@juliaelliott We all noticed that you removed the two teams that admitted sharing before teaming. Until this competition this has been a gray area, with some abusing it. You made it clear this was against the rules ( as it has always been) and you acted accordingly.
\nThis is a very positive change in Kaggle IMHO.
\nThanks for acting.
\nEdit. Theo Viel reminds me of Ahmet Erdem key role here as well. Without his posts I am not sure we would have got this great move.
", "messages": [ { "id": "1209437", "postDate": "02/18/2021 22:31:25", "content": "@juliaelliott We all noticed that you removed the two teams that admitted sharing before teaming. Until this competition this has been a gray area, with some abusing it. You made it clear this was against the rules ( as it has always been) and you acted accordingly.
\nThis is a very positive change in Kaggle IMHO.
\nThanks for acting.
\nEdit. Theo Viel reminds me of Ahmet Erdem key role here as well. Without his posts I am not sure we would have got this great move.
", "rawMarkdown": "juliaelliott We all noticed that you removed the two teams that admitted sharing before teaming. Until this competition this has been a gray area, with some abusing it. You made it clear this was against the rules ( as it has always been) and you acted accordingly.\n\nThis is a very positive change in Kaggle IMHO.\n\nThanks for acting.\n\nEdit. Theo Viel reminds me of Ahmet Erdem key role here as well. Without his posts I am not sure we would have got this great move.", "votes": null }, { "id": "1209439", "postDate": "02/18/2021 22:36:53", "content": "Thanks to @aerdem4 as well :)
\nI believe that without his threads both teams would've gone unpunished, as some other teams most likely did.
Indeed. I thanked him privately but I should have mentioned him in the post. Fixed.
", "rawMarkdown": "Indeed. I thanked him privately but I should have mentioned him in the post. Fixed.", "votes": null }, { "id": "1209469", "postDate": "02/18/2021 23:13:04", "content": "Thank you kaggle team for doing the right rules and showing fairness to all teams!
", "rawMarkdown": "Thank you kaggle team for doing the right rules and showing fairness to all teams!", "votes": null }, { "id": "1209510", "postDate": "02/18/2021 23:53:22", "content": "While I feel sad for removed teams since their efforts are wasted, I feel happy for the other teams who got their deserved places.
", "rawMarkdown": "While I feel sad for removed teams since their efforts are wasted, I feel happy for the other teams who got their deserved places.", "votes": null }, { "id": "1209823", "postDate": "02/19/2021 03:58:53", "content": "Lucky we have kagglers like Ahmet Erdem.
", "rawMarkdown": "Lucky we have kagglers like Ahmet Erdem.", "votes": null }, { "id": "1209832", "postDate": "02/19/2021 04:05:51", "content": "Thanks for removing me from competition just because I merged with another team. Making all my solo efforts go waste
", "rawMarkdown": "Thanks for removing me from competition just because I merged with another team. Making all my solo efforts go waste", "votes": null }, { "id": "1209870", "postDate": "02/19/2021 04:43:35", "content": "I guess becareful who you merge with next time? was exciting seeing your team approach a higher lb score and was fun mentally battling it out with your team guessing what your team was doing, hope you come back up stronger and safer next time
", "rawMarkdown": "I guess becareful who you merge with next time? was exciting seeing your team approach a higher lb score and was fun mentally battling it out with your team guessing what your team was doing, hope you come back up stronger and safer next time", "votes": null }, { "id": "1209892", "postDate": "02/19/2021 04:57:22", "content": "Well how do one supposed to know whom to merge with, I guess not all are great detective like @aerdem4. This raises the concern of getting an option for unmerge. Maybe after unmerging you are not allowed to make more submissions, but one will not loose his efforts from the beginning.
", "rawMarkdown": "Well how do one supposed to know whom to merge with, I guess not all are great detective like @aerdem4. This raises the concern of getting an option for unmerge. Maybe after unmerging you are not allowed to make more submissions, but one will not loose his efforts from the beginning.", "votes": null }, { "id": "1210058", "postDate": "02/19/2021 07:18:27", "content": "bit unfortunate, I agree… a tremendous amount of effort would have gone in… My sympathies.
", "rawMarkdown": "bit unfortunate, I agree... a tremendous amount of effort would have gone in... My sympathies.", "votes": null }, { "id": "1210072", "postDate": "02/19/2021 07:34:36", "content": "@rsinda I am sorry that you are removed and unmerge with submission freeze is a good option. But why did you merge with them? I had shared my concerns about them before you merged with them. You thought I wasn't right?
", "rawMarkdown": "rsinda I am sorry that you are removed and unmerge with submission freeze is a good option. But why did you merge with them? I had shared my concerns about them before you merged with them. You thought I wasn't right?", "votes": null }, { "id": "1210079", "postDate": "02/19/2021 07:39:35", "content": "@rsinda I once was removed after winning a gold medal. Because I teamed with someone who used 3 accounts and he didn't tell me.
