{ "id": 680686, "title": "Regarding the open-source issue that is malignant and offers no nutritional value.", "url": "/competitions/deep-past-initiative-machine-translation/discussion/680686", "author_name": "StarxSky", "post_date": "2026-03-10T06:22:46.451000", "votes": 45, "comment_count": 13, "views": 0, "content": "

Dear Kaggle Team and Competition Organizers,

\n

@deeppast @ryanholbrook

\n

I am writing to formally report a growing trend of malicious public sharing in this competition, which I believe violates the spirit of \"Fair Play\" and undermines the integrity of the Kaggle community.I have always regarded Kaggle as a fair and competitive data science competition platform with a great academic exchange environment, and I believe it will remain so in the future. However, the emergence of these problematic users has undermined the fairness of the competition. Their purpose in open-sourcing is merely for \"scores\" rather than for innovation. Therefore, I hope the competition organizers and Kaggle officials will take this seriously.

\n

The Issue

\n

In the final 14 days leading up to the deadline, some users have published public notebooks that are nothing more than a simple weighted average or rank-blending of existing high-scoring public notebooks. These \"no-mod\" ensemble scripts provide:\nHere notebook is one of them

\n\n

Why This is Malicious

\n

While Kaggle encourages sharing, the timing and nature of these specific shares suggest a clear intent to distort the final standings (often referred to as \"shaking up\" the leaderboard maliciously) rather than to foster learning. It creates a \"race to the bottom\" where the highest public score becomes a commodity rather than a reflection of skill.

\n

Proposed Action

\n

I kindly request the Kaggle team to:

\n
    \n

  1. Investigate the accounts frequently posting these high-scoring, zero-originality ensembles just before the deadline.

    2. \n

  2. Enforce stricter \"Public Sharing\" guidelines: Consider a \"Freeze Period\" for public notebooks (e.g., no new public notebooks allowed within 7 days of the deadline) to prevent last-minute leaderboard manipulation.

    3. \n

  3. Review the Impact: Evaluate if these specific notebooks in [Insert Competition Name] should be hidden or if the authors should be warned for unsportsmanlike conduct.

    4. \n
\n

We value Kaggle as a platform for elite data science competition. Allowing this behavior to continue unaddressed diminishes the prestige of Kaggle Medals and the platform's credibility.

\n

Sincerely,

\n

mochi

", "messages": [ { "id": 3419222, "postDate": "2026-03-10T06:22:46.450Z", "content": "

Dear Kaggle Team and Competition Organizers,

\n

@deeppast @ryanholbrook

\n

I am writing to formally report a growing trend of malicious public sharing in this competition, which I believe violates the spirit of \"Fair Play\" and undermines the integrity of the Kaggle community.I have always regarded Kaggle as a fair and competitive data science competition platform with a great academic exchange environment, and I believe it will remain so in the future. However, the emergence of these problematic users has undermined the fairness of the competition. Their purpose in open-sourcing is merely for \"scores\" rather than for innovation. Therefore, I hope the competition organizers and Kaggle officials will take this seriously.

\n

The Issue

\n

In the final 14 days leading up to the deadline, some users have published public notebooks that are nothing more than a simple weighted average or rank-blending of existing high-scoring public notebooks. These \"no-mod\" ensemble scripts provide:\nHere notebook is one of them

\n\n

Why This is Malicious

\n

While Kaggle encourages sharing, the timing and nature of these specific shares suggest a clear intent to distort the final standings (often referred to as \"shaking up\" the leaderboard maliciously) rather than to foster learning. It creates a \"race to the bottom\" where the highest public score becomes a commodity rather than a reflection of skill.

\n

Proposed Action

\n

I kindly request the Kaggle team to:

\n
    \n

  1. Investigate the accounts frequently posting these high-scoring, zero-originality ensembles just before the deadline.

    2. \n

  2. Enforce stricter \"Public Sharing\" guidelines: Consider a \"Freeze Period\" for public notebooks (e.g., no new public notebooks allowed within 7 days of the deadline) to prevent last-minute leaderboard manipulation.

