{
  "id": 2837,
  "title": "Really swift learning?",
  "url": "/competitions/MerckActivity/discussion/2837",
  "author_name": "Shea Parkes",
  "post_date": "2012-10-08T15:49:20.023000",
  "votes": 8,
  "comment_count": 26,
  "views": 12126,
  "content": "I'm hoping I just found a glitch in the user-profiles, but this seems damn impressive: http://www.kaggle.com/users/62008/finik A user for 15 hours and he's already in the top 10 of this competition? In fact, the date stamps of his first submission would\r\n mean his account was under 5 hours old when he made that. Of course, maybe I'm just getting out classed. Maybe it only takes a few hours to download, load, train and submit a winning prediction. Maybe I should extend a job offer to finik through the new contact\r\n feature? I would love to know his tool chain.",
  "messages": [
    {
      "id": 15277,
      "postDate": "2012-10-08T15:49:20.023Z",
      "content": "I'm hoping I just found a glitch in the user-profiles, but this seems damn impressive: http://www.kaggle.com/users/62008/finik A user for 15 hours and he's already in the top 10 of this competition? In fact, the date stamps of his first submission would\r\n mean his account was under 5 hours old when he made that. Of course, maybe I'm just getting out classed. Maybe it only takes a few hours to download, load, train and submit a winning prediction. Maybe I should extend a job offer to finik through the new contact\r\n feature? I would love to know his tool chain.",
      "votes": 8
    },
    {
      "id": 15632,
      "postDate": "2012-10-15T18:26:51.803Z",
      "content": "<p>[quote=Black Magic;15624]</p>\r\n<p>that could probably explain why there are 284 teams in the competition. Usually for such large datasets, there are fewer teams. I am surprised by the # of teams</p>\r\n<p>[/quote]</p>\r\n<p>Number of teams by day:</p>\r\n<p>October &nbsp;1 : 167</p>\r\n<p>October &nbsp;2: 168</p>\r\n<p>October &nbsp;3 : 177</p>\r\n<p>October &nbsp;4 : 183</p>\r\n<p>October &nbsp;5: 186</p>\r\n<p>October &nbsp;6 : 190</p>\r\n<p>October &nbsp;7 : 195</p>\r\n<p>October &nbsp;8: 199</p>\r\n<p>October &nbsp;9 : 203</p>\r\n<p>October &nbsp;10: 211</p>\r\n<p>October &nbsp;11: 234</p>\r\n<p>October &nbsp;12 : 239</p>\r\n<p>October &nbsp;13 : 243</p>\r\n<p>October &nbsp;14 : 243</p>\r\n<p>October &nbsp;1 5: 262</p>\r\n<p>now : 284</p>\r\n<p>Do we need to run competition to find anomalies?</p>\r\n<p>It is high profile competition. It will be interesting to see how Kaggle will handle this situation. It will be easy to find sock puppets.<br>\r\nWill they be able to find the maser(s)? Those who created multiple accounts, probably, already have enough information to be among winners.</p>\r\n<p>&nbsp;</p>",
      "votes": 6
    },
    {
      "id": 15653,
      "postDate": "2012-10-15T22:16:36.247Z",
      "content": "<p>Hi Sergey,</p>\r\n<p>We watch all competitions using the approaches you described, among others. &nbsp;We check all the highly-ranked teams in particular. &nbsp;Still, sometimes users may see suspicious patterns first, so we appreciate people bringing those to our attention. &nbsp;</p>\r\n<p>Public posts that discuss tactics for circumventing the rules and approaches for detecting sock-puppets may help the offenders become more sophisticated. I think it's best to keep those topics out of the forums.</p>\r\n<p>-Guy</p>",
      "votes": 1
    },
    {
      "id": 15651,
      "postDate": "2012-10-15T21:24:10.893Z",
      "content": "<p>Hi Everyone,</p>\r\n<p>Thanks to you all for helping identify puppet accounts. We take this issue seriously at Kaggle and want to create an environment where people compete in an honest and fair way. We try our best to find and adjudicate duplicate account holders, but it's a\r\n difficult problem that sometimes grows faster than we can handle.</p>\r\n<p>We continually discuss different approaches for how to reduce misuse of the public leaderboard and multiple account creation. We plan to implement some of these approaches on our future competitions.</p>\r\n<p>Please send your tips to compliance&#43;merck@kaggle.com and we'll look into any suspicious accounts.</p>",