\nTeaming with people we don't know bears a risk, and now I avoid it, or I probe the behavior of tentative team mates before teaming.
", "rawMarkdown": "rsinda I once was removed after winning a gold medal. Because I teamed with someone who used 3 accounts and he didn't tell me.\n\nTeaming with people we don't know bears a risk, and now I avoid it, or I probe the behavior of tentative team mates before teaming.", "votes": null }, { "id": "1210226", "postDate": "02/19/2021 09:23:48", "content": "@aerdem4 I didn't read your discussion forum before teaming up. So I thought its not going to result in such a bad incident. This is the first time I teamed up in my kaggle journey.
\nI think kaggle should add unmerge option now.
@cpmpml well lesson learned although hard-way.
\nI think even my team mates understand this. It will make rules clear to everyone now.
", "rawMarkdown": "aerdem4 I didn't read your discussion forum before teaming up. So I thought its not going to result in such a bad incident. This is the first time I teamed up in my kaggle journey. \nI think kaggle should add unmerge option now.\n\n@cpmpml well lesson learned although hard-way. \n\nI think even my team mates understand this. It will make rules clear to everyone now.", "votes": null }, { "id": "1210270", "postDate": "02/19/2021 10:01:10", "content": "Really good call, otherwise this would set a pretty bad precendent and people would abuse further in the future, as the saying goes \"the road to hell is paved with good intentions\".
", "rawMarkdown": "Really good call, otherwise this would set a pretty bad precendent and people would abuse further in the future, as the saying goes \"the road to hell is paved with good intentions\".", "votes": null }, { "id": "1210576", "postDate": "02/19/2021 14:28:29", "content": "Issue is that there were plenty of precedents. But this should change now that Kaggle is taking action.
", "rawMarkdown": "Issue is that there were plenty of precedents. But this should change now that Kaggle is taking action.", "votes": null }, { "id": "1211280", "postDate": "02/20/2021 05:47:26", "content": "This was my first kaggle competition in a while. I even built my own desktop, hoping to learn a lot from it. Well I did, not about data science, but about what kaggle justice is, how it is defended by a bunch of people, and how unfriendly it is to new kagglers.
\nI know many of you who are reading this are probably more senior kagglers than I am, or my teammates are. Just for a moment, please imagine you are new to kaggle, and this is your first competition. Now, think about the following two questions.
\nIs the private sharing rule clear enough?
\nIt's there on line 3-5 of the Rules page:
\n\n\nNo private sharing outside teams
\n
\n Privately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.
I did read this when joining the competition. I totally agree with it. It's unfair for competitors if some people take multiple front places using basically the same model. However, does this make it clear that it's NOT OK to work as a team in separate accounts and only merge before competition ends? No, not to me. And it seems quite natural for people to collaborate in this flexible way. I mean after all who wants to be prohibited from testing a model just because your team mate uses up all submissions. And I have worked this way in a previous competition. Yes, please go and remove me from that too. But why bother, since I wasn't in the medal zone. Still, kind of double standard isn't it? I should be removed this time just because our performance is better? Doesn't make sense to me. Actually, if I had been removed from the previous competition, I guarantee you that we would have merged at the very beginning this time.
\nSome fellow kagglers might quote section 8.A in Rules:
\n\n\n\n
\n- SUBMISSION CODE REQUIREMENTS.
\n
\n A. Private Code Sharing. Unless otherwise specifically permitted under the Competition Website or Competition Specific Rules above, during the Competition Period, you are not allowed to privately share source or executable code developed in connection with or based upon the Competition Data or other source or executable code relevant to the Competition
\n (“Competition Code”). This prohibition includes sharing Competition Code between separate Teams, unless a Team merger occurs. Any such sharing of Competition Code is a breach of these Competition Rules and may result in disqualification.
Wow, crystal clear isn't it? When I first saw Ahmet's post about us private sharing, I immediately went to the Rules, and found this section. I was thinking \"we are doomed. It's said very clear here\". But where is this section? Line 148 (yes, I counted) of the Rules page! Come on, don't tell me that most people read the Rules word by word to the bottom. People bashed us with this, accusing us of abusing the private sharing rule. I felt very much like being told by tech giants to refer to User Agreement, while they abuse our private data. If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!
\nIs the punishment proportional to the severity of the breach?
\nWe believed that the purpose of no private sharing rule is to prevent people taking multple spots in the final leader board, which we never intended to do. After realzing people's concerns about us, we decided to merge right away, making it clear that we were a team with no intention to steal more spots. Altough, by section 8.A we breached the (not so clearly stated) rules, we believed that kaggle is a place for sharing and learning, rather than a court; we believed that fellow kagglers would be open-minded and accept the possibility that this could be an oversight from a bunch of young kagglers. But no, we overestimated the kindness in people. Good lesson learned.
@aerdem4, if you can have more patience and wait until the end of competition to post you investigation, you would find us in LB as one team, and other teams will still be in their \"deserved places\". What's the point of posting it when the competition is in progress? Admittedly, you might be right about some other advantages of not merging, like more submission limits towards the end of the competition, etc. But, are such advantages big enough to disqualify us and let our hours of effort go down the drain? To me, stealing spots in the end has a qualitative difference form the advantages above.
