I believe that the data provided for this competition can not only be used to improve the detection of potential threats. It can also be used by terrorists to find better ways to conceal their weapons.\nIn my opinion, it would be naive to think that major terrorist organizations such as the IS and Al-Qaeda do not take such methods into account. They have lots of academic supporters and for sure they would find data scientists who can work on this problem.
\n\nEven though the true nature of the objects is not revealed, it is implicitly disclosed a lot about the shape, characteristics and positions of the objects Homeland Security is looking for. If terrorists get hands on such a scanner and are able to generate these kind of pictures themselves, the problem gets even worse.
\n\nMoreover, a lot of stuff on kaggle ends up being public. Even if the winning solution itself is not shared in the end, chances are that there are only minor differences to what is publicly accessible. If the winning solution is deployed in practice, this would mean that an adversary would have an algorithm whose predictions are already pretty close to the ones used in the live system. That enables him to investigate the false negatives of the algorithm, which could guide him to better ways of concealment. Furthermore, new samples in the form of concealing candidates could be generated and their efficacy tested by making use of this algorithm.
\n\nJudging from previous competitions, also leaks are quite common and that raises another potential problem. If the data was not checked properly, even internal security protocols might have been leaked somehow. I don't know about the details, but those issues and the potential dangers concern me. I hope that the guys from Homeland Security have taken such aspects into consideration when they decided to launch this challenge.
", "messages": [ { "id": 197577, "postDate": "2017-06-29T21:37:35.280Z", "content": "I believe that the data provided for this competition can not only be used to improve the detection of potential threats. It can also be used by terrorists to find better ways to conceal their weapons.\nIn my opinion, it would be naive to think that major terrorist organizations such as the IS and Al-Qaeda do not take such methods into account. They have lots of academic supporters and for sure they would find data scientists who can work on this problem.
\n\nEven though the true nature of the objects is not revealed, it is implicitly disclosed a lot about the shape, characteristics and positions of the objects Homeland Security is looking for. If terrorists get hands on such a scanner and are able to generate these kind of pictures themselves, the problem gets even worse.
\n\nMoreover, a lot of stuff on kaggle ends up being public. Even if the winning solution itself is not shared in the end, chances are that there are only minor differences to what is publicly accessible. If the winning solution is deployed in practice, this would mean that an adversary would have an algorithm whose predictions are already pretty close to the ones used in the live system. That enables him to investigate the false negatives of the algorithm, which could guide him to better ways of concealment. Furthermore, new samples in the form of concealing candidates could be generated and their efficacy tested by making use of this algorithm.
\n\nJudging from previous competitions, also leaks are quite common and that raises another potential problem. If the data was not checked properly, even internal security protocols might have been leaked somehow. I don't know about the details, but those issues and the potential dangers concern me. I hope that the guys from Homeland Security have taken such aspects into consideration when they decided to launch this challenge.
", "rawMarkdown": "I believe that the data provided for this competition can not only be used to improve the detection of potential threats. It can also be used by terrorists to find better ways to conceal their weapons.\nIn my opinion, it would be naive to think that major terrorist organizations such as the IS and Al-Qaeda do not take such methods into account. They have lots of academic supporters and for sure they would find data scientists who can work on this problem.\n\nEven though the true nature of the objects is not revealed, it is implicitly disclosed a lot about the shape, characteristics and positions of the objects Homeland Security is looking for. If terrorists get hands on such a scanner and are able to generate these kind of pictures themselves, the problem gets even worse.\n\nMoreover, a lot of stuff on kaggle ends up being public. Even if the winning solution itself is not shared in the end, chances are that there are only minor differences to what is publicly accessible. If the winning solution is deployed in practice, this would mean that an adversary would have an algorithm whose predictions are already pretty close to the ones used in the live system. That enables him to investigate the false negatives of the algorithm, which could guide him to better ways of concealment. Furthermore, new samples in the form of concealing candidates could be generated and their efficacy tested by making use of this algorithm.\n\nJudging from previous competitions, also leaks are quite common and that raises another potential problem. If the data was not checked properly, even internal security protocols might have been leaked somehow. I don't know about the details, but those issues and the potential dangers concern me. I hope that the guys from Homeland Security have taken such aspects into consideration when they decided to launch this challenge.", "votes": 6 }, { "id": 197916, "postDate": "2017-06-30T14:41:52.637Z", "content": "\n\n\nThey have lots of academic supporters and for sure they would find data scientists who can work on this problem.