    3. \n

  3. Review the Impact: Evaluate if these specific notebooks in [Insert Competition Name] should be hidden or if the authors should be warned for unsportsmanlike conduct.

    4. \n
\n

We value Kaggle as a platform for elite data science competition. Allowing this behavior to continue unaddressed diminishes the prestige of Kaggle Medals and the platform's credibility.

\n

Sincerely,

\n

mochi

", "rawMarkdown": "Dear Kaggle Team and Competition Organizers,\n\n@deeppast @ryanholbrook \n\nI am writing to formally report a growing trend of malicious public sharing in this competition, which I believe violates the spirit of \"Fair Play\" and undermines the integrity of the Kaggle community.I have always regarded Kaggle as a fair and competitive data science competition platform with a great academic exchange environment, and I believe it will remain so in the future. However, the emergence of these problematic users has undermined the fairness of the competition. Their purpose in open-sourcing is merely for \"scores\" rather than for innovation. Therefore, I hope the competition organizers and Kaggle officials will take this seriously.\n\n# The Issue\n\nIn the final 14 days leading up to the deadline, some users have published public notebooks that are nothing more than a simple weighted average or rank-blending of existing high-scoring public notebooks. These \"no-mod\" ensemble scripts provide:\n[Here notebook is one of them](https://www.kaggle.com/code/vitorhugobarbedo/lb-35-9-with-corrections-public-model)\n\n* Zero technical insight: They do not introduce new features, architectures, or methodologies.\n\n* Artificial leaderboard inflation: They allow hundreds of users to \"jump\" the leaderboard without contributing any original work.\n\n* Disruption of the competition ecosystem: This behavior punishes participants who have spent weeks on original research, only to see their rankings plummeted by \"one-click\" submissions.\n\n# Why This is Malicious\n\nWhile Kaggle encourages sharing, the timing and nature of these specific shares suggest a clear intent to distort the final standings (often referred to as \"shaking up\" the leaderboard maliciously) rather than to foster learning. It creates a \"race to the bottom\" where the highest public score becomes a commodity rather than a reflection of skill.\n\n# Proposed Action\n\nI kindly request the Kaggle team to:\n\n1. Investigate the accounts frequently posting these high-scoring, zero-originality ensembles just before the deadline.\n\n2. Enforce stricter \"Public Sharing\" guidelines: Consider a \"Freeze Period\" for public notebooks (e.g., no new public notebooks allowed within 7 days of the deadline) to prevent last-minute leaderboard manipulation.\n\n3. Review the Impact: Evaluate if these specific notebooks in [Insert Competition Name] should be hidden or if the authors should be warned for unsportsmanlike conduct.\n\nWe value Kaggle as a platform for elite data science competition. Allowing this behavior to continue unaddressed diminishes the prestige of Kaggle Medals and the platform's credibility.\n\nSincerely,\n\nmochi\n", "votes": 45 }, { "id": 3419231, "postDate": "2026-03-10T06:46:04.857Z", "content": "

I totally agree. Such behavior significantly demoralizes participants who are genuinely committed to original research. The core value of Kaggle lies in fostering innovation through competition.

", "rawMarkdown": "I totally agree. Such behavior significantly demoralizes participants who are genuinely committed to original research. The core value of Kaggle lies in fostering innovation through competition.", "votes": 6 }, { "id": 3419326, "postDate": "2026-03-10T13:30:07.073Z", "content": "

I think the rules are less concrete than they should be on this front. I dont think the rules are clear enough to retroactively punish people. However, the need at this point is clear. In the last several competitions I have been apart of I have seen medal selling, public notebooks being updated in only the last hours before competitions close (santa comp), rescores in the final days of the competition, and more. Not all of these are simple by any means, but this one certainly is. Develop clear rules about sharing and when, make it outright impossible to make a competition notebook public in the medal zone in the final month of competition, and post clear punishments for what happens if you do a work around.