\nComments here say that justice is done, and this is a \"very positive change in Kaggle\". I'm surprised that no one points to the real issue. It means nothing if whatever happened in this competition stays in this competition. It would be a real change only if what private sharing is is clearly stated in the next competition, not just punishing the next breach, which could be from another poor young kaggle team like us.
\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice. I hope in the future you can make kaggle a place with more open-mind, understanding, and generosity, not a place like a court.
", "rawMarkdown": "This was my first kaggle competition in a while. I even built my own desktop, hoping to learn a lot from it. Well I did, not about data science, but about what kaggle justice is, how it is defended by a bunch of people, and how unfriendly it is to new kagglers. \n\nI know many of you who are reading this are probably more senior kagglers than I am, or my teammates are. Just for a moment, please imagine you are new to kaggle, and this is your first competition. Now, think about the following two questions.\n\n**Is the private sharing rule clear enough?**\n\nIt's there on line 3-5 of the Rules page:\n\n> No private sharing outside teams \nPrivately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.\n\nI did read this when joining the competition. I totally agree with it. It's unfair for competitors if some people take multiple front places using basically the same model. However, does this make it clear that it's NOT OK to work as a team in separate accounts and only merge before competition ends? No, not to me. And it seems quite natural for people to collaborate in this flexible way. I mean after all who wants to be prohibited from testing a model just because your team mate uses up all submissions. And I have worked this way in a previous competition. Yes, please go and remove me from that too. But why bother, since I wasn't in the medal zone. Still, kind of double standard isn't it? I should be removed this time just because our performance is better? Doesn't make sense to me. Actually, if I had been removed from the previous competition, I guarantee you that we would have merged at the very beginning this time. \n\nSome fellow kagglers might quote section 8.A in Rules:\n> 8. SUBMISSION CODE REQUIREMENTS.\nA. Private Code Sharing. Unless otherwise specifically permitted under the Competition Website or Competition Specific Rules above, during the Competition Period, you are not allowed to privately share source or executable code developed in connection with or based upon the Competition Data or other source or executable code relevant to the Competition \n(“Competition Code”). This prohibition includes sharing Competition Code between separate Teams, unless a Team merger occurs. Any such sharing of Competition Code is a breach of these Competition Rules and may result in disqualification.\n\nWow, crystal clear isn't it? When I first saw Ahmet's post about us private sharing, I immediately went to the Rules, and found this section. I was thinking \"we are doomed. It's said very clear here\". But where is this section? **Line 148** (yes, I counted) of the Rules page! Come on, don't tell me that most people read the Rules word by word to the bottom. People bashed us with this, accusing us of abusing the private sharing rule. I felt very much like being told by tech giants to refer to User Agreement, while they abuse our private data. **If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!**\n\n**Is the punishment proportional to the severity of the breach?**\nWe believed that the purpose of no private sharing rule is to prevent people taking multple spots in the final leader board, which we never intended to do. After realzing people's concerns about us, we decided to merge right away, making it clear that we were a team with no intention to steal more spots. Altough, by section 8.A we breached the (not so clearly stated) rules, we believed that kaggle is a place for sharing and learning, rather than a court; we believed that fellow kagglers would be open-minded and accept the possibility that this could be an oversight from a bunch of young kagglers. But no, we overestimated the kindness in people. Good lesson learned. \n\n@aerdem4, if you can have more patience and wait until the end of competition to post you investigation, you would find us in LB as one team, and other teams will still be in their \"deserved places\". What's the point of posting it when the competition is in progress? Admittedly, you might be right about some other advantages of not merging, like more submission limits towards the end of the competition, etc. But, are such advantages big enough to disqualify us and let our hours of effort go down the drain? To me, stealing spots in the end has a qualitative difference form the advantages above. \n\nComments here say that justice is done, and this is a \"very positive change in Kaggle\". I'm surprised that no one points to the real issue. It means nothing if whatever happened in this competition stays in this competition. **It would be a real change only if what private sharing is is clearly stated in the next competition, not just punishing the next breach**, which could be from another poor young kaggle team like us. \n\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice. I hope in the future you can make kaggle a place with more open-mind, understanding, and generosity, not a place like a court.", "votes": null }, { "id": "1211925", "postDate": "02/20/2021 17:09:01", "content": "Well whats really sad is that kaggle staff didn't even care to reply to our comments & explain things to us. Even though I constantly tried reaching to them. But they just don't care. @juliaelliott @inversion
", "rawMarkdown": "Well whats really sad is that kaggle staff didn't even care to reply to our comments & explain things to us. Even though I constantly tried reaching to them. **But they just don't care**. @juliaelliott @inversion", "votes": null }, { "id": "1211969", "postDate": "02/20/2021 17:57:26", "content": "My take is that I think kaggle was also trying to protect other competitors in the sense that if they said certain teams would be disqualified before it ended, there is a high chance they would just share all the secrets and will either make the other competitors biased to your implementation or just plainly ruin it for the other competitors by sharing.