      "votes": 1
    },
    {
      "id": 15637,
      "postDate": "2012-10-15T18:58:37.557Z",
      "content": "<p>Sergey, nice data :) There's a clear anomaly on October 10-11 and 14-15. <br>\r\nIndeed, I'm interested to see how Kaggle will handle this. Can the cheaters be identified by their IP / other info?\r\n</p>",
      "votes": 1
    },
    {
      "id": 15841,
      "postDate": "2012-10-17T17:26:58.817Z",
      "content": "<p>Thankfully, so did Kaggle (given that they are holding off on releasing the offical Private leaderboard). I think that revealing the private LB and then altering it would have tainted things. It looks like they are taking this very seriously and handling\r\n it the best they can.</p>\r\n<p>Good luck to all!</p>"
    },
    {
      "id": 15835,
      "postDate": "2012-10-17T16:59:49.487Z",
      "content": "<p>agree with jeremy achin.</p>"
    },
    {
      "id": 15720,
      "postDate": "2012-10-16T17:41:14.743Z",
      "content": "<p>It would probably be better to disqualify the team(s) that were connected to the dummy accounts now rather than wait until after the competition ends.&nbsp;</p>\r\n<p>Does anyone else agree with me on this?</p>"
    },
    {
      "id": 15674,
      "postDate": "2012-10-16T00:47:49.550Z",
      "content": "<p>I've created a <a href=\"https://www.kaggle.com/forums/t/2898/ideas-for-competition-changes-to-reduce-cheating-multiple-accounts\">\r\nnew forum thread</a> in the General Kaggle Forum to discuss these issues with the larger community. Please see that post and send us your feedback on that thread. Thanks to everyone so far on your comments, please continue to let us know what you think will\r\n work better for our competitions.</p>"
    },
    {
      "id": 15669,
      "postDate": "2012-10-16T00:28:28.897Z",
      "content": "<p>[quote=jcnhvnhck;15651]</p>\r\n<p>We continually discuss different approaches for how to reduce misuse of the public leaderboard and multiple account creation.</p>\r\n<p>[/quote]</p>\r\n<p>What misuse you meant? In my opinion we all (and puppets too) use it by right way (submit and got score). Puppets do not misuse the leaderboard they violate your submission limit per day and make competition unfair by violating principle of equal conditions\r\n to all contestants. Similar thing you do by enabling aggregation contestants in teams, in short period of time a team will have access to bigger amount of scores/submissions than one contestant - in my opinion it violates the principle too (why team can make\r\n n*10 submissions in five days before aggregation but one contestant can't). Besides, who can prove that two submissions per day is right amount for any contest?<br>\r\n<br>\r\nLook on public leaderboard as on additional validation set (in other case what the public leaderboard is). Remove or change meaningless limit (introduce some limit to the stored submissions if you need this). You will solve puppets problem and probably will\r\n have better outcomes (at least not worse) for the customers (contestants will have ability for better using additional validation). And we all will compete in more equal conditions. Or support your restriction, discuss/try different approaches to reduce the\r\n misusing and fight with puppets - it is your choice :)</p>"
    },
    {
      "id": 15662,
      "postDate": "2012-10-15T23:14:04.517Z",
      "content": "<p>I personally would welcome a more open discussion. It sounds a lot like security though obscurity at the moment.</p>\r\n<p>Still, I'm happy the admins are taking it seriously and spending time on it.</p>"
    },
    {
      "id": 15655,
      "postDate": "2012-10-15T22:31:24.400Z",
      "content": "<p>[quote=Guy Cavet;15653]</p>\r\n<p>Hi Sergey,</p>\r\n<p>Public posts that discuss tactics for circumventing the rules and approaches for detecting sock-puppets may help the offenders become more sophisticated. I think it's best to keep those topics out of the forums.</p>\r\n<p>-Guy</p>\r\n<p>[/quote]</p>\r\n<p>&nbsp;</p>\r\n<p>I was not planning to discuss tactics. I just mentioned some obvious things. Anyway, I edited my post.</p>\r\n<p>P.S. Actually, I offered my assistance in private e-mail, but it was not accepted.</p>"
