Excited to read those modeling approach. =)
", "messages": [ { "id": 936956, "postDate": "2020-07-20T16:15:24.143Z", "content": "![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1582365%2F6ce3a030b95043a536569df49038b451%2FScreenshot%202020-07-20%2017.13.35.png?generation=1595261646416620&alt=media\")
\nIt is entirely feasible that it is a complete coincidence ... for at least 8 teams to suddenly score .936 within 2 hours of submissions**, and to join just 9 other teams on .936 after weeks and months of submissions. Isn't it? Maybe? Coincidences coincidences. Two teams have users with similar username, Two other teams share similar username in opposite team.
You DO see new joiners, and in teams, in research challenges. They can be less used to the gamification of live leaderboards, happy to make their first submissions with hours to go. Of course, teams had to be formed 7 days ago.
\n\n**Downloadable public.csv shows 7 of these new teams with .936 submissions were made within a 2 hour period. Coincidences do happen?
", "rawMarkdown": "\nIt is entirely feasible that it is a complete coincidence ... for at least 8 teams to suddenly score .936 within 2 hours of submissions**, and to join just 9 other teams on .936 after weeks and months of submissions. Isn't it? Maybe? Coincidences coincidences. Two teams have users with similar username, Two other teams share similar username in opposite team. \n\nYou DO see new joiners, and in teams, in research challenges. They can be less used to the gamification of live leaderboards, happy to make their first submissions with hours to go. Of course, teams had to be formed 7 days ago.\n\n**Downloadable public.csv shows 7 of these new teams with .936 submissions were made within a 2 hour period. Coincidences do happen?\n\n", "votes": 19, "replies": [ { "id": 936967, "postDate": "2020-07-20T16:20:29.503Z", "content": "I hesitate to take a screenshot to share...oh man π
", "rawMarkdown": "I hesitate to take a screenshot to share...oh man π ", "votes": 2 }, { "id": 936978, "postDate": "2020-07-20T16:25:37.917Z", "content": "Anyway well done all of them.
\n\nI hope in the interests of research we get a large amount of teams sharing their solutions no matter their leaderboard place. The knowledge I'll learn will make up for dropping an additional 10 positions in a day.
", "rawMarkdown": "Anyway well done all of them. \n\nI hope in the interests of research we get a large amount of teams sharing their solutions no matter their leaderboard place. The knowledge I'll learn will make up for dropping an additional 10 positions in a day.", "votes": 3 }, { "id": 936979, "postDate": "2020-07-20T16:27:13.940Z", "content": "hmm, same here. =)
", "rawMarkdown": "hmm, same here. =)", "votes": 1 }, { "id": 937593, "postDate": "2020-07-21T05:10:54.040Z", "content": "Yes @robga at the end of the day everything is knowledge.
", "rawMarkdown": "Yes @robga at the end of the day everything is knowledge. ", "votes": 1 }, { "id": 938006, "postDate": "2020-07-21T09:45:15.510Z", "content": "And here is a screenshot of the 'after'. The odds of a moving as a group to the same private score is very low indeed. Disappointed this happens on the site.\n![\"https://i.imgur.com/OTqBTYa.png\"](\"https://i.imgur.com/OTqBTYa.png\")
EDIT 27/7: Leaderboard finalised and 9 offending teams in this screenshot removed. Thank you to Kaggle for your excellent compliance team, it is a valued service to the community.
", "rawMarkdown": "And here is a screenshot of the 'after'. The odds of a moving as a group to the same private score is very low indeed. Disappointed this happens on the site.\n\n\nEDIT 27/7: Leaderboard finalised and 9 offending teams in this screenshot removed. Thank you to Kaggle for your excellent compliance team, it is a valued service to the community.", "votes": 3 } ] }, { "id": 936814, "postDate": "2020-07-20T14:44:45.420Z", "content": "Excited to read those modeling approach. =)
", "rawMarkdown": "Excited to read those modeling approach. =)", "votes": 22 }, { "id": 936949, "postDate": "2020-07-20T16:11:38.103Z", "content": "Donβt start witch-hunt too early. I easily can assume solving this challenge using only local validation. This is the only reasonable way when you have such small test set. \nIn case this is a fake account - kaggle will eventually ban it. But before this happened - just keep calm.