", "rawMarkdown": "My take is that I think kaggle was also trying to protect other competitors in the sense that if they said certain teams would be disqualified before it ended, there is a high chance they would just share all the secrets and will either make the other competitors biased to your implementation or just plainly ruin it for the other competitors by sharing.", "votes": null }, { "id": "1211976", "postDate": "02/20/2021 18:02:54", "content": "\n\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice.
\n
What version? Defended how and where?
", "rawMarkdown": "> Respect to @cpmpml, @aerdem4. You have defended your version of kaggle justice.\n\nWhat version? Defended how and where?", "votes": null }, { "id": "1211993", "postDate": "02/20/2021 18:27:31", "content": "\n\nCome on, don't tell me that most people read the Rules word by word to the bottom.
\n
I did. It is the only sensible thing to do, when you are agreeing with something that regulates potentially substantial money transfer. Same for work contracts, buying a house/car, etc etc etc.
\n\n\nIf private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!
\n
I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback
", "rawMarkdown": "> Come on, don't tell me that most people read the Rules word by word to the bottom.\n\nI did. It is the only sensible thing to do, when you are agreeing with something that regulates potentially substantial money transfer. Same for work contracts, buying a house/car, etc etc etc.\n\n> If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!\n\nI agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback", "votes": null }, { "id": "1212042", "postDate": "02/20/2021 20:07:09", "content": "@fffrrt I'm on same page as you, but most people don't read fine prints, even in their work contract. Silly maybe, but this is reality. Then they complain.
", "rawMarkdown": "fffrrt I'm on same page as you, but most people don't read fine prints, even in their work contract. Silly maybe, but this is reality. Then they complain.", "votes": null }, { "id": "1212147", "postDate": "02/20/2021 23:31:02", "content": "@fffrrt you are meticulous, that's great. But still, most people wouldn't really compare a kaggle challenge to buying a house or signing work contract. I'm just saying this is a rule that can be easily violated unintentionally. Many people ignore that when making comments, and speculated us to have really bad intentions.
\n\n\nI agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback
\n
We will provide our feedback for sure.
", "rawMarkdown": "fffrrt you are meticulous, that's great. But still, most people wouldn't really compare a kaggle challenge to buying a house or signing work contract. I'm just saying this is a rule that can be easily violated unintentionally. Many people ignore that when making comments, and speculated us to have really bad intentions.\n\n> I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\n\nWe will provide our feedback for sure.", "votes": null }, { "id": "1212715", "postDate": "02/21/2021 14:03:59", "content": "….\"I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\"…
\nSame can be said of the original complainant. It could have been done in the respective f/b forums provided by Kaggle. Of course, in this case he was proved 100% right (thanks to him for ensuring a fair play) but this can set a wrong trend going ahead. a bunch of folks from a same location or company doing well in a comp and then there is a vigilante crowd raising allegations…not exactly a pleasant environment for a comp right? what if in the next case it turned out to be mere coincidence? Wouldn't it be better to use the right forum to provide these kind of f/b and let Kaggle take a call?
\nand as for the other comment on fine print, I dont think 9 out of 10 have read all the hundreds of lines of the contract, other than skim thru the top few lines…. It definitely makes sense to have these rules stated unambiguously somewhere on top.
\nI will not speak on this topic again. However I did feel sorry for the couple of newcomers who had to face a slightly harsh welcome to Kaggle and felt I had to put across my point of view once. Not everyone has dozens of golds accumalated and not everyone may be as magnanimous in losing one.
", "rawMarkdown": "....*\"I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\"*...\n\nSame can be said of the original complainant. It could have been done in the respective f/b forums provided by Kaggle. Of course, in this case he was proved 100% right (thanks to him for ensuring a fair play) but this can set a wrong trend going ahead. a bunch of folks from a same location or company doing well in a comp and then there is a vigilante crowd raising allegations...not exactly a pleasant environment for a comp right? what if in the next case it turned out to be mere coincidence? Wouldn't it be better to use the right forum to provide these kind of f/b and let Kaggle take a call?\n\nand as for the other comment on fine print, I dont think 9 out of 10 have read all the hundreds of lines of the contract, other than skim thru the top few lines.... It definitely makes sense to have these rules stated unambiguously somewhere on top.\n\nI will not speak on this topic again. However I did feel sorry for the couple of newcomers who had to face a slightly harsh welcome to Kaggle and felt I had to put across my point of view once. Not everyone has dozens of golds accumalated and not everyone may be as magnanimous in losing one.", "votes": null } ], "comments": [ { "id": 1209439, "author_name": "theoviel", "author_url": "", "post_date": "02/18/2021 22:36:53", "content": "Thanks to @aerdem4 as well :)
\nI believe that without his threads both teams would've gone unpunished, as some other teams most likely did.
