{
  "id": 56248,
  "title": "Best way to deal with the situation (everyone happy)",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/56248",
  "author_name": "",
  "post_date": "2018-05-08T01:27:12.423795800Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Due to sharing , then removal of high score submission and kernel, there is a general resentment among Kagglers.\nCan Kaggle calculate a rank  (old) just before that kernel was published on the basis of best public LB (selection), then add one more column which mentions rank before that Kernel was published.  </p>\n\n<p>The points awarded to each competitor shall be max(points old LB ,points new LB)\nThoughts?</p>",
  "messages": [
    {
      "id": "324904",
      "postDate": "05/08/2018 01:27:12",
      "content": "<p>Due to sharing , then removal of high score submission and kernel, there is a general resentment among Kagglers.\nCan Kaggle calculate a rank  (old) just before that kernel was published on the basis of best public LB (selection), then add one more column which mentions rank before that Kernel was published.  </p>\n\n<p>The points awarded to each competitor shall be max(points old LB ,points new LB)\nThoughts?</p>",
      "rawMarkdown": "Due to sharing , then removal of high score submission and kernel, there is a general resentment among Kagglers.\nCan Kaggle calculate a rank  (old) just before that kernel was published on the basis of best public LB (selection), then add one more column which mentions rank before that Kernel was published.  \n \nThe points awarded to each competitor shall be max(points old LB ,points new LB)\nThoughts?",
      "votes": null
    },
    {
      "id": "324947",
      "postDate": "05/08/2018 02:04:39",
      "content": "<p>Taking a weighted mean over a historical window of daily/weekly best scores is a reasonable approach to reflect the persistent efforts people have dedicated, hopefully well reformed rules not getting over-complicated and pushing people away.</p>",
      "rawMarkdown": "Taking a weighted mean over a historical window of daily/weekly best scores is a reasonable approach to reflect the persistent efforts people have dedicated, hopefully well reformed rules not getting over-complicated and pushing people away.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 324947,
      "author_name": "dongxu027",
      "author_url": "",
      "post_date": "05/08/2018 02:04:39",
      "content": "<p>Taking a weighted mean over a historical window of daily/weekly best scores is a reasonable approach to reflect the persistent efforts people have dedicated, hopefully well reformed rules not getting over-complicated and pushing people away.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "324904": "Due to sharing , then removal of high score submission and kernel, there is a general resentment among Kagglers.\nCan Kaggle calculate a rank  (old) just before that kernel was published on the basis of best public LB (selection), then add one more column which mentions rank before that Kernel was published.  \n \nThe points awarded to each competitor shall be max(points old LB ,points new LB)\nThoughts?",
    "324947": "Taking a weighted mean over a historical window of daily/weekly best scores is a reasonable approach to reflect the persistent efforts people have dedicated, hopefully well reformed rules not getting over-complicated and pushing people away."
  },
  "source": "meta"
}