{
  "id": 12467,
  "title": "Cheaters Removed",
  "url": "/competitions/avazu-ctr-prediction/discussion/12467",
  "author_name": "",
  "post_date": "2015-02-10T16:08:30.013Z",
  "votes": 21,
  "comment_count": 33,
  "views": 12393,
  "content": "<p>We have removed suspected cheaters (and a shameful number, at that) from the leaderboard. If your team was removed from the leaderboard and you believe it was in error, please email compliance@kaggle.com to appeal. Include an explanation for your statistically improbable similarities to other teams.</p>\n<p>As always, thanks to those who play by the rules.</p>",
  "messages": [
    {
      "id": "63929",
      "postDate": "02/10/2015 16:08:30",
      "content": "<p>We have removed suspected cheaters (and a shameful number, at that) from the leaderboard. If your team was removed from the leaderboard and you believe it was in error, please email compliance@kaggle.com to appeal. Include an explanation for your statistically improbable similarities to other teams.</p>\n<p>As always, thanks to those who play by the rules.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63931",
      "postDate": "02/10/2015 16:34:35",
      "content": "<p>It was pretty obvious&nbsp;there was a lot of cheating. Looks like 130 or so got removed.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63934",
      "postDate": "02/10/2015 16:41:05",
      "content": "<p>130? it was 1616 earlier and 1604 now - I must have missed the original number. (I bet, as usual, the puppetmasters remain on the LB!)</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63937",
      "postDate": "02/10/2015 16:52:22",
      "content": "<p>I checked right at the contest end and&nbsp;saw 1736.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63938",
      "postDate": "02/10/2015 16:53:52",
      "content": "<p>Blimey - I've been decorating for a couple of months so out of the loop! That's a large number of cheaters!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63939",
      "postDate": "02/10/2015 17:11:19",
      "content": "<p>There will be some false positive cause a lots of people use similar beat the benchmark starter code.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63941",
      "postDate": "02/10/2015 17:20:19",
      "content": "<p>I guess there weren't many in the top 100 because I haven't moved at all from yesterday. I thought Kaggle only looked for cheaters in the top 100... but just a quick visual inspection on the LB shows some other suspects (like the 240-260 area).</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63947",
      "postDate": "02/10/2015 18:06:56",
      "content": "<p><img src=\"http://sd.keepcalm-o-matic.co.uk/i/keep-calm-and-hate-cheaters-2.png\" alt=\"Cheaters!\" width=\"250\" height=\"154\"></p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63951",
      "postDate": "02/10/2015 19:00:46",
      "content": "<p>[quote=Domcastro;63934]</p>\n<p>130? it was 1616 earlier and 1604 now - I must have missed the original number. (I bet, as usual, the puppetmasters remain on the LB!)</p>\n<p>[/quote]</p>\n<p>I'm not sure how you would know the puppetmasters remain on the LB. Because if you know this for certain, can you at least post the data and algorithms to show this is true?&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63952",
      "postDate": "02/10/2015 19:11:10",
      "content": "<p>[quote=Giulio;63941]</p>\n<p>I guess there weren't many in the top 100 because I haven't moved at all from yesterday. I thought Kaggle only looked for cheaters in the top 100... but just a quick visual inspection on the LB shows some other suspects (like the 240-260 area).</p>\n<p>[/quote]</p>\n<p>I think I was 31st on public LB and dropped to 34th or 36th for a last 2-3 days. Now I'm 31st on public LB again (and 29th on private). So there should be some.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63967",
      "postDate": "02/10/2015 20:53:35",
      "content": "<p>When will leaderboard be final?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63972",
      "postDate": "02/10/2015 21:27:38",
      "content": "<p>Tomorrow</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63975",
      "postDate": "02/10/2015 22:15:22",
      "content": "<p>Thanks Will! Also noted is the progress counter when submitting is a huge improvement over the prior counter.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63984",
      "postDate": "02/11/2015 00:30:26",
      "content": "<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63985",
      "postDate": "02/11/2015 00:36:22",