", "rawMarkdown": "I think the rules are less concrete than they should be on this front. I dont think the rules are clear enough to retroactively punish people. However, the need at this point is clear. In the last several competitions I have been apart of I have seen medal selling, public notebooks being updated in only the last hours before competitions close (santa comp), rescores in the final days of the competition, and more. Not all of these are simple by any means, but this one certainly is. Develop clear rules about sharing and when, make it outright impossible to make a competition notebook public in the medal zone in the final month of competition, and post clear punishments for what happens if you do a work around.", "votes": 5, "replies": [ { "id": 3419342, "postDate": "2026-03-10T14:18:20.487Z", "content": "

Agreed, kinda harsh to call these notebooks malicious or malignant when they are explicitly allowed or even encouraged in some ways.

", "rawMarkdown": "Agreed, kinda harsh to call these notebooks malicious or malignant when they are explicitly allowed or even encouraged in some ways. ", "votes": 2 }, { "id": 3419343, "postDate": "2026-03-10T14:18:44.833Z", "content": "

Absolutely, \nAs your mentioned , I think the rules in this regard should be improved to ensure that open-source notebook do not always appear in the public code when the final DDL is approaching. Although the official guidelines encourage users to improve code quality and model performance in various ways, such behavior (publishing high-scoring notebook close to the deadline) obviously affects most people (who usually spend hundreds of dollars and a lot of time and effort to achieve the best score at present).

", "rawMarkdown": "Absolutely, \nAs your mentioned , I think the rules in this regard should be improved to ensure that open-source notebook do not always appear in the public code when the final DDL is approaching. Although the official guidelines encourage users to improve code quality and model performance in various ways, such behavior (publishing high-scoring notebook close to the deadline) obviously affects most people (**who usually spend hundreds of dollars and a lot of time and effort to achieve the best score at present**).", "votes": 2 } ] }, { "id": 3419239, "postDate": "2026-03-10T07:21:34.527Z", "content": "

Valid point, but since no explicit rules are being broken, it's unlikely Kaggle will take action. Regarding the 7-day freeze, I believe that is already implemented. Another important thing to consider is what @aaronbornstein mentioned: most of these public models are heavily overfitted. So, if your model is built properly, you will likely surpass the majority of them in the shake-up (assuming the private data doesn't just consist of sentences from train.csv).

", "rawMarkdown": "Valid point, but since no explicit rules are being broken, it's unlikely Kaggle will take action. Regarding the 7-day freeze, I believe that is already implemented. Another important thing to consider is what @aaronbornstein mentioned: most of these public models are heavily overfitted. So, if your model is built properly, you will likely surpass the majority of them in the shake-up (assuming the private data doesn't just consist of sentences from train.csv).", "votes": 4 }, { "id": 3419327, "postDate": "2026-03-10T13:30:37.507Z", "content": "

As I mentioned in the notebook, I believe that opening a public discussion on this topic is extremely important. Kaggle could create a metric that rewards the original author of the notebook, in addition to establishing a descending scoring system as the percentage and number of reposts or derivative modifications increase.

\n

Another possibility would be to implement some kind of additional indicator that would give greater visibility in “Hot” to 100% original scripts, especially those that presented new solutions or innovative approaches.

\n

I emphasize that my post was intended to share knowledge, including with the author of the notebook. Contrary to what the author of the thread says @lau01b, the notebook made Regex modifications that could help other users with private training models that could improve their results.