    },
    {
      "id": 15652,
      "postDate": "2012-10-15T21:41:30.833Z",
      "content": "<p>[quote=jcnhvnhck;15651]</p>\r\n<p>Please send your tips to compliance&#43;merck@kaggle.com and we'll look into any suspicious accounts.</p>\r\n<p>[/quote]</p>\r\n<p>Do I need to list all accounts, let say younger than 10 days? Maybe you can just apply filter and check them without&nbsp;my e-mail?</p>\r\n<p>(removed)</p>\r\n<p>By the way, what is the story with dajiangyou? He(?) jumped today to 2nd place, then after my complain his(?) last submission was removed&nbsp; and he(?) is back to #133 but not out.</p>\r\n<p>&nbsp;</p>"
    },
    {
      "id": 15650,
      "postDate": "2012-10-15T21:16:25.407Z",
      "content": "<p>[quote=Halla;15648]</p>\r\n<p>Changing the public private split from 25/75 to 1/99 could help in future competitions</p>\r\n<p>[/quote]</p>\r\n<p>Follow your logic removing the public leaderboard makes future competitions perfect ;)</p>"
    },
    {
      "id": 15649,
      "postDate": "2012-10-15T21:14:10.167Z",
      "content": "<p>May be better to fight with a reason - low amount of submissions per day. Look on the public hold-out set as on additional validation set and there will be no problem if anyone will have ability to use it enough amount of times.</p>"
    },
    {
      "id": 15645,
      "postDate": "2012-10-15T20:46:14.497Z",
      "content": "<p>for all you know they might just overfit the leaderboard (Let's hope!)</p>\r\n<p>For the 284&#43; participants, is there no way Kaggle can find and weed out sock puppets? It is unfair on participants&nbsp; who have been honest with only 1 account - either increase number of entries for all, round off to 1 significant digit or weed out the sock\r\n puppets from the 284</p>"
    },
    {
      "id": 15628,
      "postDate": "2012-10-15T18:13:03.413Z",
      "content": "<p>the follow-up money stage must be a full-fledged competition with training and test. They ran one in Impermium where there was only a new dataset and I must say - it was one of the worst scoring datasets ever totally different from training and test</p>"
    },
    {
      "id": 15624,
      "postDate": "2012-10-15T17:44:02.697Z",
      "content": "<p>that could probably explain why there are 284 teams in the competition. Usually for such large datasets, there are fewer teams. I am surprised by the # of teams</p>"
    },
    {
      "id": 15618,
      "postDate": "2012-10-15T17:08:46.040Z",
      "content": "<p>One, probably, can use provided RF benchmark and replace prediction only for one set.\r\n</p>"
    },
    {
      "id": 15293,
      "postDate": "2012-10-08T17:08:13.813Z",
      "content": "<p>the dataset is so huge that a RF will run for 6-7 hours for 15 essaysets. Wonderful ability indeed - is he using a supercomputer or something?</p>"
    },
    {
      "id": 15292,
      "postDate": "2012-10-08T17:07:09.427Z",
      "content": "<p>Amazing! - Such people should get a special award as an incentive to disclose their methods :)</p>\r\n<p>Good catch Shea!</p>"
    },
    {
      "id": 15648,
      "postDate": "2012-10-15T21:11:35.360Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 15639,
      "postDate": "2012-10-15T19:08:53.793Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    },
    {
      "id": 15626,
      "postDate": "2012-10-15T17:56:53.013Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 15615,
      "postDate": "2012-10-15T16:50:09.857Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 15287,
      "postDate": "2012-10-08T16:49:20.777Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 15286,
      "postDate": "2012-10-08T16:46:58.110Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 15632,
      "author_name": "Sergey Yurgenson",
      "author_url": "",
      "post_date": "2012-10-15T18:26:51.803000",