\n
I suppose this statement is a ground truth remark and of course no need to dig in various and irrefutable sources to prove it ?
", "rawMarkdown": "> **They have lots of academic supporters** and for sure they would find data scientists who can work on this problem.\n\nI suppose this statement is a ground truth remark and of course no need to dig in various and irrefutable sources to prove it ? ", "votes": 1, "replies": [ { "id": 197972, "postDate": "2017-06-30T17:28:50.070Z", "content": "There are many sources available, here for instance:
\n\n\n\nSo IS recruits are more educated than average and 25% of them have university education. This is based on the data from World Bank, so it should be reliable.
\n\nMoreover, you can do a quick background check on the 9/11 attackers:
\n\nhttps://en.wikipedia.org/wiki/Hamburg_cell
\n\nYou will find that almost all of them graduated from university.
\n\nMight be surprising at first, but it's actually quite plausible. Attacks like 9/11 as well as running a whole state for years would have not been possible without smart minds planning and conducting everything. It is shocking that so many well-educated people decide for such an ideology. But it is crucial to take this into consideration, otherwise you will underestimate your enemy.
", "rawMarkdown": "There are many sources available, here for instance:\n\nhttp://www.telegraph.co.uk/news/2016/10/06/isil-recruits-better-educated-than-their-average-countryman-worl/\n\nSo IS recruits are more educated than average and 25% of them have university education. This is based on the data from World Bank, so it should be reliable.\n\nMoreover, you can do a quick background check on the 9/11 attackers:\n\nhttps://en.wikipedia.org/wiki/Hamburg_cell\n\nYou will find that almost all of them graduated from university.\n\nMight be surprising at first, but it's actually quite plausible. Attacks like 9/11 as well as running a whole state for years would have not been possible without smart minds planning and conducting everything. It is shocking that so many well-educated people decide for such an ideology. But it is crucial to take this into consideration, otherwise you will underestimate your enemy." }, { "id": 197978, "postDate": "2017-06-30T17:34:09.133Z", "content": "of course they have some people from academia in the whole pack. They are crazy, not stupid
", "rawMarkdown": "of course they have some people from academia in the whole pack. They are crazy, not stupid" } ] }, { "id": 198876, "postDate": "2017-07-03T23:58:25.387Z", "content": "Sounds analogous to the \"uncertainty principle\" in that the querying the data affects the data.
", "rawMarkdown": "Sounds analogous to the \"uncertainty principle\" in that the querying the data affects the data.", "votes": 1 }, { "id": 199162, "postDate": "2017-07-04T19:31:41.227Z", "content": "I think it's a tradeoff they've thought about, and probably decided that having (potentially) hundreds of the best data scientists in the world working on this problem was worth more than a terrorist group having access to the data.
", "rawMarkdown": "I think it's a tradeoff they've thought about, and probably decided that having (potentially) hundreds of the best data scientists in the world working on this problem was worth more than a terrorist group having access to the data." }, { "id": 197980, "postDate": "2017-06-30T17:35:12.127Z", "content": "I agree on that. There is probably ways to find which are the materials in question here and avoid them
", "rawMarkdown": "I agree on that. There is probably ways to find which are the materials in question here and avoid them" }, { "id": 199596, "postDate": "2017-07-05T23:40:49.003Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 197916, "author_name": "Serigne ", "author_url": "", "post_date": "2017-06-30T14:41:52.637000", "content": "\n\n\nThey have lots of academic supporters and for sure they would find data scientists who can work on this problem.
\n
I suppose this statement is a ground truth remark and of course no need to dig in various and irrefutable sources to prove it ?
