(I am not mentionning any names for I have no hard evidence and didn't really pay attention to the leaderboard evolution in this competiton, but what I am highlighting has been occuring in other comps as well)
\nAs most of you already now, it is forbidden to share code with other competitors that are not in your team. This also includes people you are going to team up with, when you haven't already merged.
\nIt is (sometimes) quite easy to spot people that know each other in the leaderboard by having a quick look at the profiles (previous competitions, country of origin, employer, …).
\nThe fact I am seeing people of such clusters that have been competiting solo during the whole competition merge the day of the deadline, usually makes me think they were working together before merging.
\nWhat puzzles me is that the only advantage of sharing before merging is that you can have some days where you submit more than the limit, as long as you submit less other says. Please do everybody a favor and don't wait for the last moment to merge, i.e. respect the rules.
\nI don't want to start a witch hunt here, partly because I'd rather see such people merge than try to grab a bunch of solo gold medals. I also think some people genuinely don't know the rules when joining a competition, and therefore do not mean any harm.
", "messages": [ { "id": 1054179, "postDate": "2020-10-19T18:46:44.693Z", "content": "(I am not mentionning any names for I have no hard evidence and didn't really pay attention to the leaderboard evolution in this competiton, but what I am highlighting has been occuring in other comps as well)
\nAs most of you already now, it is forbidden to share code with other competitors that are not in your team. This also includes people you are going to team up with, when you haven't already merged.
\nIt is (sometimes) quite easy to spot people that know each other in the leaderboard by having a quick look at the profiles (previous competitions, country of origin, employer, …).
\nThe fact I am seeing people of such clusters that have been competiting solo during the whole competition merge the day of the deadline, usually makes me think they were working together before merging.
\nWhat puzzles me is that the only advantage of sharing before merging is that you can have some days where you submit more than the limit, as long as you submit less other says. Please do everybody a favor and don't wait for the last moment to merge, i.e. respect the rules.
\nI don't want to start a witch hunt here, partly because I'd rather see such people merge than try to grab a bunch of solo gold medals. I also think some people genuinely don't know the rules when joining a competition, and therefore do not mean any harm.
", "rawMarkdown": "(I am not mentionning any names for I have no hard evidence and didn't really pay attention to the leaderboard evolution in this competiton, but what I am highlighting has been occuring in other comps as well)\n\nAs most of you already now, it is forbidden to share code with other competitors that are not in your team. This also includes people you are going to team up with, when you haven't already merged. \n\nIt is (sometimes) quite easy to spot people that know each other in the leaderboard by having a quick look at the profiles (previous competitions, country of origin, employer, ...). \n\nThe fact I am seeing people of such clusters that have been competiting solo during the whole competition merge the day of the deadline, usually makes me think they were working together before merging.\n\nWhat puzzles me is that the only advantage of sharing before merging is that you can have some days where you submit more than the limit, as long as you submit less other says. Please do everybody a favor and don't wait for the last moment to merge, i.e. respect the rules. \n\nI don't want to start a witch hunt here, partly because I'd rather see such people merge than try to grab a bunch of solo gold medals. I also think some people genuinely don't know the rules when joining a competition, and therefore do not mean any harm.\n\n\n\n", "votes": 28 }, { "id": 1054358, "postDate": "2020-10-19T21:05:10.423Z", "content": "@theoviel hi, nice point!
\nBut I think most people do follow the rules and don't share code before merging.
\nI often see that a lot of people who are merged make enough few submissions, and they could easily do the same number of submissions in the team too without problems with limits.
\nI think that people want more time for THEMSELVES, to try all their OWN best practices, and only then brainstorm in the team what needs to be combined
Yeah, you are absolutely right
", "rawMarkdown": "Yeah, you are absolutely right", "votes": 1 } ] }, { "id": 1060643, "postDate": "2020-10-26T12:38:24.840Z", "content": "\n\nThis also includes people you are going to team up with, when you haven't already merged.
\n
Thi is your take.
\nIt would be good to have some clarification from Kaggle on that.
", "rawMarkdown": "> This also includes people you are going to team up with, when you haven't already merged. \n\nThi is your take.\n\nIt would be good to have some clarification from Kaggle on that.", "votes": 1, "replies": [ { "id": 1061083, "postDate": "2020-10-26T18:18:32.420Z", "content": "I've already seen people getting removed for such reason in the ARC competition.
