![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F443651%2Fa918c45416e6548957e4d84d8bddfb6b%2FJaney_Email.png?generation=1674811791103145&alt=media\")
As we all know, gold medal selling has been occurring in many past competitions here on Kaggle. After each competition, Kaggle banned many teams, but often the main culprits came away with it ( 1, 2 ).
\nIn this thread, @kyoshioka47 made a very intriguing observation on how two high-scoring code bases by @shujun717 and @poteman share a lot of similarities (the exact same code), which cannot be found anywhere else on the internet. I have teamed up with @shujun717 in the past, and I am willing to put one of my organs on the line for his integrity. However, in this PANDA competition, Shujun teamed up with @ruozha001. Here is a picture below of the mail Shujun received from this user, which I received from him after just speaking with him on LinkedIn. Doesn't it look extremely familiar?
\n
I have fallen for this trap in the past as well, where I teamed up with this exact @ruozha001 in the Melanoma competition. We were scoring very high on the public LB, and the competition was FILLED with teams of 4-5 novices. All these teams were then later on banned because they bought their medals. One (or two?) master accounts were banned afterwards, but clearly the issue kept on persisting.
\nNow provided with all this information, we can finally connect some dots. @ruozha001 accessed @shujun717 code, passed this through to @poteman. So in other words, @ruozha001 and @poteman are very likely connected, and @ruozha001 is in turn connected to medal selling.
\nMoreover, the silver zone in the public LB is again filled with extremely suspicious teams: 4-5 novices who created their account very recently.
\nIt is hurtful to keep seeing these issues in many Kaggle competitons. However, the good news is that all these issues seem to be connected to each other, and that it is a rather small group of people responsible for this (and a very large group of people dumb enough to buy these medals). It is time to have (i) personal identification, (ii) release a competition so that users are incentivized to help Kaggle put an anti-cheat mechanisms in place; as it can be hard to gather labeled data for such a competition, this could perhaps be done through an analysis competition on anonymized data, (iii) dig deeper in the code submissions done in the PANDA competition, as it should be easier to identify similarities between codes of different teams there than in a CSV competition. Perhaps Kaggle can release this as a dataset?
", "messages": [ { "id": 2117471, "postDate": "2023-01-27T09:54:35.520Z", "content": "As we all know, gold medal selling has been occurring in many past competitions here on Kaggle. After each competition, Kaggle banned many teams, but often the main culprits came away with it ( 1, 2 ).
\nIn this thread, @kyoshioka47 made a very intriguing observation on how two high-scoring code bases by @shujun717 and @poteman share a lot of similarities (the exact same code), which cannot be found anywhere else on the internet. I have teamed up with @shujun717 in the past, and I am willing to put one of my organs on the line for his integrity. However, in this PANDA competition, Shujun teamed up with @ruozha001. Here is a picture below of the mail Shujun received from this user, which I received from him after just speaking with him on LinkedIn. Doesn't it look extremely familiar?
\n
I have fallen for this trap in the past as well, where I teamed up with this exact @ruozha001 in the Melanoma competition. We were scoring very high on the public LB, and the competition was FILLED with teams of 4-5 novices. All these teams were then later on banned because they bought their medals. One (or two?) master accounts were banned afterwards, but clearly the issue kept on persisting.
\nNow provided with all this information, we can finally connect some dots. @ruozha001 accessed @shujun717 code, passed this through to @poteman. So in other words, @ruozha001 and @poteman are very likely connected, and @ruozha001 is in turn connected to medal selling.
\nMoreover, the silver zone in the public LB is again filled with extremely suspicious teams: 4-5 novices who created their account very recently.
\nIt is hurtful to keep seeing these issues in many Kaggle competitons. However, the good news is that all these issues seem to be connected to each other, and that it is a rather small group of people responsible for this (and a very large group of people dumb enough to buy these medals). It is time to have (i) personal identification, (ii) release a competition so that users are incentivized to help Kaggle put an anti-cheat mechanisms in place; as it can be hard to gather labeled data for such a competition, this could perhaps be done through an analysis competition on anonymized data, (iii) dig deeper in the code submissions done in the PANDA competition, as it should be easier to identify similarities between codes of different teams there than in a CSV competition. Perhaps Kaggle can release this as a dataset?
