cellio: (avatar)
[personal profile] cellio
Interesting. I tried out "TrustFlow", which assesses your "friends" list and their lists and so on and nominates people who are popular among your friends but not directly your friends. Their model is that friend = trusted party (not a safe bet with LJ, but what can you do?), and each person distributes "trust points" evenly among his friends, and after you run this a few levels out you start to get a picture. (Algorithm description.) (I found out about this from [livejournal.com profile] dragontdc.)

I recognize many of the names on my list, which isn't surprising. What is surprising is that I don't recognize the top scorer and we have only one friend in common. I'm curious about that.

Here are the results (with their canned text):

I tried out TrustFlow II for LiveJournal. The following people not on the friends list for [livejournal.com profile] cellio are close by:

Created by ciphergoth; hosted by LShift.

TrustFlow II: Who is closest to your friends list?

(no subject)

Date: 2006-03-29 04:03 am (UTC)
From: [identity profile] profane-stencil.livejournal.com
After two minutes of, "Machine too busy, please try again later," I got a result filled with people whose journals I read often, but would never add to my fl, and people who used to be on my list.

LJ itself has something similar called Popular with your Friends (http://www.livejournal.com/portal/).

(no subject)

Date: 2006-03-29 04:46 am (UTC)
From: [identity profile] profane-stencil.livejournal.com
It probably started slowing down when several of us read your post and rushed off to try it!

(no subject)

Date: 2006-03-29 04:16 am (UTC)
From: [personal profile] rectangularcat
landsknecht_po is a really shady character....
funny you have my sister (tootsiepop0204) and iamzaphod (meadhbh_9's husband) on there as I guess those are common friends just between meadhbh_9 and I which doesn't explain the lower levels...

(no subject)

Date: 2006-03-29 04:26 am (UTC)
kayre: (Default)
From: [personal profile] kayre
After looking at my results, I'm convinced they've got the explanation backwards; those at the bottom of the list are actually "closest". On the other hand, my top person is also someone with only a few friends, one of whom we have in common, so that may be overwhelming all other data?

(no subject)

Date: 2006-03-29 06:26 am (UTC)
From: [identity profile] http://users.livejournal.com/merle_/
I'm not sure about that, because I recognize (and have traded comments with) the people near the top (smaller numbers) of my list, but do not even recognize the people at the bottom.

But it could easily depend on the person. Or the algorithm may be flawed...

Expand Cut Tags

No cut tags