      "content": "<p>It's not my place or desire to publicly discuss specific situations. That said, we don't remove teams without overwhelming evidence of a rules violation.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63987",
      "postDate": "02/11/2015 00:47:00",
      "content": "<p>[quote=David Thaler;63984]</p>\n<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>\n<p>[/quote]</p>\n<p>Irrespective of whether it is true or not,I am not surprised because of that:&nbsp;https://www.kaggle.com/c/yelp-recsys-2013/forums/t/5607/important-problem-of-invisible-cheating/29916#post29916 .</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63991",
      "postDate": "02/11/2015 01:22:38",
      "content": "<p>[quote=KazAnova;63987]</p>\n<p>[quote=David Thaler;63984]</p>\n<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>\n<p>[/quote]</p>\n<p>Irrespective of whether it is true or not,I am not surprised because of that:&nbsp;https://www.kaggle.com/c/yelp-recsys-2013/forums/t/5607/important-problem-of-invisible-cheating/29916#post29916 .</p>\n<p>[/quote]</p>\n<p>This too:</p>\n<p>https://www.kaggle.com/c/GiveMeSomeCredit/forums/t/1153/magic-team-migration/7141#post7141</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "63993",
      "postDate": "02/11/2015 02:34:29",
      "content": "<p>Hat tip to KazAnova and dynamic24 for digging up that bit of Kaggle history. It all makes sense now.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64120",
      "postDate": "02/12/2015 16:28:51",
      "content": "<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64121",
      "postDate": "02/12/2015 16:36:08",
      "content": "<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>That is an excellent point - in fact that would promote collaboration even more. so far (and that is my current way thinking), I have avoided merging with people they have not a proven record of multiple competitions in their arsenal, just because I am worried about cheating...I think it is a shame for a person-in-a-group to get penalized because another one has cheated - but difficult to track too. Maybe rc's point is to some extend the solution.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64126",
      "postDate": "02/12/2015 16:56:36",
      "content": "<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>Yikes. Do team mates of cheaters get removed from Kaggle, or just the contest?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64127",
      "postDate": "02/12/2015 17:00:50",
      "content": "<p>[quote=inversion;64126]</p>\n<p>Yikes. Do team mates of cheaters get removed from Kaggle, or just the contest?</p>\n<p>[/quote]</p>\n<p>I think just from the contest, but the&nbsp;the stink&nbsp;remains forever...</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64129",
      "postDate": "02/12/2015 17:42:13",
      "content": "<p>I'm actually going to team up for the first time. Are there some best practices so we&nbsp;don't inadvertently get flagged for cheating? Are there things I should ask the other team member before we make it official?&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64132",
      "postDate": "02/12/2015 18:13:13",
      "content": "<p>[quote=KazAnova;64121]</p>\n<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>That is an excellent point - in fact that would promote collaboration even more. so far (and that is my current way thinking), I have avoided merging with people they have not a proven record of multiple competitions in their arsenal, just because I am worried about cheating...I think it is a shame for a person-in-a-group to get penalized because another one has cheated - but difficult to track too. Maybe rc's point is to some extend the solution.&nbsp;</p>\n<p>[/quote]</p>\n<p>We have experimented with banning teams during the competition. All it does is inform the cheater that they were busted,&nbsp;gives&nbsp;time to make new accounts, and makes them more careful to hide their tracks.</p>\n<p>We'll do some thinking about your points on trust.&nbsp;I agree it requires a lot of faith (and it should, since it can&nbsp;have financial ramifications). The problem with warning about a potential cheater is firstly that the review process is manual, secondly that a warning would leak information and could be abused, and thirdly that&nbsp;we've seen cases&nbsp;where only one of the members was cheating, unbeknownst to the others (after the merger had occurred).</p>\n<p>The best way to handle now is to establish a clear, written agreement that sets&nbsp;expectations about proper behavior, division of labor, IP, sharing, &amp; prizes.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64152",
      "postDate": "02/12/2015 23:52:41",