", "rawMarkdown": "As we all know, gold medal selling has been occurring in many past competitions here on Kaggle. After each competition, Kaggle banned many teams, but often the main culprits came away with it ( [1](https://www.kaggle.com/competitions/open-problems-multimodal/discussion/366313), [2](https://www.kaggle.com/competitions/siim-isic-melanoma-classification/discussion/175156) ).\n\nIn [this thread](https://www.kaggle.com/competitions/otto-recommender-system/discussion/381321), @kyoshioka47 made a [very intriguing observation](https://www.kaggle.com/competitions/otto-recommender-system/discussion/381321#2116033) on how two high-scoring code bases by @shujun717 and @poteman share a lot of similarities (the exact same code), which cannot be found anywhere else on the internet. I have teamed up with @shujun717 in the past, and I am willing to put one of my organs on the line for his integrity. However, in this [PANDA competition](https://www.kaggle.com/c/prostate-cancer-grade-assessment/leaderboard), Shujun teamed up with @ruozha001. Here is a picture below of the mail Shujun received from this user, which I received from him after just speaking with him on LinkedIn. Doesn't it look extremely familiar?\n\n\n\nI have fallen for this trap in the past as well, where I teamed up with this exact @ruozha001 in the [Melanoma competition](https://www.kaggle.com/competitions/siim-isic-melanoma-classification). We were scoring very high on the public LB, and the competition was FILLED with teams of 4-5 novices. All these teams were then later on banned because they bought their medals. One (or two?) master accounts were banned afterwards, but clearly the issue kept on persisting.\n\nNow provided with all this information, we can finally connect some dots. @ruozha001 accessed @shujun717 code, passed this through to @poteman. So in other words, @ruozha001 and @poteman are very likely connected, and @ruozha001 is in turn connected to medal selling.\n\nMoreover, the silver zone in the public LB is again filled with extremely suspicious teams: 4-5 novices who created their account very recently.\n\nIt is hurtful to keep seeing these issues in many Kaggle competitons. However, the good news is that all these issues seem to be connected to each other, and that it is a rather small group of people responsible for this (and a very large group of people dumb enough to buy these medals). It is time to have (i) personal identification, (ii) release a competition so that users are incentivized to help Kaggle put an anti-cheat mechanisms in place; as it can be hard to gather labeled data for such a competition, this could perhaps be done through an analysis competition on anonymized data, (iii) dig deeper in the code submissions done in the PANDA competition, as it should be easier to identify similarities between codes of different teams there than in a CSV competition. Perhaps Kaggle can release this as a dataset?", "votes": 55 }, { "id": 2122499, "postDate": "2023-01-31T01:33:18.617Z", "content": "@group16 I'm new to competition and this is sad to see. Thanks for the heads up! Kaggle admin should do what USACO do to cheater, just ban for life.
\nReference: USA Computing Olympiad Contest Instruction and Rules
", "rawMarkdown": "@group16 I'm new to competition and this is sad to see. Thanks for the heads up! Kaggle admin should do what USACO do to cheater, just ban for life.\n\nReference: [USA Computing Olympiad Contest Instruction and Rules](http://www.usaco.org/index.php?page=instructions)", "votes": 1, "replies": [ { "id": 2122877, "postDate": "2023-01-31T07:20:00.393Z", "content": "I agree but unfortunately, because Kaggle accounts can be created with just one email and phone number (i.e. no personal identification), the cheater cannot be personally identified so we can't ban the \"person\". This is why Kaggle needs to implement personal identification.
", "rawMarkdown": "I agree but unfortunately, because **Kaggle accounts can be created with just one email and phone number** (i.e. no personal identification), the cheater cannot be personally identified so we can't ban the \"person\". This is why Kaggle needs to implement personal identification.", "votes": 1 } ] }, { "id": 2121655, "postDate": "2023-01-30T13:23:16.300Z", "content": "\n\nAfter each competition, Kaggle banned many teams
\n
Unfortunately, this is just a band-aid as a solution. Most of these cheaters will face no consequences and come back with one new phone number and email to make another Kaggle account. This is why we need personal identification for Kaggle accounts (like you have mentioned too). I give credit to the Kaggle organizers for their hard work to remove these cheaters at the end of competitions but it will be \"smarter\" work if they implement personal identification.
