You'd first need a set of master data with all possible 6 word snippets of text from all the existing posts. (provided someone is copying only from existing Bitcoin posts). This would then have to be compared with the set of snippets formed from every new post. While this could be done, I believe the space and memory requirements would be pretty huge. Though, doesn't google do it for like, all of the internet? And Altavista used to do it at one time. Now, google has humungous capacity of course but I don't think that the old sites like Altavista had those.
There are clever indexing methods that make this kind of search relatively quick and also can match slight variations, even word spinning to an extent.
I'm not really sure what you're proposing (checking your posts against all other posts? why exactly?) but the plagiarism problem in general is not a technical one. We can use all sorts of tricks to catch plagiarists and they'll just make their posts more and more obscure - copying from outside sources, translating from other languages, etc - as long as there is a financial incentive to do so.