", "rawMarkdown": "Donβt start witch-hunt too early. I easily can assume solving this challenge using only local validation. This is the only reasonable way when you have such small test set. \nIn case this is a fake account - kaggle will eventually ban it. But before this happened - just keep calm.", "votes": 7 }, { "id": 937057, "postDate": "2020-07-20T17:40:05.370Z", "content": "I usually dont jump to conclusions early, but this looks very obvious.
", "rawMarkdown": "I usually dont jump to conclusions early, but this looks very obvious.", "votes": 8 }, { "id": 936884, "postDate": "2020-07-20T15:13:56.710Z", "content": "it's strange to see people that are submitting less than 10 submissions and getting the Top of LB and the most of them , they were joining only this competition and kaggle recently ! Any explanation ?
", "rawMarkdown": "it's strange to see people that are submitting less than 10 submissions and getting the Top of LB and the most of them , they were joining only this competition and kaggle recently ! Any explanation ?", "votes": 5, "replies": [ { "id": 936986, "postDate": "2020-07-20T16:30:08.967Z", "content": "Nothing actually! Things are very clear to everyone.
", "rawMarkdown": "Nothing actually! Things are very clear to everyone. ", "votes": 6 } ] }, { "id": 937031, "postDate": "2020-07-20T17:16:17.290Z", "content": "Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π
", "rawMarkdown": "Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π", "votes": 3, "replies": [ { "id": 937597, "postDate": "2020-07-21T05:13:52.543Z", "content": "You are right @nischaydnk It cant be a coincidence
", "rawMarkdown": "You are right @nischaydnk It cant be a coincidence ", "votes": 1 } ] }, { "id": 937162, "postDate": "2020-07-20T19:14:54.693Z", "content": "and then there was a legendary single sub (idle) speculation
\n\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/leaderboard
", "rawMarkdown": "and then there was a legendary single sub (idle) speculation\n\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/leaderboard", "votes": 4 }, { "id": 937239, "postDate": "2020-07-20T20:55:09.217Z", "content": "You only have 5 subs per day. Having more subs is a huge advantage, that's why people open new accounts.
\n\n(this should be a reply to @remicogranne but it always ends up separately)
", "rawMarkdown": "You only have 5 subs per day. Having more subs is a huge advantage, that's why people open new accounts.\n\n(this should be a reply to @remicogranne but it always ends up separately)", "votes": 3 }, { "id": 937350, "postDate": "2020-07-21T00:34:23.073Z", "content": "A gold medal position ( in Private LB) with only 6 submission and Only one single submission got bronze medal π²
", "rawMarkdown": "A gold medal position ( in Private LB) with only 6 submission and Only one single submission got bronze medal π²", "votes": 1 }, { "id": 937199, "postDate": "2020-07-20T19:55:51.890Z", "content": "Hello there,
\nSInce I am a total newbie on Kaggle (and acknowledge this matter of fact) I don't see how you can \"cheat\" ?
\nI mean arguably, you can presume that those people, whose scores are almost identical and submission almost simultaneous, are working together and sharing information. However, in such a case it would be very silly to submit using different account (and why do so ?)
\nHowever, I also strongly believe that since the training and testing dataset are extracted from the same sources of images (subsampling of the same original dataset, images processed in the same way) you can train a lot offline and take your chance only at the very end in order not to reveal your game right ?
\nSorry for newbie question, it's just that it does not seems so obvious to me and I don't get the point of acting this was :c
By creating a new ID, the contestant can make additional submissions per day. If in the last few weeks, one has many experiments and not enough submission slots, this would allow the contestant to try out different experiments to the leaderboard.
