be their PR manager. You canāt make this up. š
", "votes": 0, "replies": [] } ] } ] }, { "id": 3248512, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T18:09:22.420000", "content": "Canāt believe after all this, someone actually showed up to defend the medal farming agencies š
", "votes": 0, "replies": [] }, { "id": 3248506, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T17:54:50.740000", "content": "\n\ntbh, I've never seen anyone in industry take these medals seriously. winning prize money is something, 1st place is usually best. eotd, your resume still trumps it all.
\n
As kaggleqrdl said, no one regards medals as the only thing to prove themselves, but isn't Kaggle a place to learn and understand various subdivisions? I don't want Kaggle to become a place to sell medals.
", "votes": 0, "replies": [] }, { "id": 3248497, "author_name": "", "author_url": "", "post_date": "2025-07-14T17:40:43.790000", "content": "", "votes": 0, "replies": [ { "id": 3248501, "author_name": "songling", "author_url": "", "post_date": "2025-07-14T17:49:08.753000", "content": "@kaggleqrdl Yes, but it's not just this competition. The earthquake wave competition a while ago was also like this. No one regards medals as the only thing. They are all for learning. But aren't you surprised by the number of cheaters? After all, Kaggle is a place to learn, and cheaters sell medals.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "3248483": "The Kaggle team just finished the final data update and issued medals, but I really hope that the Kaggle team will take care of the cheating in the competition. It is so speechless and infuriating.\n\nJust a friendly (but serious) reminder to the organizers: please donāt overlook possible cheating in this competition. Time series challenges are based on real-world data, and rankings often get locked in the final round ā which means a single unchecked submission can throw the whole leaderboard off.\n\nItās a bit suspicious when brand-new accounts with no visible activity suddenly climb the top ranks. Maybe theyāre just geniusesā¦ or maybe not. Either way, worth a closer look.\n\nGo ahead, hit me with those downvotes š\n\nThis is an example of a suspicious area.\n\n\n\n\nChecking the activity records of some accounts, we can see that:\n\nSome accounts have no active records for a long time, but suddenly come online and submit predictions during the competition;\n\nThese accounts have consistent active cycles and contribution graphs.\n\nThese signs are highly consistent with the strategy of \"controlling multiple accounts to participate in the competition together and trying to submit multiple results to test the leaderboard.\"\n\n\n\n \n\nIt might be worth investigating whether some of the top submissions are based on forked versions of the same open-source notebook ā and whether they made identical or minimal changes that coincidentally boosted their scores.\n\nAnother strong recommendation: track whether the email address, account activity, or ownership details change after medals are awarded. If someone suddenly disappears or hands off their account, thatās probably not a coincidence.\n\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward ", "3248488": "It seems weāve had an unusually high number of āquant prodigiesā emerge overnight ā all with perfect solutions and minimal public activity. Impressiveā¦ perhaps too impressive.\n@ryanholbrook @kaitoneil @mylesoneill @mdesaijs @maggiemd @wcukierski @addisonhoward ", "3248512": "Canāt believe after all this, someone actually showed up to defend the medal farming agencies š", "3248506": "> tbh, I've never seen anyone in industry take these medals seriously. winning prize money is something, 1st place is usually best. eotd, your resume still trumps it all.\n\nAs kaggleqrdl said, no one regards medals as the only thing to prove themselves, but isn't Kaggle a place to learn and understand various subdivisions? I don't want Kaggle to become a place to sell medals.", "3248497": "" } }