Indeed. I thanked him privately but I should have mentioned him in the post. Fixed.
", "votes": null, "replies": [] } ] }, { "id": 1209469, "author_name": "duykhanh99", "author_url": "", "post_date": "02/18/2021 23:13:04", "content": "Thank you kaggle team for doing the right rules and showing fairness to all teams!
", "votes": null, "replies": [] }, { "id": 1209510, "author_name": "aerdem4", "author_url": "", "post_date": "02/18/2021 23:53:22", "content": "While I feel sad for removed teams since their efforts are wasted, I feel happy for the other teams who got their deserved places.
", "votes": null, "replies": [] }, { "id": 1209823, "author_name": "mcggood", "author_url": "", "post_date": "02/19/2021 03:58:53", "content": "Lucky we have kagglers like Ahmet Erdem.
", "votes": null, "replies": [] }, { "id": 1209832, "author_name": "rsinda", "author_url": "", "post_date": "02/19/2021 04:05:51", "content": "Thanks for removing me from competition just because I merged with another team. Making all my solo efforts go waste
", "votes": null, "replies": [ { "id": 1209870, "author_name": "dicksonchin93", "author_url": "", "post_date": "02/19/2021 04:43:35", "content": "I guess becareful who you merge with next time? was exciting seeing your team approach a higher lb score and was fun mentally battling it out with your team guessing what your team was doing, hope you come back up stronger and safer next time
", "votes": null, "replies": [] }, { "id": 1209892, "author_name": "rsinda", "author_url": "", "post_date": "02/19/2021 04:57:22", "content": "Well how do one supposed to know whom to merge with, I guess not all are great detective like @aerdem4. This raises the concern of getting an option for unmerge. Maybe after unmerging you are not allowed to make more submissions, but one will not loose his efforts from the beginning.
", "votes": null, "replies": [] }, { "id": 1210058, "author_name": "allohvk", "author_url": "", "post_date": "02/19/2021 07:18:27", "content": "bit unfortunate, I agree… a tremendous amount of effort would have gone in… My sympathies.
", "votes": null, "replies": [] }, { "id": 1210072, "author_name": "aerdem4", "author_url": "", "post_date": "02/19/2021 07:34:36", "content": "@rsinda I am sorry that you are removed and unmerge with submission freeze is a good option. But why did you merge with them? I had shared my concerns about them before you merged with them. You thought I wasn't right?
", "votes": null, "replies": [] }, { "id": 1210079, "author_name": "cpmpml", "author_url": "", "post_date": "02/19/2021 07:39:35", "content": "@rsinda I once was removed after winning a gold medal. Because I teamed with someone who used 3 accounts and he didn't tell me.
\nTeaming with people we don't know bears a risk, and now I avoid it, or I probe the behavior of tentative team mates before teaming.
", "votes": null, "replies": [] }, { "id": 1210226, "author_name": "rsinda", "author_url": "", "post_date": "02/19/2021 09:23:48", "content": "@aerdem4 I didn't read your discussion forum before teaming up. So I thought its not going to result in such a bad incident. This is the first time I teamed up in my kaggle journey.
\nI think kaggle should add unmerge option now.
@cpmpml well lesson learned although hard-way.
\nI think even my team mates understand this. It will make rules clear to everyone now.
", "votes": null, "replies": [] } ] }, { "id": 1210270, "author_name": "antorsae", "author_url": "", "post_date": "02/19/2021 10:01:10", "content": "Really good call, otherwise this would set a pretty bad precendent and people would abuse further in the future, as the saying goes \"the road to hell is paved with good intentions\".
", "votes": null, "replies": [ { "id": 1210576, "author_name": "cpmpml", "author_url": "", "post_date": "02/19/2021 14:28:29", "content": "Issue is that there were plenty of precedents. But this should change now that Kaggle is taking action.
", "votes": null, "replies": [] } ] }, { "id": 1211280, "author_name": "zenglix", "author_url": "", "post_date": "02/20/2021 05:47:26", "content": "This was my first kaggle competition in a while. I even built my own desktop, hoping to learn a lot from it. Well I did, not about data science, but about what kaggle justice is, how it is defended by a bunch of people, and how unfriendly it is to new kagglers.
\nI know many of you who are reading this are probably more senior kagglers than I am, or my teammates are. Just for a moment, please imagine you are new to kaggle, and this is your first competition. Now, think about the following two questions.
\nIs the private sharing rule clear enough?
\nIt's there on line 3-5 of the Rules page:
\n\n\nNo private sharing outside teams
\n
\n Privately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.
I did read this when joining the competition. I totally agree with it. It's unfair for competitors if some people take multiple front places using basically the same model. However, does this make it clear that it's NOT OK to work as a team in separate accounts and only merge before competition ends? No, not to me. And it seems quite natural for people to collaborate in this flexible way. I mean after all who wants to be prohibited from testing a model just because your team mate uses up all submissions. And I have worked this way in a previous competition. Yes, please go and remove me from that too. But why bother, since I wasn't in the medal zone. Still, kind of double standard isn't it? I should be removed this time just because our performance is better? Doesn't make sense to me. Actually, if I had been removed from the previous competition, I guarantee you that we would have merged at the very beginning this time.
