Writting patterns

When we write we produce trackable patterns. These patterns can be used to track individuals across platforms, and across identities. How does lain protect herself from being tracked in this manner online? Are there any software tools she uses to alter her sentances?

>>4733 I've been concerned about this for some time. The inherent problem language is how a sentence meaning and tone changes when replacing one word or more with several synonyms. >I don't know That's it, I do not know the answer to what you just asked >I'm not sure I may or may not know, I'm juggling a few things in my head. Linguistically, people from different areas of the world who use English as a second (or third, etc...) language make similar mistakes when they translate mentally before they speak. For example: German has no gerund phrases, "I eat" and "I am eating" are literally the same, but in meaning they are different. The conjugation is implied in its context. Germans speaking English (poorly) can still be understood even if they fuarrrk it up by saying "I am eat strawberries." Spanish has gendered nouns and uses prepositions in a slightly different way. It is not uncommon for me to hear to "put the lights in the Christmas tree" or "put more salt on the soup." One thing I've been wanting to study is the syntax and grammar of foreign languages to identify which mistakes are most commonly made so if I do pursue something like the OP said I can have inconsistencies in my grammar that will throw off the scent of where I'm from and subsequently who I am. Or I could just do it so badly and inconsistently that "they", in this hypothetical situation where they're resorting to stylometric analysis to ID me, can identify me immediately.

I don't do too much. If I notice I've been using a phrase much more than everyone else I try to rephrase what I just typed. I also try to avoid contributing to large datasets like Discord which tie a bunch of messages to a profile. Discord has the perfect dataset for pulling stuff like this off.

I HAVE THOUGHT OF THIS, AND I USE A VERY UNIQUE WRITING STYLE... one way of getting around it would be to use different meme dialects (like chan speak) or using broken phrasing like stereotypical accents alternatively, use machine learning to generate yor syntax for you, like with those "grammar and proofreading" services like grammarly (even though it is most certainly spyware)

I also have this problem. I am pretty sure you would be able to connect my lainchan post if you look trough them xd I just have a thought, maybe stupid one, maybe not. Let's say that we strip language from synonyms and words or constructions which are not obligatory. What's left is like a RISC VM (thing about it as a subset of language). Now if one want to write something undistinguishable he just writes it using this RISC VM set of word. Then it is being processed by some scripts which enrich this text with synonyms from full language.

>>4738 Lainchan should have a conlang like this.

>>4735 often have i used deliberate style markers, and other implications to suggest both different authorship and different location, but often my fear would be consistency for a single identity, and consistent difference for yet another. probably, with enough care it would be possible to manage that well, but i worry over my own competence and would rather something like >>4737's machine learning idea -- i wonder if there are corpora avaliable of texts (say on a sentance level broken), that one could use to develope some program just to pass things through to try to rework them. as far as pure obfuscation, it also occurs to me that one could write one, in let us say his native language, and pass it through machine translation to one or several languages before passing it to the target language, if he reads it well it can be edited before release. of course, then the service can see who i am, or at least my original words.

https://github.com/pagea/unstyle https://github.com/psal/anonymouth how do i use these?

>>4740 Very valid points. Using style markers like in a way that would include colloquialisms and other suggestive syntax would be really hard to maintain over a prolonged period of time hand jamming it yourself. I think if you could "proxy" text through a locally hosted machine-assisted translation service, you would be in an ideal position. That doesn't seem very cost-friendly to the average Jose. I'd thought about coming up with an idea similar to >>4738 While I think it'd be really cool to come up with a "code" like language, I think if someone is trying to prove you authored something, you're a suspect in the first place which means you stand a high chance of getting fuarrrked some other way. I mean really what can we do with language? Strip out superfluous langage Put in MORE superfluous language Use semantically similar sentences Idk, I'm interested in hearing more lainons thoughts.

>>4735 Learn and use language perfectly in terms of grammar and be as neutral as you can and you should be fine.

>>4733 >sentances caught!

>>4733 How about just throwing your writing through google translate (or some other translating service), to another language (or more languages) and then back to english then after doing this you correct the grammar mistakes. You could translate to different languages for different personalities resulting in different kinds of speaking.

>>4735 >Spanish has gendered nouns and uses prepositions in a slightly different way. in / on are confusing for spanish/ portuguese speakers because 'en/em' functions effectively for both respectively. You can definitely identify native languages through common mistakes. For instance, Chinese frequently destroy plurals because Chinese has near no inflection for plurals. >One thing I've been wanting to study is the syntax and grammar of foreign languages to identify which mistakes are most commonly made so if I do pursue something like the OP said I can have inconsistencies in my grammar that will throw off the scent of where I'm from and subsequently who I am. It depends on your threat model. I think there is little chance that people will correlate your style with your government identity. Exception to that obviously is uni-bomber levels of publicity. If your goal is to avoid creating a consistently traceable identity consistently making the same mistakes uniquely identifies you anyway. Its in interesting idea, but I think its only relevance would be toward writing a uni-bomber like manifesto and being identified as a Chinese female. But LE are skeptical enough of this kind of profile thing to do expend all resources chasing a Spanish immigrant because you substituted 'in' for 'on'. Also, I wonder how uniquely identifying each message could be. Presumably, over a large swath of people, many would writing similarly enough to you that uniquely identifying definite authorship even of a pseudonymous identity sounds difficult - specially for short messages. Without happexlegomenmon - like my mispelling of "hapax legomenon" , how sure can you be ?