      "content": "<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one. It's got to be done though. There's also been some Masters and people with lots of competitions who have been in teams where someone has cheated - so it really is hard to discern &quot;trustworthyness&quot;. I'd be mortified if I got removed from the leaderboard. But saying that, I've been in teams with 7 different people and have had no problems. So, I think it is a minority rather than majority</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64170",
      "postDate": "02/13/2015 16:13:04",
      "content": "<p>[quote=Bluefool;64152]</p>\n<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one.</p>\n<p>[/quote]</p>\n<p>I was not offended .I felt I had to quickly destroy all the evidence (delete my extra kaggle accounts, namely, Triskelion ,Phill Culiton and Abhishek , hide plagiarized papers, burn my 3 different passports, do a plastic surgery etc &nbsp;)!&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64171",
      "postDate": "02/13/2015 16:17:30",
      "content": "<p>[quote=KazAnova;64170]</p>\n<p>[quote=Bluefool;64152]</p>\n<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one.</p>\n<p>[/quote]</p>\n<p>I was not offended .I felt I had to quickly destroy all the evidence (delete my extra kaggle accounts, namely, Triskelion ,Phill Culiton and Abhishek , hide plagiarized papers, burn my 3 different passports, do a plastic surgery etc &nbsp;)!&nbsp;</p>\n<p>[/quote]</p>\n<p>hhahahaa :D :D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64172",
      "postDate": "02/13/2015 16:18:29",
      "content": "<p>:D</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64174",
      "postDate": "02/13/2015 18:33:04",
      "content": "<p>ha!</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64227",
      "postDate": "02/14/2015 13:33:59",
      "content": "<p>Lesson learned, Don't give away your results&nbsp;if you want to exit a competition.</p>\n<p>Send your hard disk to KazAnova and ask her to dispose it.</p>\n<p>https://www.kaggle.com/c/avazu-ctr-prediction/forums/t/12320/team-up</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64230",
      "postDate": "02/14/2015 15:44:10",
      "content": "<p>[quote=superfan123;64227]</p>\n<p>ask her</p>\n<p>[/quote]</p>\n<p>I do have long hair now, but I would still call it a &quot;him&quot; :) . True, I am good at disposing.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64232",
      "postDate": "02/14/2015 16:32:52",
      "content": "<p>Oops. I guess the 'plastic surgery' part is misleading.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "64418",
      "postDate": "02/17/2015 19:59:57",
      "content": "<p>[quote=KazAnova;64230]</p>\n<p>[quote=superfan123;64227]</p>\n<p>ask her</p>\n<p>[/quote]</p>\n<p>I do have long hair now, but I would still call it a &quot;him&quot; :) . True, I am good at disposing.</p>\n<p>[/quote]</p>\n\n<p>hhahahahah cruely laughs</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "81578",
      "postDate": "06/11/2015 05:50:53",
      "content": "<p>I can see, there was a very intense discussion on this topic.</p>\n<p>https://www.kaggle.com/c/restaurant-revenue-prediction/forums/t/14131/multiple-accounts</p>\n<p>I can not be sure that above point will be the final, but I am feeling a necessity to put it here.</p>\n<p>&quot;The Home of Data Science&quot; is an illustration of another incorrect statement made by Kaggle.<br>Clearly, this claim is too ambitious if not ridiculous.</p>\n<p>Many final results announced by Kaggle contradict to the most fundamental principles of data science and machine learning, and it appears that Kaggle is not concerned about it at all..</p>\n<p>By definition, Kaggle general orientation is a commercial, but not a scientific.<br>In the other terms, 1) to write a serious scientific paper, and 2) to win any Kaggle competition <br>will require different skills.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 63931,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "02/10/2015 16:34:35",
      "content": "<p>It was pretty obvious&nbsp;there was a lot of cheating. Looks like 130 or so got removed.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63934,
      "author_name": "domcastro",
      "author_url": "",
      "post_date": "02/10/2015 16:41:05",
      "content": "<p>130? it was 1616 earlier and 1604 now - I must have missed the original number. (I bet, as usual, the puppetmasters remain on the LB!)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63937,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "02/10/2015 16:52:22",
      "content": "<p>I checked right at the contest end and&nbsp;saw 1736.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63938,