      "content": "<p>[quote=Black Magic;15624]</p>\r\n<p>that could probably explain why there are 284 teams in the competition. Usually for such large datasets, there are fewer teams. I am surprised by the # of teams</p>\r\n<p>[/quote]</p>\r\n<p>Number of teams by day:</p>\r\n<p>October &nbsp;1 : 167</p>\r\n<p>October &nbsp;2: 168</p>\r\n<p>October &nbsp;3 : 177</p>\r\n<p>October &nbsp;4 : 183</p>\r\n<p>October &nbsp;5: 186</p>\r\n<p>October &nbsp;6 : 190</p>\r\n<p>October &nbsp;7 : 195</p>\r\n<p>October &nbsp;8: 199</p>\r\n<p>October &nbsp;9 : 203</p>\r\n<p>October &nbsp;10: 211</p>\r\n<p>October &nbsp;11: 234</p>\r\n<p>October &nbsp;12 : 239</p>\r\n<p>October &nbsp;13 : 243</p>\r\n<p>October &nbsp;14 : 243</p>\r\n<p>October &nbsp;1 5: 262</p>\r\n<p>now : 284</p>\r\n<p>Do we need to run competition to find anomalies?</p>\r\n<p>It is high profile competition. It will be interesting to see how Kaggle will handle this situation. It will be easy to find sock puppets.<br>\r\nWill they be able to find the maser(s)? Those who created multiple accounts, probably, already have enough information to be among winners.</p>\r\n<p>&nbsp;</p>",
      "votes": 6,
      "replies": []
    },
    {
      "id": 15653,
      "author_name": "Guy Cavet",
      "author_url": "",
      "post_date": "2012-10-15T22:16:36.247000",
      "content": "<p>Hi Sergey,</p>\r\n<p>We watch all competitions using the approaches you described, among others. &nbsp;We check all the highly-ranked teams in particular. &nbsp;Still, sometimes users may see suspicious patterns first, so we appreciate people bringing those to our attention. &nbsp;</p>\r\n<p>Public posts that discuss tactics for circumventing the rules and approaches for detecting sock-puppets may help the offenders become more sophisticated. I think it's best to keep those topics out of the forums.</p>\r\n<p>-Guy</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 15651,
      "author_name": "joycenv",
      "author_url": "",
      "post_date": "2012-10-15T21:24:10.893000",
      "content": "<p>Hi Everyone,</p>\r\n<p>Thanks to you all for helping identify puppet accounts. We take this issue seriously at Kaggle and want to create an environment where people compete in an honest and fair way. We try our best to find and adjudicate duplicate account holders, but it's a\r\n difficult problem that sometimes grows faster than we can handle.</p>\r\n<p>We continually discuss different approaches for how to reduce misuse of the public leaderboard and multiple account creation. We plan to implement some of these approaches on our future competitions.</p>\r\n<p>Please send your tips to compliance&#43;merck@kaggle.com and we'll look into any suspicious accounts.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 15637,
      "author_name": "dmitrim",
      "author_url": "",
      "post_date": "2012-10-15T18:58:37.557000",
      "content": "<p>Sergey, nice data :) There's a clear anomaly on October 10-11 and 14-15. <br>\r\nIndeed, I'm interested to see how Kaggle will handle this. Can the cheaters be identified by their IP / other info?\r\n</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 15841,
      "author_name": "Jeremy Achin",
      "author_url": "",
      "post_date": "2012-10-17T17:26:58.817000",
      "content": "<p>Thankfully, so did Kaggle (given that they are holding off on releasing the offical Private leaderboard). I think that revealing the private LB and then altering it would have tainted things. It looks like they are taking this very seriously and handling\r\n it the best they can.</p>\r\n<p>Good luck to all!</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15835,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-17T16:59:49.487000",
      "content": "<p>agree with jeremy achin.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15720,
      "author_name": "Jeremy Achin",
      "author_url": "",
      "post_date": "2012-10-16T17:41:14.743000",
      "content": "<p>It would probably be better to disqualify the team(s) that were connected to the dummy accounts now rather than wait until after the competition ends.&nbsp;</p>\r\n<p>Does anyone else agree with me on this?</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15674,
      "author_name": "joycenv",
      "author_url": "",