", "rawMarkdown": "> After each competition, Kaggle banned many teams\n\nUnfortunately, this is just a band-aid as a solution. Most of these **cheaters will face no consequences** and come back with one new phone number and email to make another Kaggle account. This is why we need personal identification for Kaggle accounts (like you have mentioned too). I give credit to the Kaggle organizers for their hard work to remove these cheaters at the end of competitions but it will be \"smarter\" work if they implement personal identification.", "votes": 1 }, { "id": 2120964, "postDate": "2023-01-29T23:25:56.313Z", "content": "Thanks so much for sharing. More data-points like that allow us to create a complete picture of this scheme .
", "rawMarkdown": "Thanks so much for sharing. More data-points like that allow us to create a complete picture of this scheme .", "votes": 2 }, { "id": 2118017, "postDate": "2023-01-27T18:03:54.277Z", "content": "Consider upvoting my product feedback here - https://www.kaggle.com/discussions/product-feedback/373889#2118013
\nKaggle should pro-actively warn and provide suggestions to users when teaming, such as getting on a zoom call and doing a two way technical interview, talking about code sharing expections and how the team will operate. Also, using automated KYC services which add a mark to our profiles so folks know who has committed to Kaggle at least about making their identity fully known.
", "rawMarkdown": "Consider upvoting my product feedback here - https://www.kaggle.com/discussions/product-feedback/373889#2118013 \n\nKaggle should pro-actively warn and provide suggestions to users when teaming, such as getting on a zoom call and doing a two way technical interview, talking about code sharing expections and how the team will operate. Also, using automated KYC services which add a mark to our profiles so folks know who has committed to Kaggle at least about making their identity fully known.", "votes": 1 }, { "id": 2118788, "postDate": "2023-01-28T09:23:46.630Z", "rawMarkdown": "", "votes": 3, "isDeleted": true, "replies": [ { "id": 2119528, "postDate": "2023-01-28T21:08:43.737Z", "content": "I fully agree with you. But first of all Kaggle needs to stop ignoring some reports at all.
\nFight against cheating and unfairness of all kinds is my long term agenda and I really haven't seen any Kaggle team's substantial action done without a previous big public \"uprising\".
\nThis signalizes that reporting system in fact doesn't work properly and filling any form is mostly lost time.
\nIn fact I don't see any interest from Kaggle's side to solve any problem until they are massively publicly forced to do so.
Something to consider - unless it's a prize winner, I think beyond a shadow of doubt is a reasonable bar to get past. These 'gold medals' are virtual and really just bragging rights, no need to risk making a mistake over them, even if a few cheaters unfairly win out.
\nI'm fairly sure for payouts, they go over things fairly carefully.
\nHonestly, I wouldn't take the medals too seriously. If you want to evaluate someone on Kaggle, look over their published notebooks and published winning solutions.
", "rawMarkdown": "Something to consider - unless it's a prize winner, I think beyond a shadow of doubt is a reasonable bar to get past. These 'gold medals' are virtual and really just bragging rights, no need to risk making a mistake over them, even if a few cheaters unfairly win out.\n\nI'm fairly sure for payouts, they go over things fairly carefully.\n\nHonestly, I wouldn't take the medals too seriously. If you want to evaluate someone on Kaggle, look over their published notebooks and published winning solutions.", "votes": 1 } ] } ] }, { "id": 2118763, "postDate": "2023-01-28T08:56:09.277Z", "rawMarkdown": "", "votes": 3, "isDeleted": true, "replies": [ { "id": 2118769, "postDate": "2023-01-28T09:00:48.557Z", "content": "Not if we can cut off the source of these medals. If all masters & GMs responsible for these practices can be banned, it will become much more difficult for them.
", "rawMarkdown": "Not if we can cut off the source of these medals. If all masters & GMs responsible for these practices can be banned, it will become much more difficult for them.", "votes": 6, "replies": [ { "id": 2122511, "postDate": "2023-01-31T01:49:05.637Z", "content": "Maybe it has always been here, but now we might be dealing with another scale of the problem
", "rawMarkdown": "Maybe it has always been here, but now we might be dealing with another scale of the problem", "votes": 2 }, { "id": 2122879, "postDate": "2023-01-31T07:22:03.020Z", "content": "Yes, the scale of this problem is growing. Check out the scale of the cheating that occured in the Open Problems comp!: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/366313
", "rawMarkdown": "Yes, the **scale of this problem is growing**. Check out the scale of the cheating that occured in the Open Problems comp!: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/366313", "votes": 1 } ] } ] } ], "comments": [ { "id": 2122499, "author_name": "Minh T. Nguyen", "author_url": "", "post_date": "2023-01-31T01:33:18.617000", "content": "@group16 I'm new to competition and this is sad to see. Thanks for the heads up! Kaggle admin should do what USACO do to cheater, just ban for life.
