Thanks to Kaggle team for the first leaderboard cleaning. It was done unusually quickly, in my opinion the reason is that Kaggle team needed to concentrate on the not very smooth ending of Ubiquant competition.
\nMany teams were removed from the leaderboard but several other \"sudden geniuses\" were left intact. I concentrated only on the silver zone and there are apparently 5 teams showing all the signs of typical \"goose invasion teams\" (as I called it in a discussion already half a year ago).
\nThe typical signs are, as it was already written elsewhere, nearly no activity (in average less than 4 days per team member), hardly any history and of course sudden improvement. Yes, they can have some limited history from another competition, because if they succeed once, they are encouraged to repeat the scenario again.
I speak about positions 17.(doloresf), 31.(simba), 35. (some signes), 54.(GreatGBL), and my favorites 56.(Sheep101). The members of team Sheep101 are real record breakers:
\nOne member - 2 silver medals for only 2 days of activity on Kaggle at all, second member - 2 silver medals for 4 days activity on Kaggle at all and the 2 others - one silver medal for 1 day activity on Kaggle at all!
\nIf there is no cheating behind this tremendous success, they should be nominated for this year's Turing Award!
You may say that 5 teams don't matter so much. But if any seeds of unfairness are left it will hugely encourage all others to keep trying and the problem will grow monster in a short time again.
\nSo @wcukierski and @addisonhoward could you have a deeper look into these issues and create only a positive precedent for future competitions?
\nThanks in advance.
And, please, don't refer me to a compliance form, because evidently only Kaggle can always have all the evidence \"beyond any reasonable doubt\". And I don't mean my comment as any witch-hunt but as a polite call for a system improvement.
", "messages": [ { "id": "1762385", "postDate": "04/20/2022 16:41:43", "content": "Thanks to Kaggle team for the first leaderboard cleaning. It was done unusually quickly, in my opinion the reason is that Kaggle team needed to concentrate on the not very smooth ending of Ubiquant competition.
\nMany teams were removed from the leaderboard but several other \"sudden geniuses\" were left intact. I concentrated only on the silver zone and there are apparently 5 teams showing all the signs of typical \"goose invasion teams\" (as I called it in a discussion already half a year ago).
\nThe typical signs are, as it was already written elsewhere, nearly no activity (in average less than 4 days per team member), hardly any history and of course sudden improvement. Yes, they can have some limited history from another competition, because if they succeed once, they are encouraged to repeat the scenario again.
I speak about positions 17.(doloresf), 31.(simba), 35. (some signes), 54.(GreatGBL), and my favorites 56.(Sheep101). The members of team Sheep101 are real record breakers:
\nOne member - 2 silver medals for only 2 days of activity on Kaggle at all, second member - 2 silver medals for 4 days activity on Kaggle at all and the 2 others - one silver medal for 1 day activity on Kaggle at all!
\nIf there is no cheating behind this tremendous success, they should be nominated for this year's Turing Award!
You may say that 5 teams don't matter so much. But if any seeds of unfairness are left it will hugely encourage all others to keep trying and the problem will grow monster in a short time again.
\nSo @wcukierski and @addisonhoward could you have a deeper look into these issues and create only a positive precedent for future competitions?
\nThanks in advance.
And, please, don't refer me to a compliance form, because evidently only Kaggle can always have all the evidence \"beyond any reasonable doubt\". And I don't mean my comment as any witch-hunt but as a polite call for a system improvement.
", "rawMarkdown": "Thanks to Kaggle team for the first leaderboard cleaning. It was done unusually quickly, in my opinion the reason is that Kaggle team needed to concentrate on the not very smooth ending of Ubiquant competition.\n\nMany teams were removed from the leaderboard but several other \"sudden geniuses\" were left intact. I concentrated only on the silver zone and there are apparently 5 teams showing all the signs of typical \"goose invasion teams\" (as I called it in a discussion already half a year ago).\nThe typical signs are, as it was already written elsewhere, nearly no activity (in average less than 4 days per team member), hardly any history and of course sudden improvement. Yes, they can have some limited history from another competition, because if they succeed once, they are encouraged to repeat the scenario again. \n \nI speak about positions 17.(doloresf), 31.(simba), 35. (some signes), 54.(GreatGBL), and my favorites 56.(Sheep101). The members of team Sheep101 are real record breakers:\nOne member - 2 silver medals for only 2 days of activity on Kaggle at all, second member - 2 silver medals for 4 days activity on Kaggle at all and the 2 others - one silver medal for 1 day activity on Kaggle at all!\nIf there is no cheating behind this tremendous success, they should be nominated for this year's Turing Award!\n\nYou may say that 5 teams don't matter so much. But if any seeds of unfairness are left it will hugely encourage all others to keep trying and the problem will grow monster in a short time again.\nSo @wcukierski and @addisonhoward could you have a deeper look into these issues and **create only a positive precedent for future competitions**?\nThanks in advance.\n\nAnd, please, don't refer me to a compliance form, because evidently only Kaggle can always have all the evidence \"beyond any reasonable doubt\". And I don't mean my comment as any witch-hunt but as a polite call for a system improvement.", "votes": null }, { "id": "1762473", "postDate": "04/20/2022 17:53:32", "content": "You can add #23 and #32 to your list.