", "rawMarkdown": "By creating a new ID, the contestant can make additional submissions per day. If in the last few weeks, one has many experiments and not enough submission slots, this would allow the contestant to try out different experiments to the leaderboard. ", "votes": 3 } ] }, { "id": 937156, "postDate": "2020-07-20T19:11:15.797Z", "content": "I was a Kaggle competitor during the Rossmann competition. For the first month or so, the eventual winner stayed on the number one spot of the leaderboard with only a few submissions. He just relied on local validation and didn't submit very often. It's definitely possible. (But, yeah, I know what you mean.)
\nhttps://www.kaggle.com/c/rossmann-store-sales/leaderboard
", "rawMarkdown": "I was a Kaggle competitor during the Rossmann competition. For the first month or so, the eventual winner stayed on the number one spot of the leaderboard with only a few submissions. He just relied on local validation and didn't submit very often. It's definitely possible. (But, yeah, I know what you mean.)\n\nhttps://www.kaggle.com/c/rossmann-store-sales/leaderboard", "votes": 1 }, { "id": 937084, "postDate": "2020-07-20T17:58:09.580Z", "content": "Happend before.
\n\nhttps://www.kaggle.com/c/imaterialist-challenge-furniture-2018/leaderboard
", "rawMarkdown": "Happend before. \n\nhttps://www.kaggle.com/c/imaterialist-challenge-furniture-2018/leaderboard\n\n", "votes": 2, "replies": [ { "id": 937086, "postDate": "2020-07-20T18:01:11.353Z", "content": "So, things are repeated and there's no concern from kaggle team. Is that it?
", "rawMarkdown": "So, things are repeated and there's no concern from kaggle team. Is that it?" }, { "id": 937137, "postDate": "2020-07-20T18:48:39.947Z", "content": "\n\nSo, things are repeated and there's no concern from kaggle team. Is that it?
\n
We take allegations of cheating seriously and actively refine the methods we use to screen for cheating. We do not publicly speculate on ongoing accusations, nor make public examples of users who we do remove for violating the rules. We base our compliance decisions on a standard of evidence aimed at minimizing false positives. The minimum penalty for violating competition rules is removal from the competition, while repeat, egregious, or deliberate violations can warrant account removal.
\nThere is no question that some have cheated and gotten away with it, some have tried and been caught, but \"no concern\" is not an accurate characterization of our stance or history on this issue.
\nFriendly reminder: please use our compliance form to report evidence of cheating.
", "rawMarkdown": "> So, things are repeated and there's no concern from kaggle team. Is that it?\n\nWe take allegations of cheating seriously and actively refine the methods we use to screen for cheating. We do not publicly speculate on ongoing accusations, nor make public examples of users who we do remove for violating the rules. We base our compliance decisions on a standard of evidence aimed at minimizing false positives. The minimum penalty for violating competition rules is removal from the competition, while repeat, egregious, or deliberate violations can warrant account removal.\n\nThere is no question that some have cheated and gotten away with it, some have tried and been caught, but \"no concern\" is not an accurate characterization of our stance or history on this issue.\n\nFriendly reminder: please use our [compliance form](https://www.kaggle.com/compliance) to report evidence of cheating.", "votes": 8 }, { "id": 937148, "postDate": "2020-07-20T19:00:07.433Z", "content": "@wcukierski \nThank you. Understood.
", "rawMarkdown": "@wcukierski \nThank you. Understood. ", "votes": 2 } ] }, { "id": 937071, "postDate": "2020-07-20T17:47:01.770Z", "content": "\n\n\nNischay Dhankhar wrote:\n Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π
\n
Like they said, Lucky! π
No I don't call it karma at all, they just probably have the similar blends which end up in roughly same places after shake up too, it's just that their subs are not good enough for better positions. Tbh, I can't really give them the benefit of a doubt here as it's too obvious.