\nReference: USA Computing Olympiad Contest Instruction and Rules
", "votes": 1, "replies": [ { "id": 2122877, "author_name": "goto_conversion", "author_url": "", "post_date": "2023-01-31T07:20:00.393000", "content": "I agree but unfortunately, because Kaggle accounts can be created with just one email and phone number (i.e. no personal identification), the cheater cannot be personally identified so we can't ban the \"person\". This is why Kaggle needs to implement personal identification.
", "votes": 1, "replies": [] } ] }, { "id": 2121655, "author_name": "goto_conversion", "author_url": "", "post_date": "2023-01-30T13:23:16.300000", "content": "\n\nAfter each competition, Kaggle banned many teams
\n
Unfortunately, this is just a band-aid as a solution. Most of these cheaters will face no consequences and come back with one new phone number and email to make another Kaggle account. This is why we need personal identification for Kaggle accounts (like you have mentioned too). I give credit to the Kaggle organizers for their hard work to remove these cheaters at the end of competitions but it will be \"smarter\" work if they implement personal identification.
", "votes": 1, "replies": [] }, { "id": 2120964, "author_name": "narsil (jobs-in-data.com)", "author_url": "", "post_date": "2023-01-29T23:25:56.313000", "content": "Thanks so much for sharing. More data-points like that allow us to create a complete picture of this scheme .
", "votes": 2, "replies": [] }, { "id": 2118017, "author_name": "@kaggleqrdl", "author_url": "", "post_date": "2023-01-27T18:03:54.277000", "content": "Consider upvoting my product feedback here - https://www.kaggle.com/discussions/product-feedback/373889#2118013
\nKaggle should pro-actively warn and provide suggestions to users when teaming, such as getting on a zoom call and doing a two way technical interview, talking about code sharing expections and how the team will operate. Also, using automated KYC services which add a mark to our profiles so folks know who has committed to Kaggle at least about making their identity fully known.
", "votes": 1, "replies": [] }, { "id": 2118788, "author_name": "", "author_url": "", "post_date": "2023-01-28T09:23:46.630000", "content": "", "votes": 3, "replies": [ { "id": 2119528, "author_name": "Allie K.", "author_url": "", "post_date": "2023-01-28T21:08:43.737000", "content": "I fully agree with you. But first of all Kaggle needs to stop ignoring some reports at all.
\nFight against cheating and unfairness of all kinds is my long term agenda and I really haven't seen any Kaggle team's substantial action done without a previous big public \"uprising\".
\nThis signalizes that reporting system in fact doesn't work properly and filling any form is mostly lost time.
\nIn fact I don't see any interest from Kaggle's side to solve any problem until they are massively publicly forced to do so.
Something to consider - unless it's a prize winner, I think beyond a shadow of doubt is a reasonable bar to get past. These 'gold medals' are virtual and really just bragging rights, no need to risk making a mistake over them, even if a few cheaters unfairly win out.
\nI'm fairly sure for payouts, they go over things fairly carefully.
\nHonestly, I wouldn't take the medals too seriously. If you want to evaluate someone on Kaggle, look over their published notebooks and published winning solutions.
", "votes": 1, "replies": [] } ] } ] }, { "id": 2118763, "author_name": "", "author_url": "", "post_date": "2023-01-28T08:56:09.277000", "content": "", "votes": 3, "replies": [ { "id": 2118769, "author_name": "Gilles Vandewiele", "author_url": "", "post_date": "2023-01-28T09:00:48.557000", "content": "Not if we can cut off the source of these medals. If all masters & GMs responsible for these practices can be banned, it will become much more difficult for them.