", "votes": 1, "replies": [ { "id": 197972, "author_name": "JStr", "author_url": "", "post_date": "2017-06-30T17:28:50.070000", "content": "There are many sources available, here for instance:
\n\n\n\nSo IS recruits are more educated than average and 25% of them have university education. This is based on the data from World Bank, so it should be reliable.
\n\nMoreover, you can do a quick background check on the 9/11 attackers:
\n\nhttps://en.wikipedia.org/wiki/Hamburg_cell
\n\nYou will find that almost all of them graduated from university.
\n\nMight be surprising at first, but it's actually quite plausible. Attacks like 9/11 as well as running a whole state for years would have not been possible without smart minds planning and conducting everything. It is shocking that so many well-educated people decide for such an ideology. But it is crucial to take this into consideration, otherwise you will underestimate your enemy.
", "votes": 0, "replies": [] }, { "id": 197978, "author_name": "Charles Jansen", "author_url": "", "post_date": "2017-06-30T17:34:09.133000", "content": "of course they have some people from academia in the whole pack. They are crazy, not stupid
", "votes": 0, "replies": [] } ] }, { "id": 198876, "author_name": "Paolo Pignatelli", "author_url": "", "post_date": "2017-07-03T23:58:25.387000", "content": "Sounds analogous to the \"uncertainty principle\" in that the querying the data affects the data.
", "votes": 1, "replies": [] }, { "id": 199162, "author_name": "moefasa", "author_url": "", "post_date": "2017-07-04T19:31:41.227000", "content": "I think it's a tradeoff they've thought about, and probably decided that having (potentially) hundreds of the best data scientists in the world working on this problem was worth more than a terrorist group having access to the data.
", "votes": 0, "replies": [] }, { "id": 197980, "author_name": "Charles Jansen", "author_url": "", "post_date": "2017-06-30T17:35:12.127000", "content": "I agree on that. There is probably ways to find which are the materials in question here and avoid them
", "votes": 0, "replies": [] }, { "id": 199596, "author_name": "", "author_url": "", "post_date": "2017-07-05T23:40:49.003000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "197577": "I believe that the data provided for this competition can not only be used to improve the detection of potential threats. It can also be used by terrorists to find better ways to conceal their weapons.\nIn my opinion, it would be naive to think that major terrorist organizations such as the IS and Al-Qaeda do not take such methods into account. They have lots of academic supporters and for sure they would find data scientists who can work on this problem.\n\nEven though the true nature of the objects is not revealed, it is implicitly disclosed a lot about the shape, characteristics and positions of the objects Homeland Security is looking for. If terrorists get hands on such a scanner and are able to generate these kind of pictures themselves, the problem gets even worse.\n\nMoreover, a lot of stuff on kaggle ends up being public. Even if the winning solution itself is not shared in the end, chances are that there are only minor differences to what is publicly accessible. If the winning solution is deployed in practice, this would mean that an adversary would have an algorithm whose predictions are already pretty close to the ones used in the live system. That enables him to investigate the false negatives of the algorithm, which could guide him to better ways of concealment. Furthermore, new samples in the form of concealing candidates could be generated and their efficacy tested by making use of this algorithm.\n\nJudging from previous competitions, also leaks are quite common and that raises another potential problem. If the data was not checked properly, even internal security protocols might have been leaked somehow. I don't know about the details, but those issues and the potential dangers concern me. I hope that the guys from Homeland Security have taken such aspects into consideration when they decided to launch this challenge.", "197916": "> **They have lots of academic supporters** and for sure they would find data scientists who can work on this problem.\n\nI suppose this statement is a ground truth remark and of course no need to dig in various and irrefutable sources to prove it ? ", "198876": "Sounds analogous to the \"uncertainty principle\" in that the querying the data affects the data.", "199162": "I think it's a tradeoff they've thought about, and probably decided that having (potentially) hundreds of the best data scientists in the world working on this problem was worth more than a terrorist group having access to the data.", "197980": "I agree on that. There is probably ways to find which are the materials in question here and avoid them", "199596": "" } }