{
  "id": 64366,
  "title": "Appeal to Kaggle",
  "url": "/competitions/airbus-ship-detection/discussion/64366",
  "author_name": "",
  "post_date": "2018-08-28T15:57:59.388117300Z",
  "votes": 34,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Given the current situation with the data leak, I decided to share my concerns about Kaggle's future as a top-tier data science platform. </p>\n\n<p>I think many will agree with the statement that one's Kaggle performance is currently something that is valuable both in academia and industry, even despite some speculations about non-realistic approaches and perfectly (not always, haha) clean datasets.</p>\n\n<p>I firmly convinced such leaky competitions are a big threat to Kaggle reputation. I believe Kaggle team is interested in professional data scientists to compete on this platform, but I fear if data will not be checked properly the majority of good specialists will abandon Kaggle in favor of other platforms. </p>\n\n<p>In addition to the reputation damage, leak disclosure after the start of a competition is always discouraging, often making efforts of all participants to be worthless.</p>\n\n<p>I ask Kaggle team to put more effort on data preparation and leak disclosure in the future competitions. I also think that current competition should be restarted with another leak-free dataset provided.</p>\n\n<p>Happy kaggling,</p>\n\n<p>Dmitriy</p>",
  "messages": [
    {
      "id": "377128",
      "postDate": "08/28/2018 15:57:59",
      "content": "<p>Given the current situation with the data leak, I decided to share my concerns about Kaggle's future as a top-tier data science platform. </p>\n\n<p>I think many will agree with the statement that one's Kaggle performance is currently something that is valuable both in academia and industry, even despite some speculations about non-realistic approaches and perfectly (not always, haha) clean datasets.</p>\n\n<p>I firmly convinced such leaky competitions are a big threat to Kaggle reputation. I believe Kaggle team is interested in professional data scientists to compete on this platform, but I fear if data will not be checked properly the majority of good specialists will abandon Kaggle in favor of other platforms. </p>\n\n<p>In addition to the reputation damage, leak disclosure after the start of a competition is always discouraging, often making efforts of all participants to be worthless.</p>\n\n<p>I ask Kaggle team to put more effort on data preparation and leak disclosure in the future competitions. I also think that current competition should be restarted with another leak-free dataset provided.</p>\n\n<p>Happy kaggling,</p>\n\n<p>Dmitriy</p>",
      "rawMarkdown": "Given the current situation with the data leak, I decided to share my concerns about Kaggle's future as a top-tier data science platform. \n\nI think many will agree with the statement that one's Kaggle performance is currently something that is valuable both in academia and industry, even despite some speculations about non-realistic approaches and perfectly (not always, haha) clean datasets.\n\nI firmly convinced such leaky competitions are a big threat to Kaggle reputation. I believe Kaggle team is interested in professional data scientists to compete on this platform, but I fear if data will not be checked properly the majority of good specialists will abandon Kaggle in favor of other platforms. \n\nIn addition to the reputation damage, leak disclosure after the start of a competition is always discouraging, often making efforts of all participants to be worthless.\n\nI ask Kaggle team to put more effort on data preparation and leak disclosure in the future competitions. I also think that current competition should be restarted with another leak-free dataset provided.\n\nHappy kaggling,\n\nDmitriy",
      "votes": null
    },
    {
      "id": "377374",
      "postDate": "08/29/2018 03:40:14",
      "content": "<p>Hi Dmitriy -</p>\n\n<p>Thanks for the perspective. I agree with your reasons to keep leaks to a minimum. What happened in this competition was quite the blunder on my part, and we've added a new protocol to catch obvious issues like this. <a href=\"https://www.kaggle.com/c/airbus-ship-detection/discussion/64388\">Details in this thread</a>.</p>",
      "rawMarkdown": "Hi Dmitriy -\n\nThanks for the perspective. I agree with your reasons to keep leaks to a minimum. What happened in this competition was quite the blunder on my part, and we've added a new protocol to catch obvious issues like this. [Details in this thread][1].\n\n\n  [1]: https://www.kaggle.com/c/airbus-ship-detection/discussion/64388",
      "votes": null
    },
    {
      "id": "377482",
      "postDate": "08/29/2018 08:31:28",
      "content": "<p><a href=\"/inversion\">@inversion</a></p>\n\n<p>Thanks for such a detailed and quick response! From what you wrote it's clear this leak was just a silly mistake everyone makes from time to time, so I believe there is no reason for you to reproach yourself. </p>\n\n<p>Looking forward to seeing new test set :)</p>",
      "rawMarkdown": "inversion\n\nThanks for such a detailed and quick response! From what you wrote it's clear this leak was just a silly mistake everyone makes from time to time, so I believe there is no reason for you to reproach yourself. \n\nLooking forward to seeing new test set :)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 377374,
      "author_name": "inversion",
      "author_url": "",
      "post_date": "08/29/2018 03:40:14",
      "content": "<p>Hi Dmitriy -</p>\n\n<p>Thanks for the perspective. I agree with your reasons to keep leaks to a minimum. What happened in this competition was quite the blunder on my part, and we've added a new protocol to catch obvious issues like this. <a href=\"https://www.kaggle.com/c/airbus-ship-detection/discussion/64388\">Details in this thread</a>.</p>",
      "votes": null,
      "replies": [
        {
          "id": 377482,
          "author_name": "ddanevskyi",
          "author_url": "",
          "post_date": "08/29/2018 08:31:28",
          "content": "<p><a href=\"/inversion\">@inversion</a></p>\n\n<p>Thanks for such a detailed and quick response! From what you wrote it's clear this leak was just a silly mistake everyone makes from time to time, so I believe there is no reason for you to reproach yourself. </p>\n\n<p>Looking forward to seeing new test set :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "377128": "Given the current situation with the data leak, I decided to share my concerns about Kaggle's future as a top-tier data science platform. \n\nI think many will agree with the statement that one's Kaggle performance is currently something that is valuable both in academia and industry, even despite some speculations about non-realistic approaches and perfectly (not always, haha) clean datasets.\n\nI firmly convinced such leaky competitions are a big threat to Kaggle reputation. I believe Kaggle team is interested in professional data scientists to compete on this platform, but I fear if data will not be checked properly the majority of good specialists will abandon Kaggle in favor of other platforms. \n\nIn addition to the reputation damage, leak disclosure after the start of a competition is always discouraging, often making efforts of all participants to be worthless.\n\nI ask Kaggle team to put more effort on data preparation and leak disclosure in the future competitions. I also think that current competition should be restarted with another leak-free dataset provided.\n\nHappy kaggling,\n\nDmitriy",
    "377374": "Hi Dmitriy -\n\nThanks for the perspective. I agree with your reasons to keep leaks to a minimum. What happened in this competition was quite the blunder on my part, and we've added a new protocol to catch obvious issues like this. [Details in this thread][1].\n\n\n  [1]: https://www.kaggle.com/c/airbus-ship-detection/discussion/64388",
    "377482": "inversion\n\nThanks for such a detailed and quick response! From what you wrote it's clear this leak was just a silly mistake everyone makes from time to time, so I believe there is no reason for you to reproach yourself. \n\nLooking forward to seeing new test set :)"
  },
  "source": "meta"
}