", "rawMarkdown": "No I don't call it karma at all, they just probably have the similar blends which end up in roughly same places after shake up too, it's just that their subs are not good enough for better positions. Tbh, I can't really give them the benefit of a doubt here as it's too obvious. ", "votes": 1 } ] }, { "id": 937286, "postDate": "2020-07-20T22:46:21.377Z", "rawMarkdown": "", "isDeleted": true }, { "id": 937238, "postDate": "2020-07-20T20:54:47.440Z", "rawMarkdown": "", "isDeleted": true }, { "id": 937207, "postDate": "2020-07-20T20:11:33.713Z", "rawMarkdown": "", "votes": 3, "isDeleted": true }, { "id": 937206, "postDate": "2020-07-20T20:10:58.200Z", "rawMarkdown": "", "isDeleted": true }, { "id": 937069, "postDate": "2020-07-20T17:46:17.107Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 936956, "author_name": "robga", "author_url": "", "post_date": "2020-07-20T16:15:24.143000", "content": "\nIt is entirely feasible that it is a complete coincidence ... for at least 8 teams to suddenly score .936 within 2 hours of submissions**, and to join just 9 other teams on .936 after weeks and months of submissions. Isn't it? Maybe? Coincidences coincidences. Two teams have users with similar username, Two other teams share similar username in opposite team.
You DO see new joiners, and in teams, in research challenges. They can be less used to the gamification of live leaderboards, happy to make their first submissions with hours to go. Of course, teams had to be formed 7 days ago.
\n\n**Downloadable public.csv shows 7 of these new teams with .936 submissions were made within a 2 hour period. Coincidences do happen?
", "votes": 19, "replies": [ { "id": 936967, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T16:20:29.503000", "content": "I hesitate to take a screenshot to share...oh man π
", "votes": 2, "replies": [] }, { "id": 936978, "author_name": "robga", "author_url": "", "post_date": "2020-07-20T16:25:37.917000", "content": "Anyway well done all of them.
\n\nI hope in the interests of research we get a large amount of teams sharing their solutions no matter their leaderboard place. The knowledge I'll learn will make up for dropping an additional 10 positions in a day.
", "votes": 3, "replies": [] }, { "id": 936979, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T16:27:13.940000", "content": "hmm, same here. =)
", "votes": 1, "replies": [] }, { "id": 937593, "author_name": "Vishnu R", "author_url": "", "post_date": "2020-07-21T05:10:54.040000", "content": "Yes @robga at the end of the day everything is knowledge.
", "votes": 1, "replies": [] }, { "id": 938006, "author_name": "robga", "author_url": "", "post_date": "2020-07-21T09:45:15.510000", "content": "And here is a screenshot of the 'after'. The odds of a moving as a group to the same private score is very low indeed. Disappointed this happens on the site.\n
EDIT 27/7: Leaderboard finalised and 9 offending teams in this screenshot removed. Thank you to Kaggle for your excellent compliance team, it is a valued service to the community.
", "votes": 3, "replies": [] } ] }, { "id": 936949, "author_name": "Eugene Khvedchenya", "author_url": "", "post_date": "2020-07-20T16:11:38.103000", "content": "Donβt start witch-hunt too early. I easily can assume solving this challenge using only local validation. This is the only reasonable way when you have such small test set. \nIn case this is a fake account - kaggle will eventually ban it. But before this happened - just keep calm.
", "votes": 7, "replies": [] }, { "id": 937057, "author_name": "Psi", "author_url": "", "post_date": "2020-07-20T17:40:05.370000", "content": "I usually dont jump to conclusions early, but this looks very obvious.
", "votes": 8, "replies": [] }, { "id": 936884, "author_name": "Ashraf Mahdhi", "author_url": "", "post_date": "2020-07-20T15:13:56.710000", "content": "it's strange to see people that are submitting less than 10 submissions and getting the Top of LB and the most of them , they were joining only this competition and kaggle recently ! Any explanation ?
", "votes": 5, "replies": [ { "id": 936986, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T16:30:08.967000", "content": "Nothing actually! Things are very clear to everyone.
