Paul Crowley (

ciphergoth) wrote in

trustmetrics,
@ 2006-03-28 14:01:00

TrustFlow frequently asked questions

TrustFlow for LiveJournal

Frequently asked questions

What do the results mean?: TrustFlow is making a guess at who is "near" your friends list; who might be on it, but isn't. It does this by looking at your friends list, and the friends list of your friends, and so on.
Is this based on who reads my journal, or interests, or what?: No. TrustFlow looks only at who is on whose friends list to make the determination; no other information is taken into account. In particular, it doesn't know anything about whose journals you are actually reading, or who is reading your journal, except what friends lists tell you.
Are the people listed in any order?: Yes; the numbers are a measure of "distance", so the first person listed is "closest".
How exactly does it determine who to list? What do the numbers mean?: A description of the algorithm appeared earlier in this journal.
I get an error!: Please read the error carefully before telling me about it, and please quote the error exactly - otherwise how can I do anything about it?
I've changed my friends list, but the results haven't changed: When it fetches friends data, it keeps that data for around 24 hours; if you wait about that long after a change you'll see that change reflected in the results.
It doesn't work at all - it lists my worst enemy first!: That means it's working. That person is someone quite close to your circle of friends, who you would list except that you've deliberately decided not to. It can't tell that you don't like them; all it can tell is that they are close enough to your circle of friends that you would list them if you didn't feel that way.

More will be added here as the questions are asked...

Page 1 of 2
<<	[1] [2]	>>

(Post a new comment)

	immorak 2006-03-28 01:28 pm UTC (link)
	What does it really have to do with trust? Doesn't it only list the friends of your friends? (Reply to this) (Thread)

	ciphergoth 2006-03-28 01:38 pm UTC (link)
	It's designed to be a trust metric; see the description in the community. If LJ friending indicates trust then TrustFlow finds users who you might consider conferring small amounts of trust to. It doesn't quite, of course, which skews the results. Under some circumstances it will look beyond the friends of your friends to their friends and so on. (Reply to this) (Parent)

(no subject) -

tommytesto, 2006-03-30 03:27 am UTC

(no subject) -

brokenbeast, 2006-03-30 09:19 am UTC

(no subject) -

brokenbeast, 2006-03-30 09:20 am UTC

	rosefox 2006-03-28 02:16 pm UTC (link)
	Does it break for me because my flist is too big? *) (Reply to this) (Thread)

	ciphergoth 2006-03-28 02:19 pm UTC (link)
	What error do you get? I just tried as you and it seems to work so far... (Reply to this) (Parent)(Thread)

(no subject) -

rosefox, 2006-03-28 02:44 pm UTC

(no subject) -

cest_la_vie, 2006-11-07 05:46 am UTC

(no subject) -

ciphergoth, 2006-03-28 02:23 pm UTC

	(Anonymous) 2006-03-28 02:21 pm UTC (link)
	fast flying thing whipping across the screen ... (Reply to this) (Thread)

	ciphergoth 2006-03-28 02:23 pm UTC (link)
	A better animated graphic would be quite welcome! (Reply to this) (Parent)(Thread)

progress indicator -

ablueskyboy, 2006-03-28 10:14 pm UTC

Re: progress indicator -

sewcute, 2006-03-29 04:09 am UTC

Re: progress indicator -

ablueskyboy, 2006-03-29 03:58 pm UTC

Re: progress indicator -

ablueskyboy, 2006-03-29 04:23 pm UTC

(no subject) -

adudeabides, 2006-03-29 04:37 am UTC

dennyd
2006-03-28 02:52 pm UTC (link)

There seems to be a strong bias towards listing people who are (friends of (my friends with small friends lists)) high up in the results, which doesn't seem right to me... are you postulating that the trust of those without many friends is more valuable? If so, do you think this factor should outweigh the (more distributed/diluted) trust of many friends to the extent that it currently seems to do?

(Reply to this) (Thread)

ciphergoth
2006-03-28 03:13 pm UTC (link)

Yes, that's right. It's hard to know exactly how to modify the algorithm to reduce that bias; I can't remove it altogether because that destroys the attack resistance, so it's a question of introducing a fudge factor that cuts in when your friends list is short. The tricky think is making sure the fudge factor doesn't introduce a different weird behaviour.