\nUnfortunately I think the Kaggle team has to review user profiles one by one to detect cheaters, the current algorithm is not really working.
Thank you for your note, Theo.
\nYou are right, there are also team members there who have 1 or 2 activity days only.
\nI would like somebody to show me how to legally get a silver medal after 1 day on Kaggle! I was working for it for several weeks.
ID verified accounts (I requested this a while ago) would have solved such issues once and for all. If people aren't willing to verify their accounts, then limit what they can do on kaggle such as no participation in ranking point awarding (tiers worth) competitions, unable to cast votes/down votes, etc..
", "rawMarkdown": "ID verified accounts ([I requested this a while ago](https://www.kaggle.com/discussions/product-feedback/300028)) would have solved such issues once and for all. If people aren't willing to verify their accounts, then limit what they can do on kaggle such as **no participation in ranking point awarding (tiers worth) competitions**, **unable to cast votes/down votes**, etc..", "votes": null }, { "id": "1763574", "postDate": "04/21/2022 16:20:57", "content": "Cannot agree more.
", "rawMarkdown": "Cannot agree more.", "votes": null } ], "comments": [ { "id": 1762473, "author_name": "theoviel", "author_url": "", "post_date": "04/20/2022 17:53:32", "content": "You can add #23 and #32 to your list.
\nUnfortunately I think the Kaggle team has to review user profiles one by one to detect cheaters, the current algorithm is not really working.
Thank you for your note, Theo.
\nYou are right, there are also team members there who have 1 or 2 activity days only.
\nI would like somebody to show me how to legally get a silver medal after 1 day on Kaggle! I was working for it for several weeks.
ID verified accounts (I requested this a while ago) would have solved such issues once and for all. If people aren't willing to verify their accounts, then limit what they can do on kaggle such as no participation in ranking point awarding (tiers worth) competitions, unable to cast votes/down votes, etc..
", "votes": null, "replies": [] }, { "id": 1763574, "author_name": "haqishen", "author_url": "", "post_date": "04/21/2022 16:20:57", "content": "Cannot agree more.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1762385": "Thanks to Kaggle team for the first leaderboard cleaning. It was done unusually quickly, in my opinion the reason is that Kaggle team needed to concentrate on the not very smooth ending of Ubiquant competition.\n\nMany teams were removed from the leaderboard but several other \"sudden geniuses\" were left intact. I concentrated only on the silver zone and there are apparently 5 teams showing all the signs of typical \"goose invasion teams\" (as I called it in a discussion already half a year ago).\nThe typical signs are, as it was already written elsewhere, nearly no activity (in average less than 4 days per team member), hardly any history and of course sudden improvement. Yes, they can have some limited history from another competition, because if they succeed once, they are encouraged to repeat the scenario again. \n \nI speak about positions 17.(doloresf), 31.(simba), 35. (some signes), 54.(GreatGBL), and my favorites 56.(Sheep101). The members of team Sheep101 are real record breakers:\nOne member - 2 silver medals for only 2 days of activity on Kaggle at all, second member - 2 silver medals for 4 days activity on Kaggle at all and the 2 others - one silver medal for 1 day activity on Kaggle at all!\nIf there is no cheating behind this tremendous success, they should be nominated for this year's Turing Award!\n\nYou may say that 5 teams don't matter so much. But if any seeds of unfairness are left it will hugely encourage all others to keep trying and the problem will grow monster in a short time again.\nSo @wcukierski and @addisonhoward could you have a deeper look into these issues and **create only a positive precedent for future competitions**?\nThanks in advance.\n\nAnd, please, don't refer me to a compliance form, because evidently only Kaggle can always have all the evidence \"beyond any reasonable doubt\". And I don't mean my comment as any witch-hunt but as a polite call for a system improvement.", "1762473": "You can add #23 and #32 to your list. \nUnfortunately I think the Kaggle team has to review user profiles one by one to detect cheaters, the current algorithm is not really working.", "1762488": "Thank you for your note, Theo. \nYou are right, there are also team members there who have 1 or 2 activity days only. \nI would like somebody to show me how to legally get a silver medal after 1 day on Kaggle! I was working for it for several weeks.", "1762992": "ID verified accounts ([I requested this a while ago](https://www.kaggle.com/discussions/product-feedback/300028)) would have solved such issues once and for all. If people aren't willing to verify their accounts, then limit what they can do on kaggle such as **no participation in ranking point awarding (tiers worth) competitions**, **unable to cast votes/down votes**, etc..", "1763574": "Cannot agree more." }, "source": "meta" }