{ "id": 382440, "title": "Should we be careful about sharing our solutions after the contest?", "url": "/competitions/otto-recommender-system/discussion/382440", "author_name": "KeigoOka", "post_date": "2023-01-31T01:33:10.734000", "votes": 7, "comment_count": 12, "views": 0, "content": "

Under the current circumstance that there could be cheaters seeing our discussion, should we be careful about sharing our approach or source code publicly after the contest is over?

\n

I think contestants who get medals would be extremely excited after the contest and tempted to share their solutions, but I'm concerned if such an act would benefit cheaters.

\n

I hope some kind of guidelines are provided before the contest ends!

", "messages": [ { "id": 2122498, "postDate": "2023-01-31T01:33:10.733Z", "content": "

Under the current circumstance that there could be cheaters seeing our discussion, should we be careful about sharing our approach or source code publicly after the contest is over?

\n

I think contestants who get medals would be extremely excited after the contest and tempted to share their solutions, but I'm concerned if such an act would benefit cheaters.

\n

I hope some kind of guidelines are provided before the contest ends!

", "rawMarkdown": "Under the current circumstance that there could be cheaters seeing our discussion, should we be careful about sharing our approach or source code publicly after the contest is over?\n\nI think contestants who get medals would be extremely excited after the contest and tempted to share their solutions, but I'm concerned if such an act would benefit cheaters.\n\nI hope some kind of guidelines are provided *before* the contest ends!", "votes": 7 }, { "id": 2127697, "postDate": "2023-02-03T04:33:08.683Z", "content": "

funny. the author was removed from lb (he was at gold zone) 👀

", "rawMarkdown": "funny. the author was removed from lb (he was at gold zone) 👀", "votes": 2, "replies": [ { "id": 2127709, "postDate": "2023-02-03T04:53:14.027Z", "content": "

😂 good catch.

", "rawMarkdown": "😂 good catch." }, { "id": 2127710, "postDate": "2023-02-03T04:53:26.403Z", "content": "

Yeah, that's weird.

", "rawMarkdown": "Yeah, that's weird." } ] }, { "id": 2122547, "postDate": "2023-01-31T02:29:32.620Z", "content": "
\n

but I'm concerned if such an act would benefit cheaters.

\n
\n

What kind of concerns do you have?

\n

Do they copy notebooks/solutions after competition ended, and claims their(cheater's) final submissions are created by this notebooks/solutions? If they do, I think that won't be any profit of cheaters because Kaggle's staffs can know they are similar/the same as the already shared solutions.

\n

However, I agree with we should be care about any possible risks.

\n

Are there any use-case for cheaters where they can exploit solution/notebooks shared after competitions?

", "rawMarkdown": "> but I'm concerned if such an act would benefit cheaters.\n\nWhat kind of concerns do you have?\n\nDo they copy notebooks/solutions after competition ended, and claims their(cheater's) final submissions are created by this notebooks/solutions? If they do, I think that won't be any profit of cheaters because Kaggle's staffs can know they are similar/the same as the already shared solutions.\n\nHowever, I agree with we should be care about any possible risks.\n\nAre there any use-case for cheaters where they can exploit solution/notebooks shared after competitions?", "votes": 2, "replies": [ { "id": 2122583, "postDate": "2023-01-31T03:09:40.303Z", "content": "

I don't know about any past use-cases (I'm a new joiner), and I might be worring too much, but if cheaters are pro, they could rewrite the source code preserving the core ideas, and claim that they wrote it. It would be easier than coming up with original ideas and writing everything from scratch.

", "rawMarkdown": "I don't know about any past use-cases (I'm a new joiner), and I might be worring too much, but if cheaters are pro, they could rewrite the source code preserving the core ideas, and claim that they wrote it. It would be easier than coming up with original ideas and writing everything from scratch.", "votes": 1, "replies": [ { "id": 2122613, "postDate": "2023-01-31T03:52:58.280Z", "content": "
\n

they could rewrite the source code preserving the core ideas, and claim that they wrote it

\n
\n

Well, they might do if the fraud check only checks if the code reproduces the final private LB scores.

\n

If the fraud check examines if the code exactly reproduces the final submitted file, that is difficult by copying/editing shared code. However, it surely depends on how Kaggle's fraud check process are taken.

", "rawMarkdown": "> they could rewrite the source code preserving the core ideas, and claim that they wrote it\n\nWell, they might do if the fraud check only checks if the code reproduces the final private LB scores.\n\nIf the fraud check examines if the code exactly reproduces the final submitted file, that is difficult by copying/editing shared code. However, it surely depends on how Kaggle's fraud check process are taken." }, { "id": 2122890, "postDate": "2023-01-31T07:35:05.140Z", "content": "
\n

If the fraud check examines if the code exactly reproduces the final submitted file

\n
\n

Just to note that reproducing the exact same file is a high bar even for innocent contestants, because of randomness.

", "rawMarkdown": ">If the fraud check examines if the code exactly reproduces the final submitted file\n\nJust to note that reproducing the exact same file is a high bar even for innocent contestants, because of randomness.", "votes": 1 }, { "id": 2123047, "postDate": "2023-01-31T09:59:45.697Z", "content": "

Yeah, I agree with this.

", "rawMarkdown": "Yeah, I agree with this." }, { "id": 2123055, "postDate": "2023-01-31T10:08:26.313Z", "content": "

@pnormann @wcukierski Can you look through this thread?