", "votes": 6, "replies": [] } ] }, { "id": 937031, "author_name": "Nischay Dhankhar", "author_url": "", "post_date": "2020-07-20T17:16:17.290000", "content": "Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π
", "votes": 3, "replies": [ { "id": 937597, "author_name": "Vishnu R", "author_url": "", "post_date": "2020-07-21T05:13:52.543000", "content": "You are right @nischaydnk It cant be a coincidence
", "votes": 1, "replies": [] } ] }, { "id": 937162, "author_name": "Dieter", "author_url": "", "post_date": "2020-07-20T19:14:54.693000", "content": "and then there was a legendary single sub (idle) speculation
\n\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/leaderboard
", "votes": 4, "replies": [] }, { "id": 937239, "author_name": "Psi", "author_url": "", "post_date": "2020-07-20T20:55:09.217000", "content": "You only have 5 subs per day. Having more subs is a huge advantage, that's why people open new accounts.
\n\n(this should be a reply to @remicogranne but it always ends up separately)
", "votes": 3, "replies": [] }, { "id": 937350, "author_name": "Md Fahim", "author_url": "", "post_date": "2020-07-21T00:34:23.073000", "content": "A gold medal position ( in Private LB) with only 6 submission and Only one single submission got bronze medal π²
", "votes": 1, "replies": [] }, { "id": 937199, "author_name": "RΓ©mi Cogranne", "author_url": "", "post_date": "2020-07-20T19:55:51.890000", "content": "Hello there,
\nSInce I am a total newbie on Kaggle (and acknowledge this matter of fact) I don't see how you can \"cheat\" ?
\nI mean arguably, you can presume that those people, whose scores are almost identical and submission almost simultaneous, are working together and sharing information. However, in such a case it would be very silly to submit using different account (and why do so ?)
\nHowever, I also strongly believe that since the training and testing dataset are extracted from the same sources of images (subsampling of the same original dataset, images processed in the same way) you can train a lot offline and take your chance only at the very end in order not to reveal your game right ?
\nSorry for newbie question, it's just that it does not seems so obvious to me and I don't get the point of acting this was :c
By creating a new ID, the contestant can make additional submissions per day. If in the last few weeks, one has many experiments and not enough submission slots, this would allow the contestant to try out different experiments to the leaderboard.
", "votes": 3, "replies": [] } ] }, { "id": 937156, "author_name": "inversion", "author_url": "", "post_date": "2020-07-20T19:11:15.797000", "content": "I was a Kaggle competitor during the Rossmann competition. For the first month or so, the eventual winner stayed on the number one spot of the leaderboard with only a few submissions. He just relied on local validation and didn't submit very often. It's definitely possible. (But, yeah, I know what you mean.)
\nhttps://www.kaggle.com/c/rossmann-store-sales/leaderboard
", "votes": 1, "replies": [] }, { "id": 937084, "author_name": "Dieter", "author_url": "", "post_date": "2020-07-20T17:58:09.580000", "content": "Happend before.
\n\nhttps://www.kaggle.com/c/imaterialist-challenge-furniture-2018/leaderboard
", "votes": 2, "replies": [ { "id": 937086, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T18:01:11.353000", "content": "So, things are repeated and there's no concern from kaggle team. Is that it?
", "votes": 0, "replies": [] }, { "id": 937137, "author_name": "Will Cukierski", "author_url": "", "post_date": "2020-07-20T18:48:39.947000", "content": "\n\nSo, things are repeated and there's no concern from kaggle team. Is that it?
\n
We take allegations of cheating seriously and actively refine the methods we use to screen for cheating. We do not publicly speculate on ongoing accusations, nor make public examples of users who we do remove for violating the rules. We base our compliance decisions on a standard of evidence aimed at minimizing false positives. The minimum penalty for violating competition rules is removal from the competition, while repeat, egregious, or deliberate violations can warrant account removal.
\nThere is no question that some have cheated and gotten away with it, some have tried and been caught, but \"no concern\" is not an accurate characterization of our stance or history on this issue.
\nFriendly reminder: please use our compliance form to report evidence of cheating.
", "votes": 8, "replies": [] }, { "id": 937148, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T19:00:07.433000", "content": "@wcukierski \nThank you. Understood.