\nSome fellow kagglers might quote section 8.A in Rules:
\n\n\n\n
\n- SUBMISSION CODE REQUIREMENTS.
\n
\n A. Private Code Sharing. Unless otherwise specifically permitted under the Competition Website or Competition Specific Rules above, during the Competition Period, you are not allowed to privately share source or executable code developed in connection with or based upon the Competition Data or other source or executable code relevant to the Competition
\n (“Competition Code”). This prohibition includes sharing Competition Code between separate Teams, unless a Team merger occurs. Any such sharing of Competition Code is a breach of these Competition Rules and may result in disqualification.
Wow, crystal clear isn't it? When I first saw Ahmet's post about us private sharing, I immediately went to the Rules, and found this section. I was thinking \"we are doomed. It's said very clear here\". But where is this section? Line 148 (yes, I counted) of the Rules page! Come on, don't tell me that most people read the Rules word by word to the bottom. People bashed us with this, accusing us of abusing the private sharing rule. I felt very much like being told by tech giants to refer to User Agreement, while they abuse our private data. If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!
\nIs the punishment proportional to the severity of the breach?
\nWe believed that the purpose of no private sharing rule is to prevent people taking multple spots in the final leader board, which we never intended to do. After realzing people's concerns about us, we decided to merge right away, making it clear that we were a team with no intention to steal more spots. Altough, by section 8.A we breached the (not so clearly stated) rules, we believed that kaggle is a place for sharing and learning, rather than a court; we believed that fellow kagglers would be open-minded and accept the possibility that this could be an oversight from a bunch of young kagglers. But no, we overestimated the kindness in people. Good lesson learned.
@aerdem4, if you can have more patience and wait until the end of competition to post you investigation, you would find us in LB as one team, and other teams will still be in their \"deserved places\". What's the point of posting it when the competition is in progress? Admittedly, you might be right about some other advantages of not merging, like more submission limits towards the end of the competition, etc. But, are such advantages big enough to disqualify us and let our hours of effort go down the drain? To me, stealing spots in the end has a qualitative difference form the advantages above.
\nComments here say that justice is done, and this is a \"very positive change in Kaggle\". I'm surprised that no one points to the real issue. It means nothing if whatever happened in this competition stays in this competition. It would be a real change only if what private sharing is is clearly stated in the next competition, not just punishing the next breach, which could be from another poor young kaggle team like us.
\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice. I hope in the future you can make kaggle a place with more open-mind, understanding, and generosity, not a place like a court.
", "votes": null, "replies": [ { "id": 1211925, "author_name": "rsinda", "author_url": "", "post_date": "02/20/2021 17:09:01", "content": "Well whats really sad is that kaggle staff didn't even care to reply to our comments & explain things to us. Even though I constantly tried reaching to them. But they just don't care. @juliaelliott @inversion
", "votes": null, "replies": [] }, { "id": 1211969, "author_name": "dicksonchin93", "author_url": "", "post_date": "02/20/2021 17:57:26", "content": "My take is that I think kaggle was also trying to protect other competitors in the sense that if they said certain teams would be disqualified before it ended, there is a high chance they would just share all the secrets and will either make the other competitors biased to your implementation or just plainly ruin it for the other competitors by sharing.
", "votes": null, "replies": [] }, { "id": 1211976, "author_name": "cpmpml", "author_url": "", "post_date": "02/20/2021 18:02:54", "content": "\n\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice.
\n
What version? Defended how and where?
", "votes": null, "replies": [] }, { "id": 1211993, "author_name": "fffrrt", "author_url": "", "post_date": "02/20/2021 18:27:31", "content": "\n\nCome on, don't tell me that most people read the Rules word by word to the bottom.
\n
I did. It is the only sensible thing to do, when you are agreeing with something that regulates potentially substantial money transfer. Same for work contracts, buying a house/car, etc etc etc.
\n\n\nIf private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!
\n
I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback
", "votes": null, "replies": [] }, { "id": 1212042, "author_name": "cpmpml", "author_url": "", "post_date": "02/20/2021 20:07:09", "content": "@fffrrt I'm on same page as you, but most people don't read fine prints, even in their work contract. Silly maybe, but this is reality. Then they complain.
", "votes": null, "replies": [] }, { "id": 1212147, "author_name": "zenglix", "author_url": "", "post_date": "02/20/2021 23:31:02", "content": "@fffrrt you are meticulous, that's great. But still, most people wouldn't really compare a kaggle challenge to buying a house or signing work contract. I'm just saying this is a rule that can be easily violated unintentionally. Many people ignore that when making comments, and speculated us to have really bad intentions.
\n\n\nI agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback
\n
We will provide our feedback for sure.