", "rawMarkdown": "As I mentioned in the notebook, I believe that opening a public discussion on this topic is extremely important. Kaggle could create a metric that rewards the original author of the notebook, in addition to establishing a descending scoring system as the percentage and number of reposts or derivative modifications increase.\n\nAnother possibility would be to implement some kind of additional indicator that would give greater visibility in “Hot” to 100% original scripts, especially those that presented new solutions or innovative approaches.\n\nI emphasize that my post was intended to share knowledge, including with the author of the notebook. Contrary to what the author of the thread says @lau01b, the notebook made Regex modifications that could help other users with private training models that could improve their results.", "votes": 2, "replies": [ { "id": 3419335, "postDate": "2026-03-10T13:54:37.883Z", "content": "

Don't worry, I don't think anyone will blame you. You didn't do anything wrong. They were just discussing and suggesting that notebook sharing should be turned off in the final few days—and obviously it's already closed now.

\n

I'm really amazed by your coding skills; you've fully leveraged the power of public models. It's a tough task, great job! :)

", "rawMarkdown": "Don't worry, I don't think anyone will blame you. You didn't do anything wrong. They were just discussing and suggesting that notebook sharing should be turned off in the final few days—and obviously it's already closed now.\n\nI'm really amazed by your coding skills; you've fully leveraged the power of public models. It's a tough task, great job! :)", "votes": 4, "replies": [ { "id": 3419341, "postDate": "2026-03-10T14:14:56.517Z", "content": "

Thank you, and good luck in the final stretch!!!!

", "rawMarkdown": "Thank you, and good luck in the final stretch!!!!" } ] } ] }, { "id": 3419267, "postDate": "2026-03-10T09:31:11.453Z", "content": "

The overfitting will stay at their notebook ：）

", "rawMarkdown": "The overfitting will stay at their notebook ：）" }, { "id": 3421564, "postDate": "2026-03-15T20:11:00.410Z", "content": "

The freeze period seems to have started, preventing public notebook sharing. Such restrictions can demoralize committed participants and undermine Kaggle’s goal of fostering innovation

", "rawMarkdown": "The freeze period seems to have started, preventing public notebook sharing. Such restrictions can demoralize committed participants and undermine Kaggle’s goal of fostering innovation", "votes": -9 }, { "id": 3419285, "postDate": "2026-03-10T10:34:30.063Z", "rawMarkdown": "", "votes": -4, "isDeleted": true }, { "id": 3419281, "postDate": "2026-03-10T10:20:06.810Z", "rawMarkdown": "", "votes": 2, "isDeleted": true }, { "id": 3421229, "postDate": "2026-03-15T01:03:00.103Z", "content": "

Thank you for the issue info @lau01b

", "rawMarkdown": "Thank you for the issue info @lau01b ", "votes": 1 } ], "comments": [ { "id": 3419231, "author_name": "Colhomy", "author_url": "", "post_date": "2026-03-10T06:46:04.857000", "content": "

I totally agree. Such behavior significantly demoralizes participants who are genuinely committed to original research. The core value of Kaggle lies in fostering innovation through competition.

", "votes": 6, "replies": [] }, { "id": 3419326, "author_name": "Cody_Null", "author_url": "", "post_date": "2026-03-10T13:30:07.073000", "content": "

I think the rules are less concrete than they should be on this front. I dont think the rules are clear enough to retroactively punish people. However, the need at this point is clear. In the last several competitions I have been apart of I have seen medal selling, public notebooks being updated in only the last hours before competitions close (santa comp), rescores in the final days of the competition, and more. Not all of these are simple by any means, but this one certainly is. Develop clear rules about sharing and when, make it outright impossible to make a competition notebook public in the medal zone in the final month of competition, and post clear punishments for what happens if you do a work around.

", "votes": 5, "replies": [ { "id": 3419342, "author_name": "tennogh", "author_url": "", "post_date": "2026-03-10T14:18:20.487000", "content": "

Agreed, kinda harsh to call these notebooks malicious or malignant when they are explicitly allowed or even encouraged in some ways.

", "votes": 2, "replies": [] }, { "id": 3419343, "author_name": "StarxSky", "author_url": "", "post_date": "2026-03-10T14:18:44.833000", "content": "

Absolutely, \nAs your mentioned , I think the rules in this regard should be improved to ensure that open-source notebook do not always appear in the public code when the final DDL is approaching. Although the official guidelines encourage users to improve code quality and model performance in various ways, such behavior (publishing high-scoring notebook close to the deadline) obviously affects most people (who usually spend hundreds of dollars and a lot of time and effort to achieve the best score at present).

