Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated: 1 00:00:00,780 --> 00:00:07,050 Based on the way you write your spelling grammar punctuation and general writing style can be used to 2 00:00:07,050 --> 00:00:08,150 identify you. 3 00:00:08,160 --> 00:00:15,060 This is called style amatory which is a statistical analysis of variations in a literary style between 4 00:00:15,060 --> 00:00:18,130 one writer or genre and another recession. 5 00:00:18,150 --> 00:00:24,360 Yes if someone has a sample of around say five thousand to six thousand five hundred words of text you 6 00:00:24,360 --> 00:00:30,180 have written they can confirm within about 80 percent accuracy what all the text you might have written 7 00:00:30,180 --> 00:00:30,250 . 8 00:00:30,300 --> 00:00:36,300 This means if you have an alias it could be potentially linked to your real identity if you write in 9 00:00:36,300 --> 00:00:41,550 the same style and English is also easier to analyze at the moment because more tools are available 10 00:00:41,550 --> 00:00:44,750 to do it and more research has been put into English. 11 00:00:44,760 --> 00:00:49,290 Let me give you some examples as I'm available tools that you could potentially play with so there's 12 00:00:49,290 --> 00:00:54,770 JSAN which is an Open-Source writing style analysis and anonymising framework. 13 00:00:54,790 --> 00:01:01,900 It consists of two parts the J.S. Stylo tour which is the Authorship Attribution framework. 14 00:01:01,920 --> 00:01:07,530 So as the name suggests it helps connect authors to writings and then you've got a nano mounth which 15 00:01:07,530 --> 00:01:11,980 is the authorship evasion framework which is this one here. 16 00:01:12,120 --> 00:01:18,720 You also have things like signiture which is another Authorship Attribution tool and here's another 17 00:01:18,740 --> 00:01:23,220 anti-establishment tool I'm not sure how effective it is it's pretty simple. 18 00:01:23,220 --> 00:01:26,840 I'm just going to paste in some writing here. 19 00:01:27,030 --> 00:01:29,880 This is Isaac Asimov's writing style. 20 00:01:30,070 --> 00:01:34,150 Let's see what we get from it. 21 00:01:34,350 --> 00:01:38,120 And there you see the input from I Robot. 22 00:01:39,060 --> 00:01:41,030 And then we can see the output. 23 00:01:41,310 --> 00:01:48,720 You can see it's capsulized beginning of words and potentially some of the things I can't say at the 24 00:01:48,720 --> 00:01:49,250 moment. 25 00:01:49,320 --> 00:01:55,050 So that's an antique style almost free tool all nation states are sure to be working on this capability 26 00:01:55,290 --> 00:01:57,830 although I haven't heard much on it yet. 27 00:01:57,850 --> 00:02:04,430 Now that means that nation states aren't experts at it already or at least some may be experts at it 28 00:02:04,430 --> 00:02:05,160 already. 29 00:02:05,200 --> 00:02:09,740 There's an interesting video here if you want to understand more of the detail Einstein a tree. 30 00:02:09,750 --> 00:02:13,450 These guys are behind the Jason anonymization framework. 31 00:02:13,650 --> 00:02:19,720 And if this is something that interests you his report on the feasibility of Internet scale authorship 32 00:02:19,780 --> 00:02:27,690 identification quite a scary prospect if an attacker can combine stylometry a tree with all the OPSEC 33 00:02:27,690 --> 00:02:32,790 failures and profile leaks attrib. is going to be much easier for them. 34 00:02:32,790 --> 00:02:39,010 There are also mitigations that you can consider against the automated recognition tools. 35 00:02:39,030 --> 00:02:42,710 The first is use authorship recognition evasion tools. 36 00:02:42,720 --> 00:02:49,490 Next consider using multiple aliases because this could make attrib. back to your real identity harder 37 00:02:49,630 --> 00:02:50,290 generally. 38 00:02:50,310 --> 00:02:53,010 Maybe always the less you write is better. 39 00:02:53,010 --> 00:02:59,820 If you can post texts from other sources that you haven't written if you can try to impersonate other 40 00:02:59,820 --> 00:03:05,740 users styles you can include extra data in your posts and writings. 41 00:03:05,820 --> 00:03:09,320 For example a list of objects or a poem. 42 00:03:09,330 --> 00:03:15,340 This makes automation the fourth attribution harder because you are into mixing different styles. 43 00:03:15,360 --> 00:03:22,560 If everyone uses leetspeak this makes identification harder leetspeak if you're not aware is converting 44 00:03:22,560 --> 00:03:33,270 your letters into things like this for Leite server to put in Isaac Asimov's writing into here and code 45 00:03:33,520 --> 00:03:36,690 with. 46 00:03:36,790 --> 00:03:44,100 Basically it is going to look something like that but obviously everybody has to be using Leite among 47 00:03:44,100 --> 00:03:46,410 the people that you're trying to hide within. 48 00:03:46,410 --> 00:03:51,480 Otherwise you'll stand out because you're using leetspeak anonymous programmers can be identified by 49 00:03:51,480 --> 00:03:53,190 analyzing coding style. 50 00:03:53,200 --> 00:03:59,940 So if you're writing code consider this and this here is a link to the hidden wiki which provides more 51 00:03:59,940 --> 00:04:04,440 information on anonymous writing styles. 5694