People usually vote their stuff. You gather all the posts that been voted by alice. You use text analysis software to group those posts into buckets with the same author.
If you can do this across a large span of the network you get buckets of anonymous authors and voting patterns. Alice's bucket is the one where everything is upvoted(most likely).
That's one way. It's hard to counteract. Even if your software just counteracted this while using something like TOR to supply the rest of the anonimity, it would be pretty cool.
Another way:
Tap the internet. Gather a list of nodes connected to alice. Connect your nodes to all those nodes.
Watch whether message X is first delivered to your nodes from alice or from someone else. If it's first from alice, you got her. Statistically this will catch her at least some times.
Anyway,don't get discouraged. anonimity is hard. really hard.
Two very good points. As you said, Tor can handle the first, but the second is much like the 51% attack on Bitcoin. In any case, if you have a possible solution to the second, I would be glad to have a coffee with you.
The field of detecting authorship is called stylometry, and preventing that is called adversarial stylometry. Here's a java open source software that you could use :
BTW , another issue. If you really want to achieve strong anonimity you have to have large number of users, and large number of researchers trying to break your system.
The best way to build anonymous forums in this context , is to build popular anonymous email(currently there's none. only 2 we might see in the future "the dark mail alliance" and "pinchon gate") , and use it for forums.
People usually vote their stuff. You gather all the posts that been voted by alice. You use text analysis software to group those posts into buckets with the same author.
If you can do this across a large span of the network you get buckets of anonymous authors and voting patterns. Alice's bucket is the one where everything is upvoted(most likely).
That's one way. It's hard to counteract. Even if your software just counteracted this while using something like TOR to supply the rest of the anonimity, it would be pretty cool.
Another way:
Tap the internet. Gather a list of nodes connected to alice. Connect your nodes to all those nodes.
Watch whether message X is first delivered to your nodes from alice or from someone else. If it's first from alice, you got her. Statistically this will catch her at least some times.
Anyway,don't get discouraged. anonimity is hard. really hard.