Software, technology, sysadmin war stories, and more. Feed
Sunday, December 30, 2012

Secret forums: you don't know why you're there

When you go to a forum today, you probably have a pretty good idea of what sort of people who will be hanging out there. It might be a bunch of startup-minded entrepreneurs, gamers, knitters, or something else entirely. These tend to clump around forums which spring up for the express purpose of discussing a given topic.

My half-baked idea for this afternoon is to have "secret forums". Everyone who wanted access to this system would first go through some kind of analysis process where all sorts of things are determined. They'd figure out what sort of car you drive (if any), where you like to go on vacation, your favorite movies, and your favorite foods. Then they'd work out all sorts of other interesting things that you might convey without even realizing it.

Remember The Game? Without going into too many details (in case anyone reading hasn't seen a 15 year old movie), they used some interesting techniques to learn things about someone's background, and then use that data to craft a new environment.

Imagine that sort of technique applied to people who want to join a forum. It would try to find the things they actually enjoy, rather than the ones they just publicly talk about. Think of it as targeting the subconscious instead of whatever's on the surface.

Once an applicant had been analyzed, they'd be auto-joined into a number of forums with opaque identifiers. For example, there might be a group for people who live in New England and make at least $200,000 a year. Another group might be for people who study black and white sci-fi movies from the '50s. The trick is that only the forum administrators would know that. Everyone else would just see that they are a member of one or more forums with a bunch of other people who also have no idea how they were selected.

To keep things interesting, I would suggest some "polite requests" which encourage forum members to not try to figure out whatever common elements brought them together. Alternatively, it might be the sort of thing which could be confined to special threads within a forum, and members who don't want their illusion spoiled could just avoid such posts.

Imagine how it might work. You sign up and find yourself as a member of 6 nondescript communities. You write a short message to say hello and push it out there. Maybe you see some other threads which are going on and start posting to them.

Would it work? I don't know. I'm mostly interested in it as an experiment since I have never heard of anyone doing something like that before. I'd want to see if the groups naturally converged on discussions of the actual thing which caused them to be selected. It might also be interesting to see if they converge on another topic and then see if there is some link from that back to the forum's "seed".

Maybe you'd learn some things, like maybe people who live in New England and make more than $200,000 a year eat a lot of lobster or frequently take helicopter shuttle trips to get around. I'm just making stuff up now, but you get the idea.

This half-baked idea was born in my notebook of random ideas and other thoughts as just a simple notion: "secret forums where nobody knows what the parameters are". It wasn't intended to connect to anything in the real world. However, as I was writing this post, something occurred to me: a certain giant advertising company which masquerades as a search engine probably already knows all of this stuff about many, if not most, of its users. It might even already have them sorted into neat little groups. I imagine they're not going to go around telling people which groups they are in, especially for anything which might "hit too close to home".

Oddly enough, they could probably create this system without too much work simply by creating one forum for each significant group and auto-joining people to it. The hard part is doing the analysis, and it's likely that data already exists in massive quantities.

For their users, the analysis already happened a long time ago.