Thinking about it just now, I think I have an idea (extending dead end detection), but it would take a little while to implement it...

(Reply to this) (Parent)(Thread)

(no subject) -

amberdulen, 2006-03-28 05:06 pm UTC

(no subject) -

ciphergoth, 2006-03-28 05:09 pm UTC

How long I had my journal -

ablueskyboy, 2006-03-28 10:17 pm UTC

Re: How long I had my journal -

ciphergoth, 2006-03-28 10:36 pm UTC

(no subject) -

wbm, 2006-05-05 04:14 am UTC

	figg 2006-03-28 07:04 pm UTC (link)
	Is the referer information used for anything? (Reply to this) (Thread)

	ciphergoth 2006-03-28 10:44 pm UTC (link)
	To pre-fill the form with a guessed name; that's it. (Reply to this) (Parent)(Thread)

(no subject) -

figg, 2006-03-28 10:48 pm UTC

(no subject) -

ciphergoth, 2006-03-28 11:03 pm UTC

(no subject) -

skx, 2006-03-29 11:06 am UTC

Off topic icon worship -

xoder, 2006-03-30 07:54 pm UTC

Re: Off topic icon worship -

figg, 2006-03-30 09:14 pm UTC

Re: Off topic icon worship -

xoder, 2006-03-30 09:42 pm UTC

	cyberglamour 2006-03-28 07:50 pm UTC (link)
	I don't get it.I just don't get the whole point or meaning of it. (Reply to this) (Thread)

	raventhon 2006-03-28 08:41 pm UTC (link)
	It lists people that are near you in lj-space. Basically, people friended by a lot of your friends but not you. (Reply to this) (Parent)(Thread)

(no subject) -

cyberglamour, 2006-03-28 09:09 pm UTC

(no subject) -

ciphergoth, 2006-03-28 10:42 pm UTC

(no subject) -

qryztufre, 2006-03-29 08:37 pm UTC

(no subject) -

illusi0nist, 2006-04-05 11:58 pm UTC

(no subject) -

raventhon, 2006-04-06 01:16 am UTC

thank you -

mousee23, 2007-04-04 05:52 am UTC

	daemonnoire 2006-03-28 08:28 pm UTC (link)
	It's funny... The two people I trust the least came up first. I don't know what that says about my friends. (Reply to this) (Thread)

	emberleo 2006-03-31 03:31 am UTC (link)
	Well, there may well be total strangers you'd trust even less if you were exposed to them as often. So really, it just says those two people have, via mutual friends, had more opportunities to lose your trust. --Ember-- (Reply to this) (Parent)

	flummoxicated 2006-03-28 09:22 pm UTC (link)
	I was a bit startled when the results popped up in my "create new entry" page; I do tend to be a Internet safety freak though. Wasn't sure how you were able to do that. (Reply to this) (Thread)

ciphergoth
2006-03-28 10:48 pm UTC (link)

As it happens, you don't have to worry - I invented the mechanism that means your LJ can't be attacked that way. There was a time when such a button could not only have created such an entry, but posted it too. With JavaScript to press the button automatically, this resulted in a relatively harmless but annoying attack on LJ called the "sausage attack". For more, see here:

http://community.livejournal.com/lj_dev/641972.html

With my defence in place, I can pre-fill the "create new entry" page, but I can't make you post it. So it's just a convenience to you.

You are wise to be an internet safety freak.

(Reply to this) (Parent)

	elkor 2006-03-28 09:27 pm UTC (link)
	You might want to make it not-case sensitive. (Reply to this) (Thread)

	ciphergoth 2006-03-28 10:36 pm UTC (link)
	Good idea - thanks! Will do tomorrow. (Reply to this) (Parent)

(no subject) -

ciphergoth, 2006-03-29 09:16 am UTC

mauser
2006-03-29 03:09 am UTC (link)

One issue with the "Create New Entry" page is that apparently it's time zone sensitive. After I posted the meme, I went to make another entry and my LJ Client pointed out to me that my latest entry had a newer timestamp than the entry I was trying to post, and that I hadn't set the backdate checkbox. Lo and Behold, the output from the meme was hours ahead for me (in the Pacific time zone).