      "author_name": "domcastro",
      "author_url": "",
      "post_date": "02/10/2015 16:53:52",
      "content": "<p>Blimey - I've been decorating for a couple of months so out of the loop! That's a large number of cheaters!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63939,
      "author_name": "superfan123",
      "author_url": "",
      "post_date": "02/10/2015 17:11:19",
      "content": "<p>There will be some false positive cause a lots of people use similar beat the benchmark starter code.&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63941,
      "author_name": "adjgiulio",
      "author_url": "",
      "post_date": "02/10/2015 17:20:19",
      "content": "<p>I guess there weren't many in the top 100 because I haven't moved at all from yesterday. I thought Kaggle only looked for cheaters in the top 100... but just a quick visual inspection on the LB shows some other suspects (like the 240-260 area).</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63947,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/10/2015 18:06:56",
      "content": "<p><img src=\"http://sd.keepcalm-o-matic.co.uk/i/keep-calm-and-hate-cheaters-2.png\" alt=\"Cheaters!\" width=\"250\" height=\"154\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63951,
      "author_name": "mikeskim",
      "author_url": "",
      "post_date": "02/10/2015 19:00:46",
      "content": "<p>[quote=Domcastro;63934]</p>\n<p>130? it was 1616 earlier and 1604 now - I must have missed the original number. (I bet, as usual, the puppetmasters remain on the LB!)</p>\n<p>[/quote]</p>\n<p>I'm not sure how you would know the puppetmasters remain on the LB. Because if you know this for certain, can you at least post the data and algorithms to show this is true?&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63952,
      "author_name": "trufanov",
      "author_url": "",
      "post_date": "02/10/2015 19:11:10",
      "content": "<p>[quote=Giulio;63941]</p>\n<p>I guess there weren't many in the top 100 because I haven't moved at all from yesterday. I thought Kaggle only looked for cheaters in the top 100... but just a quick visual inspection on the LB shows some other suspects (like the 240-260 area).</p>\n<p>[/quote]</p>\n<p>I think I was 31st on public LB and dropped to 34th or 36th for a last 2-3 days. Now I'm 31st on public LB again (and 29th on private). So there should be some.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63967,
      "author_name": "abhishek",
      "author_url": "",
      "post_date": "02/10/2015 20:53:35",
      "content": "<p>When will leaderboard be final?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63972,
      "author_name": "wcukierski",
      "author_url": "",
      "post_date": "02/10/2015 21:27:38",
      "content": "<p>Tomorrow</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63975,
      "author_name": "mikeskim",
      "author_url": "",
      "post_date": "02/10/2015 22:15:22",
      "content": "<p>Thanks Will! Also noted is the progress counter when submitting is a huge improvement over the prior counter.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63984,
      "author_name": "davidthaler",
      "author_url": "",
      "post_date": "02/11/2015 00:30:26",
      "content": "<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63985,
      "author_name": "wcukierski",
      "author_url": "",
      "post_date": "02/11/2015 00:36:22",
      "content": "<p>It's not my place or desire to publicly discuss specific situations. That said, we don't remove teams without overwhelming evidence of a rules violation.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63987,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/11/2015 00:47:00",
      "content": "<p>[quote=David Thaler;63984]</p>\n<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>\n<p>[/quote]</p>\n<p>Irrespective of whether it is true or not,I am not surprised because of that:&nbsp;https://www.kaggle.com/c/yelp-recsys-2013/forums/t/5607/important-problem-of-invisible-cheating/29916#post29916 .</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63991,
      "author_name": "dynamic24",
      "author_url": "",
      "post_date": "02/11/2015 01:22:38",
      "content": "<p>[quote=KazAnova;63987]</p>\n<p>[quote=David Thaler;63984]</p>\n<p>Will,&nbsp;</p>\n<p>It looks to me like Silogram got nailed. That surprises me. Are you sure that's right? I think he was on a team with Vladimir Nikulin.</p>\n<p>[/quote]</p>\n<p>Irrespective of whether it is true or not,I am not surprised because of that:&nbsp;https://www.kaggle.com/c/yelp-recsys-2013/forums/t/5607/important-problem-of-invisible-cheating/29916#post29916 .</p>\n<p>[/quote]</p>\n<p>This too:</p>\n<p>https://www.kaggle.com/c/GiveMeSomeCredit/forums/t/1153/magic-team-migration/7141#post7141</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 63993,