\nBut it is likely that some (if not a lot of) people get away with such practices.
", "rawMarkdown": "I've already seen people getting removed for such reason in the [ARC](https://www.kaggle.com/c/abstraction-and-reasoning-challenge/discussion/150294) competition.\n\nBut it is likely that some (if not a lot of) people get away with such practices." } ] }, { "id": 1060299, "postDate": "2020-10-26T03:07:23.740Z", "content": "Well, sadly, some people didn't even merge.
", "rawMarkdown": "Well, sadly, some people didn't even merge.", "votes": 1 }, { "id": 1054851, "postDate": "2020-10-20T08:15:49.437Z", "content": "1st of PB ?
", "rawMarkdown": "1st of PB ?" }, { "id": 1056026, "postDate": "2020-10-21T11:18:58.583Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1054358, "author_name": "Yaroslav Isaienkov", "author_url": "", "post_date": "2020-10-19T21:05:10.423000", "content": "@theoviel hi, nice point!
\nBut I think most people do follow the rules and don't share code before merging.
\nI often see that a lot of people who are merged make enough few submissions, and they could easily do the same number of submissions in the team too without problems with limits.
\nI think that people want more time for THEMSELVES, to try all their OWN best practices, and only then brainstorm in the team what needs to be combined
Yeah, you are absolutely right
", "votes": 1, "replies": [] } ] }, { "id": 1060643, "author_name": "CPMP", "author_url": "", "post_date": "2020-10-26T12:38:24.840000", "content": "\n\nThis also includes people you are going to team up with, when you haven't already merged.
\n
Thi is your take.
\nIt would be good to have some clarification from Kaggle on that.
", "votes": 1, "replies": [ { "id": 1061083, "author_name": "Theo Viel", "author_url": "", "post_date": "2020-10-26T18:18:32.420000", "content": "I've already seen people getting removed for such reason in the ARC competition.
\nBut it is likely that some (if not a lot of) people get away with such practices.
", "votes": 0, "replies": [] } ] }, { "id": 1060299, "author_name": "sin", "author_url": "", "post_date": "2020-10-26T03:07:23.740000", "content": "Well, sadly, some people didn't even merge.
", "votes": 1, "replies": [] }, { "id": 1054851, "author_name": "Monojito", "author_url": "", "post_date": "2020-10-20T08:15:49.437000", "content": "1st of PB ?
", "votes": 0, "replies": [] }, { "id": 1056026, "author_name": "", "author_url": "", "post_date": "2020-10-21T11:18:58.583000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1054179": "(I am not mentionning any names for I have no hard evidence and didn't really pay attention to the leaderboard evolution in this competiton, but what I am highlighting has been occuring in other comps as well)\n\nAs most of you already now, it is forbidden to share code with other competitors that are not in your team. This also includes people you are going to team up with, when you haven't already merged. \n\nIt is (sometimes) quite easy to spot people that know each other in the leaderboard by having a quick look at the profiles (previous competitions, country of origin, employer, ...). \n\nThe fact I am seeing people of such clusters that have been competiting solo during the whole competition merge the day of the deadline, usually makes me think they were working together before merging.\n\nWhat puzzles me is that the only advantage of sharing before merging is that you can have some days where you submit more than the limit, as long as you submit less other says. Please do everybody a favor and don't wait for the last moment to merge, i.e. respect the rules. \n\nI don't want to start a witch hunt here, partly because I'd rather see such people merge than try to grab a bunch of solo gold medals. I also think some people genuinely don't know the rules when joining a competition, and therefore do not mean any harm.\n\n\n\n", "1054358": "@theoviel hi, nice point! \nBut I think most people do follow the rules and don't share code before merging.\nI often see that a lot of people who are merged make enough few submissions, and they could easily do the same number of submissions in the team too without problems with limits.\nI think that people want more time for THEMSELVES, to try all their OWN best practices, and only then brainstorm in the team what needs to be combined", "1060643": "> This also includes people you are going to team up with, when you haven't already merged. \n\nThi is your take.\n\nIt would be good to have some clarification from Kaggle on that.", "1060299": "Well, sadly, some people didn't even merge.", "1054851": "1st of PB ?", "1056026": "" } }