(I'd still like to find a mechanism out there to try to find people I don't know yet with the greatest number of common interests).0

(Reply to this) (Thread)

	ciphergoth 2006-03-29 08:06 am UTC (link)
	whooops! The correct fix is to fill out those fields with JavaScript. like the usual "post new entry" form does. That way I get your local time. I'll sort it out today. (Reply to this) (Parent)

(no subject) -

silverrose, 2006-03-31 05:30 pm UTC

(no subject) -

mauser, 2006-03-31 07:56 pm UTC

(no subject) -

silverrose, 2006-03-31 08:01 pm UTC

(no subject) -

metallic_ah, 2006-04-08 03:26 pm UTC

(no subject) -

silverrose, 2006-04-08 03:48 pm UTC

(no subject) -

mauser, 2006-04-16 04:52 am UTC

(no subject) -

ciphergoth, 2006-04-16 08:05 am UTC

(no subject) -

mauser, 2006-04-16 08:25 am UTC

(no subject) -

ciphergoth, 2006-04-16 08:37 am UTC

(no subject) -

mauser, 2006-04-16 08:43 am UTC

(no subject) -

silverrose, 2006-04-16 10:41 pm UTC

	ocicat 2006-03-29 03:23 am UTC (link)
	I notice that none of the people who have friended me (but who I haven't friended back) show up. Does this mean they are counted as "trusted" by me? Because, you know, they're not. They are friended by some of my friends, so I would expect them to be listed, but it's not the case. (Reply to this) (Thread)

	ocicat 2006-03-29 03:34 am UTC (link)
	I take that back... there is one on there. I guess most my friends just haven't friended my stalkers back! (Reply to this) (Parent)(Thread)

(no subject) -

sewcute, 2006-03-29 04:11 am UTC

	kunzite1 2006-03-29 04:22 am UTC (link)
	could you explain the numbers more? they don't seem to be number of hops or anything obvious like that. (Reply to this) (Thread)

	ciphergoth 2006-03-29 08:46 am UTC (link)
	You'll have to read the algorithm, I'm afraid; I can't do a simpler explanation. (Reply to this) (Parent)(Thread)

(no subject) -

emberleo, 2006-03-31 03:34 am UTC

	radiantsun 2006-03-29 05:07 am UTC (link)
	Someone whom I absolutley do not trust is on the trust flow list, which I find amusing, but may be deceiving for some of my friends. (Reply to this) (Thread)

	ciphergoth 2006-03-29 08:45 am UTC (link)
	Anyone who uses this to determine serious matters of trust is crazy. The sort of trust that trustmetrics are good for is things like "deserves half a second of my attention" - the sort of trust spammers abuse. (Reply to this) (Parent)

	skx 2006-03-29 11:02 am UTC (link)
	Minor bug report: "foo-bar" is not a valid LiveJournal user name. Instead "foo_bar" must be used. (Reply to this) (Thread)

	ciphergoth 2006-03-29 04:05 pm UTC (link)
	I don't quite understand - can you make explicit how to reproduce the bug, what should happen and what happens? Thanks! (Reply to this) (Parent)(Thread)

(no subject) -

skx, 2006-03-29 04:10 pm UTC

(no subject) -

ciphergoth, 2006-03-29 04:15 pm UTC

(no subject) -

skx, 2006-03-29 04:34 pm UTC

(no subject) -

ciphergoth, 2006-03-29 04:26 pm UTC

	lalato 2006-03-29 11:53 am UTC (link)
	I'm not sure I understand the numbers. Should lower or higher numbers be more "trustworthy"? --sam (Reply to this)

	californicate 2006-03-29 01:10 pm UTC (link)
	So, basically, these are friends of my friends . . . but I just don't understand "Close to me". If it's not based on any of my personal information then what is it based on? How "popular" these people are within my friends LJ? - i.e. how many times they are mentioned, tagged, or respond to enteries? (Reply to this) (Thread)

	dossier 2006-03-29 01:52 pm UTC (link)
	As far as I can tell, friends of your friends that appear to be physically close to you in space based on the location in their user info. (Reply to this) (Parent)(Thread)

(no subject) -

ciphergoth, 2006-03-29 02:04 pm UTC

shadowcub67
2006-03-29 04:22 pm UTC (link)

After running the metric, I looked over the results and realised there were a few people who should ALREADY have been on my friends list. Also, a few more that shared a lot of common interests with me that I decided to friend & follow their journals for a while to see if we are a match.

Unfortunately, there seems to be no way to cause the metric to be recalculated based on the revised friend's list.