", "votes": 2, "replies": [] } ] }, { "id": 3419239, "author_name": "Giovanny Rodríguez", "author_url": "", "post_date": "2026-03-10T07:21:34.527000", "content": "

Valid point, but since no explicit rules are being broken, it's unlikely Kaggle will take action. Regarding the 7-day freeze, I believe that is already implemented. Another important thing to consider is what @aaronbornstein mentioned: most of these public models are heavily overfitted. So, if your model is built properly, you will likely surpass the majority of them in the shake-up (assuming the private data doesn't just consist of sentences from train.csv).

", "votes": 4, "replies": [] }, { "id": 3419327, "author_name": "Vitor Hugo Barbedo", "author_url": "", "post_date": "2026-03-10T13:30:37.507000", "content": "

As I mentioned in the notebook, I believe that opening a public discussion on this topic is extremely important. Kaggle could create a metric that rewards the original author of the notebook, in addition to establishing a descending scoring system as the percentage and number of reposts or derivative modifications increase.

\n

Another possibility would be to implement some kind of additional indicator that would give greater visibility in “Hot” to 100% original scripts, especially those that presented new solutions or innovative approaches.

\n

I emphasize that my post was intended to share knowledge, including with the author of the notebook. Contrary to what the author of the thread says @lau01b, the notebook made Regex modifications that could help other users with private training models that could improve their results.

", "votes": 2, "replies": [ { "id": 3419335, "author_name": "耶✌", "author_url": "", "post_date": "2026-03-10T13:54:37.883000", "content": "

Don't worry, I don't think anyone will blame you. You didn't do anything wrong. They were just discussing and suggesting that notebook sharing should be turned off in the final few days—and obviously it's already closed now.

\n

I'm really amazed by your coding skills; you've fully leveraged the power of public models. It's a tough task, great job! :)

", "votes": 4, "replies": [ { "id": 3419341, "author_name": "Vitor Hugo Barbedo", "author_url": "", "post_date": "2026-03-10T14:14:56.517000", "content": "

Thank you, and good luck in the final stretch!!!!

", "votes": 0, "replies": [] } ] } ] }, { "id": 3419267, "author_name": "Zejun_", "author_url": "", "post_date": "2026-03-10T09:31:11.453000", "content": "

The overfitting will stay at their notebook ：）

", "votes": 0, "replies": [] }, { "id": 3421564, "author_name": "", "author_url": "", "post_date": "2026-03-15T20:11:00.410000", "content": "

The freeze period seems to have started, preventing public notebook sharing. Such restrictions can demoralize committed participants and undermine Kaggle’s goal of fostering innovation

", "votes": -9, "replies": [] }, { "id": 3419285, "author_name": "", "author_url": "", "post_date": "2026-03-10T10:34:30.063000", "content": "", "votes": -4, "replies": [] }, { "id": 3419281, "author_name": "", "author_url": "", "post_date": "2026-03-10T10:20:06.810000", "content": "", "votes": 2, "replies": [] }, { "id": 3421229, "author_name": "Navneet", "author_url": "", "post_date": "2026-03-15T01:03:00.103000", "content": "