      "author_name": "davidthaler",
      "author_url": "",
      "post_date": "02/11/2015 02:34:29",
      "content": "<p>Hat tip to KazAnova and dynamic24 for digging up that bit of Kaggle history. It all makes sense now.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64120,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "02/12/2015 16:28:51",
      "content": "<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64121,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/12/2015 16:36:08",
      "content": "<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>That is an excellent point - in fact that would promote collaboration even more. so far (and that is my current way thinking), I have avoided merging with people they have not a proven record of multiple competitions in their arsenal, just because I am worried about cheating...I think it is a shame for a person-in-a-group to get penalized because another one has cheated - but difficult to track too. Maybe rc's point is to some extend the solution.&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64126,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "02/12/2015 16:56:36",
      "content": "<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>Yikes. Do team mates of cheaters get removed from Kaggle, or just the contest?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64127,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/12/2015 17:00:50",
      "content": "<p>[quote=inversion;64126]</p>\n<p>Yikes. Do team mates of cheaters get removed from Kaggle, or just the contest?</p>\n<p>[/quote]</p>\n<p>I think just from the contest, but the&nbsp;the stink&nbsp;remains forever...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64129,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "02/12/2015 17:42:13",
      "content": "<p>I'm actually going to team up for the first time. Are there some best practices so we&nbsp;don't inadvertently get flagged for cheating? Are there things I should ask the other team member before we make it official?&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64132,
      "author_name": "wcukierski",
      "author_url": "",
      "post_date": "02/12/2015 18:13:13",
      "content": "<p>[quote=KazAnova;64121]</p>\n<p>[quote=rcarson;64120]</p>\n<p>Team merging is risky. Is it possible that kaggle starts cheater detection from the beginning of the contest, so that we could get warnings from kaggle when we try to team up with someone who is suspicious to violate rules?</p>\n<p>[/quote]</p>\n<p>That is an excellent point - in fact that would promote collaboration even more. so far (and that is my current way thinking), I have avoided merging with people they have not a proven record of multiple competitions in their arsenal, just because I am worried about cheating...I think it is a shame for a person-in-a-group to get penalized because another one has cheated - but difficult to track too. Maybe rc's point is to some extend the solution.&nbsp;</p>\n<p>[/quote]</p>\n<p>We have experimented with banning teams during the competition. All it does is inform the cheater that they were busted,&nbsp;gives&nbsp;time to make new accounts, and makes them more careful to hide their tracks.</p>\n<p>We'll do some thinking about your points on trust.&nbsp;I agree it requires a lot of faith (and it should, since it can&nbsp;have financial ramifications). The problem with warning about a potential cheater is firstly that the review process is manual, secondly that a warning would leak information and could be abused, and thirdly that&nbsp;we've seen cases&nbsp;where only one of the members was cheating, unbeknownst to the others (after the merger had occurred).</p>\n<p>The best way to handle now is to establish a clear, written agreement that sets&nbsp;expectations about proper behavior, division of labor, IP, sharing, &amp; prizes.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64152,
      "author_name": "domcastro",
      "author_url": "",
      "post_date": "02/12/2015 23:52:41",
      "content": "<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one. It's got to be done though. There's also been some Masters and people with lots of competitions who have been in teams where someone has cheated - so it really is hard to discern &quot;trustworthyness&quot;. I'd be mortified if I got removed from the leaderboard. But saying that, I've been in teams with 7 different people and have had no problems. So, I think it is a minority rather than majority</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64170,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/13/2015 16:13:04",