", "rawMarkdown": "@pnormann @wcukierski Can you look through this thread?" }, { "id": 2123460, "postDate": "2023-01-31T14:30:18.420Z", "content": "

Confirming there is no need to withhold sharing code/solutions after the deadline. The information we use to determine rules compliance is already collected by then, and further appeals/decisions are not made based on ability to produce realistic-looking source code.

", "rawMarkdown": "Confirming there is no need to withhold sharing code/solutions after the deadline. The information we use to determine rules compliance is already collected by then, and further appeals/decisions are not made based on ability to produce realistic-looking source code.", "votes": 6 }, { "id": 2123550, "postDate": "2023-01-31T15:20:42.140Z", "content": "

@wcukierski I see. Thank you for confirmation.

", "rawMarkdown": "@wcukierski I see. Thank you for confirmation." } ] } ] }, { "id": 2123052, "postDate": "2023-01-31T10:06:59.947Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2127697, "author_name": "empty", "author_url": "", "post_date": "2023-02-03T04:33:08.683000", "content": "

funny. the author was removed from lb (he was at gold zone) 👀

", "votes": 2, "replies": [ { "id": 2127709, "author_name": "ADAM.", "author_url": "", "post_date": "2023-02-03T04:53:14.027000", "content": "

😂 good catch.

", "votes": 0, "replies": [] }, { "id": 2127710, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2023-02-03T04:53:26.403000", "content": "

Yeah, that's weird.

", "votes": 0, "replies": [] } ] }, { "id": 2122547, "author_name": "Bilzard", "author_url": "", "post_date": "2023-01-31T02:29:32.620000", "content": "
\n

but I'm concerned if such an act would benefit cheaters.

\n
\n

What kind of concerns do you have?

\n

Do they copy notebooks/solutions after competition ended, and claims their(cheater's) final submissions are created by this notebooks/solutions? If they do, I think that won't be any profit of cheaters because Kaggle's staffs can know they are similar/the same as the already shared solutions.

\n

However, I agree with we should be care about any possible risks.

\n

Are there any use-case for cheaters where they can exploit solution/notebooks shared after competitions?

", "votes": 2, "replies": [ { "id": 2122583, "author_name": "KeigoOka", "author_url": "", "post_date": "2023-01-31T03:09:40.303000", "content": "

I don't know about any past use-cases (I'm a new joiner), and I might be worring too much, but if cheaters are pro, they could rewrite the source code preserving the core ideas, and claim that they wrote it. It would be easier than coming up with original ideas and writing everything from scratch.

", "votes": 1, "replies": [ { "id": 2122613, "author_name": "Bilzard", "author_url": "", "post_date": "2023-01-31T03:52:58.280000", "content": "
\n

they could rewrite the source code preserving the core ideas, and claim that they wrote it

\n
\n

Well, they might do if the fraud check only checks if the code reproduces the final private LB scores.

\n

If the fraud check examines if the code exactly reproduces the final submitted file, that is difficult by copying/editing shared code. However, it surely depends on how Kaggle's fraud check process are taken.

", "votes": 0, "replies": [] }, { "id": 2122890, "author_name": "KeigoOka", "author_url": "", "post_date": "2023-01-31T07:35:05.140000", "content": "
\n

If the fraud check examines if the code exactly reproduces the final submitted file

\n
\n

Just to note that reproducing the exact same file is a high bar even for innocent contestants, because of randomness.

", "votes": 1, "replies": [] }, { "id": 2123047, "author_name": "Bilzard", "author_url": "", "post_date": "2023-01-31T09:59:45.697000", "content": "

Yeah, I agree with this.

", "votes": 0, "replies": [] }, { "id": 2123055, "author_name": "Bilzard", "author_url": "", "post_date": "2023-01-31T10:08:26.313000", "content": "

@pnormann @wcukierski Can you look through this thread?

", "votes": 0, "replies": [] }, { "id": 2123460, "author_name": "Will Cukierski", "author_url": "", "post_date": "2023-01-31T14:30:18.420000", "content": "

Confirming there is no need to withhold sharing code/solutions after the deadline. The information we use to determine rules compliance is already collected by then, and further appeals/decisions are not made based on ability to produce realistic-looking source code.

", "votes": 6, "replies": [] }, { "id": 2123550, "author_name": "Bilzard", "author_url": "", "post_date": "2023-01-31T15:20:42.140000", "content": "

@wcukierski I see. Thank you for confirmation.

", "votes": 0, "replies": [] } ] } ] }, { "id": 2123052, "author_name": "", "author_url": "", "post_date": "2023-01-31T10:06:59.947000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2122498": "Under the current circumstance that there could be cheaters seeing our discussion, should we be careful about sharing our approach or source code publicly after the contest is over?\n\nI think contestants who get medals would be extremely excited after the contest and tempted to share their solutions, but I'm concerned if such an act would benefit cheaters.\n\nI hope some kind of guidelines are provided *before* the contest ends!", "2127697": "funny. the author was removed from lb (he was at gold zone) 👀", "2122547": "> but I'm concerned if such an act would benefit cheaters.\n\nWhat kind of concerns do you have?\n\nDo they copy notebooks/solutions after competition ended, and claims their(cheater's) final submissions are created by this notebooks/solutions? If they do, I think that won't be any profit of cheaters because Kaggle's staffs can know they are similar/the same as the already shared solutions.\n\nHowever, I agree with we should be care about any possible risks.\n\nAre there any use-case for cheaters where they can exploit solution/notebooks shared after competitions?", "2123052": "" } }