>>4738 >What's left is like a RISC VM (thing about it as a subset of language). Now if one want to write something undistinguishable he just writes it using this RISC VM set of word. Then it is being processed by some scripts which enrich this text with synonyms from full language. >>4739 >Lainchan should have a conlang like this. esperanto Truely, I am skeptical of the utility. You sacrafice the beauty of language, and your self expression, possibly even the very quality of your thoughts by limiting yourself to a newspeak designed to be as unambiguous and unidentifying as possible. There is a huge cost to this kind of thing, a cost which could merit paying if your momentarily had need for such security. I am highly skeptical that slytometry would be useful for pervasive surveillance or even most targeted operations. For instance, I am skeptical law enforcement would value trying to identify a drug dealer by reading the text messages his clients receive from him. I think counter measures against stylometry might be useful for single purpose organizations facing a government threat. For instance, guerilla fighters might communicate in a restricted subset of language, using brief standardized communications to avoid providing intelligence on there numbers should communications be intercepted, or to disallow confirming deaths or presence of specific members at specific actions based upon style. Also, for long form manifesto like documents that might receive national attention and have consequences should the author be identified. This would involve sacrificing the expression of the manifesto to protect personal security. Without being involved directly in acting against a government, it seems to be a high cost to pay for little gain. When you have covered your eyes in tin foil to keep out the aliens, but blinded yourself, you have to ask, is it worth it ?

>>4733 >>6259 Romance langauge speakers frequently make the following mistakes: 1. Use to make instead of do because make/do are the same verb fazer / facere etc. >I need to make a question. 2. Use 'he / she ' personal pronouns instead of it. Frequently with the gender of the object in their own language. because often indefinite articles are used as pronouns. > me o da >Can you hand him to me 3. Confuse prepositions in / on

>>6254 I know someone who had an IRC bot that would do this.

>>6254 i was just going to suggest this. it makes you stick out in some ways, but i'm sure there are enough people using google translate to post in their non-native language that you wouldn't be too easily identifiable. more importantly, it erases your regional dialect.

>>6268 I think alot of that is highly speculative. Language is incredibly expressive, and rare words in english may map to specific enough words in spanish/taglog / etc. then back to english in a relatively identifiable way. Or altneratively a common word may be mangled such that its clear the google translate tactic is being applied. I think its far from aprori clear that google translate will mangle your identifying characteristics enough to prevent persistent identification. Also, no real threat model has been discussed for any of this. What are people actually trying to do ? 1. Avoid being distinguished from each other to thrawrt efforts to identify unique posters (without connection to other identies) 2. Avoid being identifyied anywhere to some characteristic ? (ex. poster is an imageboard user, young male) 3. Avoid document / post / writing being associated with govenment identity by government Because these things might be different. IE. google may mangle your writing in a identifying way indefitying you as a hyper-paranoid imageboard user. Posting like this , writing like this, etc may actually leak information about you. However, a single block of people trying not to be distinguished from each other, might find this uniformity of mangling useful. If all lains post this way, the number of lains to an outsider without access to server laws, or targeted serveilance might be hard to detect from textual analysis. What are people trying to prevent?

>>4733 How effectively do these tracking technologies handle code switching, people deliberately or sobconsciously changing the way they speak in different cultural contexts? Also, isn't there software (anonymouth?) to mitigate the risk of this by somehow mangling written text?

This only matters in a specific operation. If you're starting something, building a new identity etc... Then you have to worry about how you write, what passwords you use, etc... But on normal internet usage? This shouldn't be a concern. Tracking someone by their word pattern requires resources, it's not a source of income, so it's not done by default. It's a tool used when necessary.

The ability to track based on writing style is overrated. Lie detectors are more accurate. It would require a large corpus of material to make any connections and those would still be suspect. (Just stating the opposite of OP with just as little supporting evidence for what I said. Don't have an aneurysm if you disagree and have no evidence too. That bulging vein on your head is the point I'm making.)

>>6727 to expand on this: I don't know about you lainons, but I write very differently on imageboards than I do on other platforms. The way anyone writes changes based on format and audience. Only on sites like facebook is writing style recognition at all usable.

>>6728 Totally irrelevant. It's little unconscious quirks like how you spell "gray," common misspellings, etc. that will always give you away.

>>4742 >>6706 https://www.youtube.com/watch?v=-b0Ta9h62_E

Serial killers often wrote their letters in broken grammer and with spelling mistakes in order to thwart profilers. Could be wrong but BTK, Zodicac and Unabomber

>>7788 Conversely, cereal killers write their letters in Alpha Bits.

>>7788 >Unabomber He was caught because of his recognizable style. His brother read his manifesto, and that was it.

>>6259 wow, being a native spanish speaker I've never been able to catch on this (lol), but it's really an issue for me at least, and I consider my english to be almost native level what other things do you notice when speaking to a spanish speaker

>>5963 Or you can do the opposite, no? Be as varied as possible on purpose so that there's no pattern in your use of words/phrases.

>4738 >4739 this exists. it is called 'simple english'. the rule is to use only simple words and simple grammar. each sentence should say exactly one thing. each paragraph should say exactly one thought. one problem with simple english is that it takes longer to say what you mean. because of this, it is probably better to follow a popular style guide, and use only words that are have been common, and are still common, in the news, because the news is usually written so that almost everyone can understand it. because i like using fancy words, this post has been hard to write, but i think that it would be easier to use simple english if i use simple english often.

>>6734 His style change can potentially avoid/replace words that are quirks and thus fulfill the role of cover. >>7848 Could just mirror the style of who you reply to. Would give a wider range without losing any randomness.

Would it not be possible to make a script that intentionally and randomly replaces prepositions with each other, keeping the meaning identical but changing the writing style? Of course, something like this would work better with wider adoption since if only an individual were to use something like this it would immediately become apparent that the posts that are consistently getting prepositions and other such simple types of words wrong are made by one person.