pgpckt's Journal: Friend/Foe System...How about a web of trust?

This was a spur of the moment idea, but it sounded like it could be useful on slashdot, so feel free to share your opinion.

I like the friends/foe system, because it gives me the chance to read some of the fine journal entries some of you put out. But I was wondering...there are half a million slashdot users, how can I find people's journals I like?

Right now, it is hit-or-miss. I see a comment I think is really good, and add that person as my friend. But what if I think they are REALLY insightful? What about THEIR friends? Hey, if I like the style of writing of one guy, and he likes some other guys, maybe I will like them too.

So, how about a web of trust type system? Journals from people who are my friends would gain a point in my fan's account (see, my fans are trusting my friends). Now, for any given user, take all their friends, find all their fans, and give them a point. Users can sort by most points, and thereby would be exposed to users that are like-minded. This process could also be done recursively, perhaps to a certain number of levels. Also, you could use the foe/freak system to automatically downgrade certain users.

If this could be implemented, the result would be cliques. People would start to segment into groups, which could be a good thing. You could probably map these groups, and you could also find out which user in any group was the most popular. So, a new slashdotter would be able to tell the flavor of the group by the writings of its most popular members, and decide if they want to join.

I am proposing an automated faction system using the foe/friend/freak/fan designation to automatically correlate people. Of course, it is all optional. You could stay independent. You could manually choose people you wanted. This is sort of like Tivo for slashdot. Slashdot would merely suggest other users you would find interesting based on the information collected thus far. I think the idea is intriguing.

Well, it probably would never happen because it would be so hard to code and implement, but I thought it was a fun idea. After all, who knows what other users are out there that I might like to read?

interesting...but

[Em Emalb]

Would be hard as hell to implement. Usually what I do is actually look at the comments in the articles. If they are insightful and/or interesting, I check and see if they have a journal. If they do, then I can look and see who is posting comments in the journal, and if they have a journal. Some people, myself included, have posted "lists" of people they consider interesting. If more people did this, it would be easier.

Not the easiest way to do things, but it seems to be the best way so far.

cool idea though

Not hard to implement

[gmhowell]

I actually don't think it would be that hard to implement. I do think it would be hard to implement well and in a way that scaled. Nested SELECT statements would get the info in a jiffy.

What would make the friend/foe system MUCH easier to use would be to make links, so that you didn't have to type in the freakin' URL.

I would also argue that there are already cliques. Scan your own friends list, and see how many people are fans. Check theirs out, etc. When I open up the comment level (I usually browse at 3+), I ALWAYS see the same folks posting in my friends' journals whose journals I also read. As an example, I see FortKnox, banky, and TechnoLust posting in many of the journals I read.

Probably not because it's hard to code...

[X86Daddy]

It wouldn't necessarily be the difficulty of code. It actually sounds rather straightforward.

However, the scalability and efficiency of this system would be a lot less than the current friend / foe. Enough of a performance hit per operation, and a site with a huge user base like Slashdot crumbles.

Think of it this way (I have not examined Slash, I'm just hypothesizing):

• currently there is a lookup on everyone's comments to see if they appear in your friend / foe list
• new Journal entries trigger some sort of notification to friends or are looked up when you log in.
• You'd have to do x more lookups/notifications to find new journal entries, and x more lookups on every comment displayed on every page.... where x is some web depth you've set.

You basically go from O(n) O(n^2).

Slash may be up to it; I don't know, but it sounds like increasing O a step... All that aside, I really like the idea. I've been thinking about how Webs of Trust could also help destroy the music cartels by providing the recommendation engine that is so lacking in current P2P apps.