      "post_date": "2012-10-16T00:47:49.550000",
      "content": "<p>I've created a <a href=\"https://www.kaggle.com/forums/t/2898/ideas-for-competition-changes-to-reduce-cheating-multiple-accounts\">\r\nnew forum thread</a> in the General Kaggle Forum to discuss these issues with the larger community. Please see that post and send us your feedback on that thread. Thanks to everyone so far on your comments, please continue to let us know what you think will\r\n work better for our competitions.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15669,
      "author_name": "alegro",
      "author_url": "",
      "post_date": "2012-10-16T00:28:28.897000",
      "content": "<p>[quote=jcnhvnhck;15651]</p>\r\n<p>We continually discuss different approaches for how to reduce misuse of the public leaderboard and multiple account creation.</p>\r\n<p>[/quote]</p>\r\n<p>What misuse you meant? In my opinion we all (and puppets too) use it by right way (submit and got score). Puppets do not misuse the leaderboard they violate your submission limit per day and make competition unfair by violating principle of equal conditions\r\n to all contestants. Similar thing you do by enabling aggregation contestants in teams, in short period of time a team will have access to bigger amount of scores/submissions than one contestant - in my opinion it violates the principle too (why team can make\r\n n*10 submissions in five days before aggregation but one contestant can't). Besides, who can prove that two submissions per day is right amount for any contest?<br>\r\n<br>\r\nLook on public leaderboard as on additional validation set (in other case what the public leaderboard is). Remove or change meaningless limit (introduce some limit to the stored submissions if you need this). You will solve puppets problem and probably will\r\n have better outcomes (at least not worse) for the customers (contestants will have ability for better using additional validation). And we all will compete in more equal conditions. Or support your restriction, discuss/try different approaches to reduce the\r\n misusing and fight with puppets - it is your choice :)</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15662,
      "author_name": "Shea Parkes",
      "author_url": "",
      "post_date": "2012-10-15T23:14:04.517000",
      "content": "<p>I personally would welcome a more open discussion. It sounds a lot like security though obscurity at the moment.</p>\r\n<p>Still, I'm happy the admins are taking it seriously and spending time on it.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15655,
      "author_name": "Sergey Yurgenson",
      "author_url": "",
      "post_date": "2012-10-15T22:31:24.400000",
      "content": "<p>[quote=Guy Cavet;15653]</p>\r\n<p>Hi Sergey,</p>\r\n<p>Public posts that discuss tactics for circumventing the rules and approaches for detecting sock-puppets may help the offenders become more sophisticated. I think it's best to keep those topics out of the forums.</p>\r\n<p>-Guy</p>\r\n<p>[/quote]</p>\r\n<p>&nbsp;</p>\r\n<p>I was not planning to discuss tactics. I just mentioned some obvious things. Anyway, I edited my post.</p>\r\n<p>P.S. Actually, I offered my assistance in private e-mail, but it was not accepted.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15652,
      "author_name": "Sergey Yurgenson",
      "author_url": "",
      "post_date": "2012-10-15T21:41:30.833000",
      "content": "<p>[quote=jcnhvnhck;15651]</p>\r\n<p>Please send your tips to compliance&#43;merck@kaggle.com and we'll look into any suspicious accounts.</p>\r\n<p>[/quote]</p>\r\n<p>Do I need to list all accounts, let say younger than 10 days? Maybe you can just apply filter and check them without&nbsp;my e-mail?</p>\r\n<p>(removed)</p>\r\n<p>By the way, what is the story with dajiangyou? He(?) jumped today to 2nd place, then after my complain his(?) last submission was removed&nbsp; and he(?) is back to #133 but not out.</p>\r\n<p>&nbsp;</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15650,
      "author_name": "alegro",
      "author_url": "",
      "post_date": "2012-10-15T21:16:25.407000",