(Reply to this)

mirukux
2006-03-29 05:22 pm UTC (link)

i have a query about my results;

# jimjamthehuman (350 - 400)
# mandraked, metalmikey666 (400 - 450)
# challenged, hannah_honey (450 - 500)
# geesuv (500 - 550)

first three are fine, i know them or have known them previously and we have several shared friends, but i've only one friend in common with challenged and hannah_honey. things seem fine after that though, with 9 friends in common with geesuv. is there some rational i'm missing for this result?

(Reply to this)

	confessionalbox 2006-03-29 06:18 pm UTC (link)
	very cool concept, almost like a spatial regression are there any weights applied to users as the link becomes more tenuous? You might answer a lot of questions if you gave a chart of the score scale.. (Reply to this)

	kali921 2006-03-29 07:25 pm UTC (link)
	Can you explain what the numerical values attached to a particular group mean? How does that measure "distance"? It seems completely arbitrary and confusing. (Reply to this)

anarchy_lime
2006-03-29 08:19 pm UTC (link)

Here's a long explanation I gave a friend who posted this meme in their blog. Maybe this'll help.

---

Basically, yes. Those people that appear most often throughout your LJ friends' friends lists will show up with the lower numbers. The number can be thought of as 'distance' from you -- so, lower numbers mean they're closer.

But it's not necessarily just people that are friends of friends. It could be friends of friends of friends. Or friends of friends of friends of friends. Etc., Etc. This is why you may get names you've never heard of.

For those people who have very few (LJ) friends, it will start to include the friends of that LJer's friends. And if those friends similarly have a small number of friends, then it will continue to cascade down the line. The situation is the opposite for people with lots of friends.

Whether or not this is a valid social organizing tool isn't clear. If you had two friends who were part of a very small community and 2 other friends who gregariously were part of a very large community, all the small community people would come up first.

Perhaps, if the algorithm description of a trust flow doesn't make sense, use a different metaphor. Compare 2 friends who have 1 hour each to read their respective friends' blogs. If friend 1 only has 10 friends, that's 6 minutes per person, so he'll start looking at friends of his friends. And in any case, those 10 people must be pretty important since he's limited himself to just 10. Whereas, if friend 2 has 100 friends, then he only has 36 seconds to devote to each friend. It's unlikely that he looks beyond his 100 friend list since he's already pushed for time. And since he's spreading himself so thin, friend 2 doesn't think as highly of each individual as friend 1 thinks of his friends. If friend 2 did think some of his friends were more important, he'd cut down on the excess so he could focus on true friends.

So, both this relative importance, and the effect of friends w/ few friends reading friends of friends are the two complicating factors on top of just 'Who is friends w/ my friends that I am not friends with.'

(Reply to this) (Thread)

	i am lost lost_rose 2006-03-29 08:36 pm UTC (link)
	hell i guess thats why i am called lost rose LOL. I dont understand any of this at all could someone please explain. (Reply to this) (Parent)

	musical_sarcasm 2006-03-30 02:09 am UTC (link)
	Okay. Everything works until I try to post it in my lj. I'm already sighed in, like always, and when I click on the button to have it put it in my lj, nothing shows up in the update. (Reply to this) (Thread)

	ciphergoth 2006-03-30 07:17 am UTC (link)
	That's weird. If I press that button on your results it looks right. What browser/OS are you using? (Reply to this) (Parent)(Thread)

(no subject) -

musical_sarcasm, 2006-03-31 04:12 am UTC

	cyclophile 2006-03-30 02:32 am UTC (link)
	I am completely confused as to both the point and the result of this feature. and nope...the explanations that are already out there do not make sense. Plain english, please.....what the heck? (Reply to this)

marco235
2006-03-30 02:50 am UTC (link)

I've got not an error message per se, but something more disturbing: a person appears at the top of the list (in the 0-50 range) who I've never heard of at all, and upon clicking that particular user profile, I see absolutely nobody on said user's friends list who I've heard of either. In other words, someone who would be a large number of degrees removed has managed to appear at the top of the board.

The remainder appear to be ordering correctly.

Very cool feature.

(Reply to this)

	garneyloom 2006-03-30 05:23 am UTC (link)
	Um, it worked pretty well except it included someone who IS on my friends list. That's confusing. (Reply to this) (Thread)