", "votes": 6, "replies": [ { "id": 2122511, "author_name": "HinePo", "author_url": "", "post_date": "2023-01-31T01:49:05.637000", "content": "Maybe it has always been here, but now we might be dealing with another scale of the problem
", "votes": 2, "replies": [] }, { "id": 2122879, "author_name": "goto_conversion", "author_url": "", "post_date": "2023-01-31T07:22:03.020000", "content": "Yes, the scale of this problem is growing. Check out the scale of the cheating that occured in the Open Problems comp!: https://www.kaggle.com/competitions/open-problems-multimodal/discussion/366313
", "votes": 1, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2117471": "As we all know, gold medal selling has been occurring in many past competitions here on Kaggle. After each competition, Kaggle banned many teams, but often the main culprits came away with it ( [1](https://www.kaggle.com/competitions/open-problems-multimodal/discussion/366313), [2](https://www.kaggle.com/competitions/siim-isic-melanoma-classification/discussion/175156) ).\n\nIn [this thread](https://www.kaggle.com/competitions/otto-recommender-system/discussion/381321), @kyoshioka47 made a [very intriguing observation](https://www.kaggle.com/competitions/otto-recommender-system/discussion/381321#2116033) on how two high-scoring code bases by @shujun717 and @poteman share a lot of similarities (the exact same code), which cannot be found anywhere else on the internet. I have teamed up with @shujun717 in the past, and I am willing to put one of my organs on the line for his integrity. However, in this [PANDA competition](https://www.kaggle.com/c/prostate-cancer-grade-assessment/leaderboard), Shujun teamed up with @ruozha001. Here is a picture below of the mail Shujun received from this user, which I received from him after just speaking with him on LinkedIn. Doesn't it look extremely familiar?\n\n\n\nI have fallen for this trap in the past as well, where I teamed up with this exact @ruozha001 in the [Melanoma competition](https://www.kaggle.com/competitions/siim-isic-melanoma-classification). We were scoring very high on the public LB, and the competition was FILLED with teams of 4-5 novices. All these teams were then later on banned because they bought their medals. One (or two?) master accounts were banned afterwards, but clearly the issue kept on persisting.\n\nNow provided with all this information, we can finally connect some dots. @ruozha001 accessed @shujun717 code, passed this through to @poteman. So in other words, @ruozha001 and @poteman are very likely connected, and @ruozha001 is in turn connected to medal selling.\n\nMoreover, the silver zone in the public LB is again filled with extremely suspicious teams: 4-5 novices who created their account very recently.\n\nIt is hurtful to keep seeing these issues in many Kaggle competitons. However, the good news is that all these issues seem to be connected to each other, and that it is a rather small group of people responsible for this (and a very large group of people dumb enough to buy these medals). It is time to have (i) personal identification, (ii) release a competition so that users are incentivized to help Kaggle put an anti-cheat mechanisms in place; as it can be hard to gather labeled data for such a competition, this could perhaps be done through an analysis competition on anonymized data, (iii) dig deeper in the code submissions done in the PANDA competition, as it should be easier to identify similarities between codes of different teams there than in a CSV competition. Perhaps Kaggle can release this as a dataset?", "2122499": "@group16 I'm new to competition and this is sad to see. Thanks for the heads up! Kaggle admin should do what USACO do to cheater, just ban for life.\n\nReference: [USA Computing Olympiad Contest Instruction and Rules](http://www.usaco.org/index.php?page=instructions)", "2121655": "> After each competition, Kaggle banned many teams\n\nUnfortunately, this is just a band-aid as a solution. Most of these **cheaters will face no consequences** and come back with one new phone number and email to make another Kaggle account. This is why we need personal identification for Kaggle accounts (like you have mentioned too). I give credit to the Kaggle organizers for their hard work to remove these cheaters at the end of competitions but it will be \"smarter\" work if they implement personal identification.", "2120964": "Thanks so much for sharing. More data-points like that allow us to create a complete picture of this scheme .", "2118017": "Consider upvoting my product feedback here - https://www.kaggle.com/discussions/product-feedback/373889#2118013 \n\nKaggle should pro-actively warn and provide suggestions to users when teaming, such as getting on a zoom call and doing a two way technical interview, talking about code sharing expections and how the team will operate. Also, using automated KYC services which add a mark to our profiles so folks know who has committed to Kaggle at least about making their identity fully known.", "2118788": "", "2118763": "" } }