      "content": "<p>[quote=Halla;15648]</p>\r\n<p>Changing the public private split from 25/75 to 1/99 could help in future competitions</p>\r\n<p>[/quote]</p>\r\n<p>Follow your logic removing the public leaderboard makes future competitions perfect ;)</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15649,
      "author_name": "alegro",
      "author_url": "",
      "post_date": "2012-10-15T21:14:10.167000",
      "content": "<p>May be better to fight with a reason - low amount of submissions per day. Look on the public hold-out set as on additional validation set and there will be no problem if anyone will have ability to use it enough amount of times.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15645,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-15T20:46:14.497000",
      "content": "<p>for all you know they might just overfit the leaderboard (Let's hope!)</p>\r\n<p>For the 284&#43; participants, is there no way Kaggle can find and weed out sock puppets? It is unfair on participants&nbsp; who have been honest with only 1 account - either increase number of entries for all, round off to 1 significant digit or weed out the sock\r\n puppets from the 284</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15628,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-15T18:13:03.413000",
      "content": "<p>the follow-up money stage must be a full-fledged competition with training and test. They ran one in Impermium where there was only a new dataset and I must say - it was one of the worst scoring datasets ever totally different from training and test</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15624,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-15T17:44:02.697000",
      "content": "<p>that could probably explain why there are 284 teams in the competition. Usually for such large datasets, there are fewer teams. I am surprised by the # of teams</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15618,
      "author_name": "Sergey Yurgenson",
      "author_url": "",
      "post_date": "2012-10-15T17:08:46.040000",
      "content": "<p>One, probably, can use provided RF benchmark and replace prediction only for one set.\r\n</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15293,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-08T17:08:13.813000",
      "content": "<p>the dataset is so huge that a RF will run for 6-7 hours for 15 essaysets. Wonderful ability indeed - is he using a supercomputer or something?</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15292,
      "author_name": "Dr Kiran R",
      "author_url": "",
      "post_date": "2012-10-08T17:07:09.427000",
      "content": "<p>Amazing! - Such people should get a special award as an incentive to disclose their methods :)</p>\r\n<p>Good catch Shea!</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15648,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-15T21:11:35.360000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15639,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-15T19:08:53.793000",
      "content": "",
      "votes": 1,
      "replies": []
    },
    {
      "id": 15626,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-15T17:56:53.013000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15615,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-15T16:50:09.857000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15287,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-08T16:49:20.777000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 15286,
      "author_name": "",
      "author_url": "",
      "post_date": "2012-10-08T16:46:58.110000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "15277": "",
    "15632": "",
    "15653": "",
    "15651": "",
    "15637": "",
    "15841": "",
    "15835": "",
    "15720": "",
    "15674": "",
    "15669": "",
    "15662": "",
    "15655": "",
    "15652": "",
    "15650": "",
    "15649": "",
    "15645": "",
    "15628": "",
    "15624": "",
    "15618": "",
    "15293": "",
    "15292": "",
    "15648": "",
    "15639": "",
    "15626": "",
    "15615": "",
    "15287": "",
    "15286": ""
  }
}