Thank you for the issue info @lau01b

", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "3419222": "Dear Kaggle Team and Competition Organizers,\n\n@deeppast @ryanholbrook \n\nI am writing to formally report a growing trend of malicious public sharing in this competition, which I believe violates the spirit of \"Fair Play\" and undermines the integrity of the Kaggle community.I have always regarded Kaggle as a fair and competitive data science competition platform with a great academic exchange environment, and I believe it will remain so in the future. However, the emergence of these problematic users has undermined the fairness of the competition. Their purpose in open-sourcing is merely for \"scores\" rather than for innovation. Therefore, I hope the competition organizers and Kaggle officials will take this seriously.\n\n# The Issue\n\nIn the final 14 days leading up to the deadline, some users have published public notebooks that are nothing more than a simple weighted average or rank-blending of existing high-scoring public notebooks. These \"no-mod\" ensemble scripts provide:\n[Here notebook is one of them](https://www.kaggle.com/code/vitorhugobarbedo/lb-35-9-with-corrections-public-model)\n\n* Zero technical insight: They do not introduce new features, architectures, or methodologies.\n\n* Artificial leaderboard inflation: They allow hundreds of users to \"jump\" the leaderboard without contributing any original work.\n\n* Disruption of the competition ecosystem: This behavior punishes participants who have spent weeks on original research, only to see their rankings plummeted by \"one-click\" submissions.\n\n# Why This is Malicious\n\nWhile Kaggle encourages sharing, the timing and nature of these specific shares suggest a clear intent to distort the final standings (often referred to as \"shaking up\" the leaderboard maliciously) rather than to foster learning. It creates a \"race to the bottom\" where the highest public score becomes a commodity rather than a reflection of skill.\n\n# Proposed Action\n\nI kindly request the Kaggle team to:\n\n1. Investigate the accounts frequently posting these high-scoring, zero-originality ensembles just before the deadline.\n\n2. Enforce stricter \"Public Sharing\" guidelines: Consider a \"Freeze Period\" for public notebooks (e.g., no new public notebooks allowed within 7 days of the deadline) to prevent last-minute leaderboard manipulation.\n\n3. Review the Impact: Evaluate if these specific notebooks in [Insert Competition Name] should be hidden or if the authors should be warned for unsportsmanlike conduct.\n\nWe value Kaggle as a platform for elite data science competition. Allowing this behavior to continue unaddressed diminishes the prestige of Kaggle Medals and the platform's credibility.\n\nSincerely,\n\nmochi\n", "3419231": "I totally agree. Such behavior significantly demoralizes participants who are genuinely committed to original research. The core value of Kaggle lies in fostering innovation through competition.", "3419326": "I think the rules are less concrete than they should be on this front. I dont think the rules are clear enough to retroactively punish people. However, the need at this point is clear. In the last several competitions I have been apart of I have seen medal selling, public notebooks being updated in only the last hours before competitions close (santa comp), rescores in the final days of the competition, and more. Not all of these are simple by any means, but this one certainly is. Develop clear rules about sharing and when, make it outright impossible to make a competition notebook public in the medal zone in the final month of competition, and post clear punishments for what happens if you do a work around.", "3419239": "Valid point, but since no explicit rules are being broken, it's unlikely Kaggle will take action. Regarding the 7-day freeze, I believe that is already implemented. Another important thing to consider is what @aaronbornstein mentioned: most of these public models are heavily overfitted. So, if your model is built properly, you will likely surpass the majority of them in the shake-up (assuming the private data doesn't just consist of sentences from train.csv).", "3419327": "As I mentioned in the notebook, I believe that opening a public discussion on this topic is extremely important. Kaggle could create a metric that rewards the original author of the notebook, in addition to establishing a descending scoring system as the percentage and number of reposts or derivative modifications increase.\n\nAnother possibility would be to implement some kind of additional indicator that would give greater visibility in “Hot” to 100% original scripts, especially those that presented new solutions or innovative approaches.\n\nI emphasize that my post was intended to share knowledge, including with the author of the notebook. Contrary to what the author of the thread says @lau01b, the notebook made Regex modifications that could help other users with private training models that could improve their results.", "3419267": "The overfitting will stay at their notebook ：）", "3421564": "The freeze period seems to have started, preventing public notebook sharing. Such restrictions can demoralize committed participants and undermine Kaggle’s goal of fostering innovation", "3419285": "", "3419281": "", "3421229": "Thank you for the issue info @lau01b " } }