", "votes": null, "replies": [] }, { "id": 1212715, "author_name": "allohvk", "author_url": "", "post_date": "02/21/2021 14:03:59", "content": "….\"I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\"…
\nSame can be said of the original complainant. It could have been done in the respective f/b forums provided by Kaggle. Of course, in this case he was proved 100% right (thanks to him for ensuring a fair play) but this can set a wrong trend going ahead. a bunch of folks from a same location or company doing well in a comp and then there is a vigilante crowd raising allegations…not exactly a pleasant environment for a comp right? what if in the next case it turned out to be mere coincidence? Wouldn't it be better to use the right forum to provide these kind of f/b and let Kaggle take a call?
\nand as for the other comment on fine print, I dont think 9 out of 10 have read all the hundreds of lines of the contract, other than skim thru the top few lines…. It definitely makes sense to have these rules stated unambiguously somewhere on top.
\nI will not speak on this topic again. However I did feel sorry for the couple of newcomers who had to face a slightly harsh welcome to Kaggle and felt I had to put across my point of view once. Not everyone has dozens of golds accumalated and not everyone may be as magnanimous in losing one.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1209437": "juliaelliott We all noticed that you removed the two teams that admitted sharing before teaming. Until this competition this has been a gray area, with some abusing it. You made it clear this was against the rules ( as it has always been) and you acted accordingly.\n\nThis is a very positive change in Kaggle IMHO.\n\nThanks for acting.\n\nEdit. Theo Viel reminds me of Ahmet Erdem key role here as well. Without his posts I am not sure we would have got this great move.", "1209439": "Thanks to @aerdem4 as well :)\nI believe that without his threads both teams would've gone unpunished, as some other teams most likely did.", "1209448": "Indeed. I thanked him privately but I should have mentioned him in the post. Fixed.", "1209469": "Thank you kaggle team for doing the right rules and showing fairness to all teams!", "1209510": "While I feel sad for removed teams since their efforts are wasted, I feel happy for the other teams who got their deserved places.", "1209823": "Lucky we have kagglers like Ahmet Erdem.", "1209832": "Thanks for removing me from competition just because I merged with another team. Making all my solo efforts go waste", "1209870": "I guess becareful who you merge with next time? was exciting seeing your team approach a higher lb score and was fun mentally battling it out with your team guessing what your team was doing, hope you come back up stronger and safer next time", "1209892": "Well how do one supposed to know whom to merge with, I guess not all are great detective like @aerdem4. This raises the concern of getting an option for unmerge. Maybe after unmerging you are not allowed to make more submissions, but one will not loose his efforts from the beginning.", "1210058": "bit unfortunate, I agree... a tremendous amount of effort would have gone in... My sympathies.", "1210072": "rsinda I am sorry that you are removed and unmerge with submission freeze is a good option. But why did you merge with them? I had shared my concerns about them before you merged with them. You thought I wasn't right?", "1210079": "rsinda I once was removed after winning a gold medal. Because I teamed with someone who used 3 accounts and he didn't tell me.\n\nTeaming with people we don't know bears a risk, and now I avoid it, or I probe the behavior of tentative team mates before teaming.", "1210226": "aerdem4 I didn't read your discussion forum before teaming up. So I thought its not going to result in such a bad incident. This is the first time I teamed up in my kaggle journey. \nI think kaggle should add unmerge option now.\n\n@cpmpml well lesson learned although hard-way. \n\nI think even my team mates understand this. It will make rules clear to everyone now.", "1210270": "Really good call, otherwise this would set a pretty bad precendent and people would abuse further in the future, as the saying goes \"the road to hell is paved with good intentions\".", "1210576": "Issue is that there were plenty of precedents. But this should change now that Kaggle is taking action.", "1211280": "This was my first kaggle competition in a while. I even built my own desktop, hoping to learn a lot from it. Well I did, not about data science, but about what kaggle justice is, how it is defended by a bunch of people, and how unfriendly it is to new kagglers. \n\nI know many of you who are reading this are probably more senior kagglers than I am, or my teammates are. Just for a moment, please imagine you are new to kaggle, and this is your first competition. Now, think about the following two questions.\n\n**Is the private sharing rule clear enough?**\n\nIt's there on line 3-5 of the Rules page:\n\n> No private sharing outside teams \nPrivately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.\n\nI did read this when joining the competition. I totally agree with it. It's unfair for competitors if some people take multiple front places using basically the same model. However, does this make it clear that it's NOT OK to work as a team in separate accounts and only merge before competition ends? No, not to me. And it seems quite natural for people to collaborate in this flexible way. I mean after all who wants to be prohibited from testing a model just because your team mate uses up all submissions. And I have worked this way in a previous competition. Yes, please go and remove me from that too. But why bother, since I wasn't in the medal zone. Still, kind of double standard isn't it? I should be removed this time just because our performance is better? Doesn't make sense to me. Actually, if I had been removed from the previous competition, I guarantee you that we would have merged at the very beginning this time. \n\nSome fellow kagglers might quote section 8.A in Rules:\n> 8. SUBMISSION CODE REQUIREMENTS.