", "votes": 2, "replies": [] } ] }, { "id": 937071, "author_name": "Innat", "author_url": "", "post_date": "2020-07-20T17:47:01.770000", "content": "\n\n\nNischay Dhankhar wrote:\n Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π
\n
Like they said, Lucky! π
No I don't call it karma at all, they just probably have the similar blends which end up in roughly same places after shake up too, it's just that their subs are not good enough for better positions. Tbh, I can't really give them the benefit of a doubt here as it's too obvious.
", "votes": 1, "replies": [] } ] }, { "id": 937286, "author_name": "", "author_url": "", "post_date": "2020-07-20T22:46:21.377000", "content": "", "votes": 0, "replies": [] }, { "id": 937238, "author_name": "", "author_url": "", "post_date": "2020-07-20T20:54:47.440000", "content": "", "votes": 0, "replies": [] }, { "id": 937207, "author_name": "", "author_url": "", "post_date": "2020-07-20T20:11:33.713000", "content": "", "votes": 3, "replies": [] }, { "id": 937206, "author_name": "", "author_url": "", "post_date": "2020-07-20T20:10:58.200000", "content": "", "votes": 0, "replies": [] }, { "id": 937069, "author_name": "", "author_url": "", "post_date": "2020-07-20T17:46:17.107000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "936956": "\nIt is entirely feasible that it is a complete coincidence ... for at least 8 teams to suddenly score .936 within 2 hours of submissions**, and to join just 9 other teams on .936 after weeks and months of submissions. Isn't it? Maybe? Coincidences coincidences. Two teams have users with similar username, Two other teams share similar username in opposite team. \n\nYou DO see new joiners, and in teams, in research challenges. They can be less used to the gamification of live leaderboards, happy to make their first submissions with hours to go. Of course, teams had to be formed 7 days ago.\n\n**Downloadable public.csv shows 7 of these new teams with .936 submissions were made within a 2 hour period. Coincidences do happen?\n\n", "936814": "Excited to read those modeling approach. =)", "936949": "Donβt start witch-hunt too early. I easily can assume solving this challenge using only local validation. This is the only reasonable way when you have such small test set. \nIn case this is a fake account - kaggle will eventually ban it. But before this happened - just keep calm.", "937057": "I usually dont jump to conclusions early, but this looks very obvious.", "936884": "it's strange to see people that are submitting less than 10 submissions and getting the Top of LB and the most of them , they were joining only this competition and kaggle recently ! Any explanation ?", "937031": "Amongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π", "937162": "and then there was a legendary single sub (idle) speculation\n\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/leaderboard", "937239": "You only have 5 subs per day. Having more subs is a huge advantage, that's why people open new accounts.\n\n(this should be a reply to @remicogranne but it always ends up separately)", "937350": "A gold medal position ( in Private LB) with only 6 submission and Only one single submission got bronze medal π²", "937199": "Hello there,\nSInce I am a total newbie on Kaggle (and acknowledge this matter of fact) I don't see how you can \"cheat\" ?\nI mean arguably, you can presume that those people, whose scores are almost identical and submission almost simultaneous, are working together and sharing information. However, in such a case it would be very silly to submit using different account (and why do so ?)\nHowever, I also strongly believe that since the training and testing dataset are extracted from the same sources of images (subsampling of the same original dataset, images processed in the same way) you can train a lot offline and take your chance only at the very end in order not to reveal your game right ?\nSorry for newbie question, it's just that it does not seems so obvious to me and I don't get the point of acting this was :c", "937156": "I was a Kaggle competitor during the Rossmann competition. For the first month or so, the eventual winner stayed on the number one spot of the leaderboard with only a few submissions. He just relied on local validation and didn't submit very often. It's definitely possible. (But, yeah, I know what you mean.)\n\nhttps://www.kaggle.com/c/rossmann-store-sales/leaderboard", "937084": "Happend before. \n\nhttps://www.kaggle.com/c/imaterialist-challenge-furniture-2018/leaderboard\n\n", "937071": "> **Nischay Dhankhar** wrote:\nAmongst them, different teams have some members who've joined 7 days ago on Kaggle. Is it a coincidence too?π\n\nLike they said, `Lucky`! π ", "937340": "", "937286": "", "937238": "", "937207": "", "937206": "", "937069": "" } }