      "content": "<p>[quote=Bluefool;64152]</p>\n<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one.</p>\n<p>[/quote]</p>\n<p>I was not offended .I felt I had to quickly destroy all the evidence (delete my extra kaggle accounts, namely, Triskelion ,Phill Culiton and Abhishek , hide plagiarized papers, burn my 3 different passports, do a plastic surgery etc &nbsp;)!&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64171,
      "author_name": "abhishek",
      "author_url": "",
      "post_date": "02/13/2015 16:17:30",
      "content": "<p>[quote=KazAnova;64170]</p>\n<p>[quote=Bluefool;64152]</p>\n<p>It is scary teaming up. I remember offending Kazanova asking if he had ever cheated before we teamed up in the Africa one.</p>\n<p>[/quote]</p>\n<p>I was not offended .I felt I had to quickly destroy all the evidence (delete my extra kaggle accounts, namely, Triskelion ,Phill Culiton and Abhishek , hide plagiarized papers, burn my 3 different passports, do a plastic surgery etc &nbsp;)!&nbsp;</p>\n<p>[/quote]</p>\n<p>hhahahaa :D :D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64172,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/13/2015 16:18:29",
      "content": "<p>:D</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64174,
      "author_name": "domcastro",
      "author_url": "",
      "post_date": "02/13/2015 18:33:04",
      "content": "<p>ha!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64227,
      "author_name": "superfan123",
      "author_url": "",
      "post_date": "02/14/2015 13:33:59",
      "content": "<p>Lesson learned, Don't give away your results&nbsp;if you want to exit a competition.</p>\n<p>Send your hard disk to KazAnova and ask her to dispose it.</p>\n<p>https://www.kaggle.com/c/avazu-ctr-prediction/forums/t/12320/team-up</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64230,
      "author_name": "kazanova",
      "author_url": "",
      "post_date": "02/14/2015 15:44:10",
      "content": "<p>[quote=superfan123;64227]</p>\n<p>ask her</p>\n<p>[/quote]</p>\n<p>I do have long hair now, but I would still call it a &quot;him&quot; :) . True, I am good at disposing.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64232,
      "author_name": "superfan123",
      "author_url": "",
      "post_date": "02/14/2015 16:32:52",
      "content": "<p>Oops. I guess the 'plastic surgery' part is misleading.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 64418,
      "author_name": "domcastro",
      "author_url": "",
      "post_date": "02/17/2015 19:59:57",
      "content": "<p>[quote=KazAnova;64230]</p>\n<p>[quote=superfan123;64227]</p>\n<p>ask her</p>\n<p>[/quote]</p>\n<p>I do have long hair now, but I would still call it a &quot;him&quot; :) . True, I am good at disposing.</p>\n<p>[/quote]</p>\n\n<p>hhahahahah cruely laughs</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 81578,
      "author_name": "",
      "author_url": "",
      "post_date": "06/11/2015 05:50:53",
      "content": "<p>I can see, there was a very intense discussion on this topic.</p>\n<p>https://www.kaggle.com/c/restaurant-revenue-prediction/forums/t/14131/multiple-accounts</p>\n<p>I can not be sure that above point will be the final, but I am feeling a necessity to put it here.</p>\n<p>&quot;The Home of Data Science&quot; is an illustration of another incorrect statement made by Kaggle.<br>Clearly, this claim is too ambitious if not ridiculous.</p>\n<p>Many final results announced by Kaggle contradict to the most fundamental principles of data science and machine learning, and it appears that Kaggle is not concerned about it at all..</p>\n<p>By definition, Kaggle general orientation is a commercial, but not a scientific.<br>In the other terms, 1) to write a serious scientific paper, and 2) to win any Kaggle competition <br>will require different skills.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "63929": "",
    "63931": "",
    "63934": "",
    "63937": "",
    "63938": "",
    "63939": "",
    "63941": "",
    "63947": "",
    "63951": "",
    "63952": "",
    "63967": "",
    "63972": "",
    "63975": "",
    "63984": "",
    "63985": "",
    "63987": "",
    "63991": "",
    "63993": "",
    "64120": "",
    "64121": "",
    "64126": "",
    "64127": "",
    "64129": "",
    "64132": "",
    "64152": "",
    "64170": "",
    "64171": "",
    "64172": "",
    "64174": "",
    "64227": "",
    "64230": "",
    "64232": "",
    "64418": "",
    "81578": ""
  },
  "source": "meta"
}