\nA. Private Code Sharing. Unless otherwise specifically permitted under the Competition Website or Competition Specific Rules above, during the Competition Period, you are not allowed to privately share source or executable code developed in connection with or based upon the Competition Data or other source or executable code relevant to the Competition \n(“Competition Code”). This prohibition includes sharing Competition Code between separate Teams, unless a Team merger occurs. Any such sharing of Competition Code is a breach of these Competition Rules and may result in disqualification.\n\nWow, crystal clear isn't it? When I first saw Ahmet's post about us private sharing, I immediately went to the Rules, and found this section. I was thinking \"we are doomed. It's said very clear here\". But where is this section? **Line 148** (yes, I counted) of the Rules page! Come on, don't tell me that most people read the Rules word by word to the bottom. People bashed us with this, accusing us of abusing the private sharing rule. I felt very much like being told by tech giants to refer to User Agreement, while they abuse our private data. **If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!**\n\n**Is the punishment proportional to the severity of the breach?**\nWe believed that the purpose of no private sharing rule is to prevent people taking multple spots in the final leader board, which we never intended to do. After realzing people's concerns about us, we decided to merge right away, making it clear that we were a team with no intention to steal more spots. Altough, by section 8.A we breached the (not so clearly stated) rules, we believed that kaggle is a place for sharing and learning, rather than a court; we believed that fellow kagglers would be open-minded and accept the possibility that this could be an oversight from a bunch of young kagglers. But no, we overestimated the kindness in people. Good lesson learned. \n\n@aerdem4, if you can have more patience and wait until the end of competition to post you investigation, you would find us in LB as one team, and other teams will still be in their \"deserved places\". What's the point of posting it when the competition is in progress? Admittedly, you might be right about some other advantages of not merging, like more submission limits towards the end of the competition, etc. But, are such advantages big enough to disqualify us and let our hours of effort go down the drain? To me, stealing spots in the end has a qualitative difference form the advantages above. \n\nComments here say that justice is done, and this is a \"very positive change in Kaggle\". I'm surprised that no one points to the real issue. It means nothing if whatever happened in this competition stays in this competition. **It would be a real change only if what private sharing is is clearly stated in the next competition, not just punishing the next breach**, which could be from another poor young kaggle team like us. \n\nRespect to @cpmpml, @aerdem4. You have defended your version of kaggle justice. I hope in the future you can make kaggle a place with more open-mind, understanding, and generosity, not a place like a court.", "1211925": "Well whats really sad is that kaggle staff didn't even care to reply to our comments & explain things to us. Even though I constantly tried reaching to them. **But they just don't care**. @juliaelliott @inversion", "1211969": "My take is that I think kaggle was also trying to protect other competitors in the sense that if they said certain teams would be disqualified before it ended, there is a high chance they would just share all the secrets and will either make the other competitors biased to your implementation or just plainly ruin it for the other competitors by sharing.", "1211976": "> Respect to @cpmpml, @aerdem4. You have defended your version of kaggle justice.\n\nWhat version? Defended how and where?", "1211993": "> Come on, don't tell me that most people read the Rules word by word to the bottom.\n\nI did. It is the only sensible thing to do, when you are agreeing with something that regulates potentially substantial money transfer. Same for work contracts, buying a house/car, etc etc etc.\n\n> If private sharing is such an issue with many precedences, why not making it clear in top rows in bold, rather than hiding it in line 148+!\n\nI agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback", "1212042": "fffrrt I'm on same page as you, but most people don't read fine prints, even in their work contract. Silly maybe, but this is reality. Then they complain.", "1212147": "fffrrt you are meticulous, that's great. But still, most people wouldn't really compare a kaggle challenge to buying a house or signing work contract. I'm just saying this is a rule that can be easily violated unintentionally. Many people ignore that when making comments, and speculated us to have really bad intentions.\n\n> I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\n\nWe will provide our feedback for sure.", "1212715": "....*\"I agree, but it is probably better to post your suggestions to https://www.kaggle.com/product-feedback\"*...\n\nSame can be said of the original complainant. It could have been done in the respective f/b forums provided by Kaggle. Of course, in this case he was proved 100% right (thanks to him for ensuring a fair play) but this can set a wrong trend going ahead. a bunch of folks from a same location or company doing well in a comp and then there is a vigilante crowd raising allegations...not exactly a pleasant environment for a comp right? what if in the next case it turned out to be mere coincidence? Wouldn't it be better to use the right forum to provide these kind of f/b and let Kaggle take a call?\n\nand as for the other comment on fine print, I dont think 9 out of 10 have read all the hundreds of lines of the contract, other than skim thru the top few lines.... It definitely makes sense to have these rules stated unambiguously somewhere on top.\n\nI will not speak on this topic again. However I did feel sorry for the couple of newcomers who had to face a slightly harsh welcome to Kaggle and felt I had to put across my point of view once. Not everyone has dozens of golds accumalated and not everyone may be as magnanimous in losing one." }, "source": "meta" }