TrustFlow results for ciphergoth

[ciphergoth]

I tried out TrustFlow II for LiveJournal. The following people not on the friends list for ciphergoth are close by:

• 500 - 550: potatoprint
• 550 - 600: elfgeek
• 600 - 650: thehalibutkid, sanjibabes

• 650 - 700: ippola, kingginger, lozette, delomelas, nou
• 700 - 750: vampwillow, heteronormative, vicki_t_veg
• 750 - 800: eather, louis_mallow, marymortis, angelicdance, flufybumblebee, emmy_mallow, poannie, topbit, avalon_bliss, occular, princealbert
• 800 - 850: theauldhoose, zapruder, inanimatepeanut, babyholly, hfnuala, memevector, sublevel3
• 850 - 900: graydon, hypermallard, naturalborn, just_becky, karen2205, cherryblossom1, sheridanwilde, ankh, fuzzygoth, snagglepat
• 900 - 950: nickmc, feanelwa, nitoda, calendril, original_aj, cyberpunkgrrl, princess_leanna
• 950 - 1000: maleficent, bethanthepurple
• 1000 - 1050: earwigmc, superdus, preachermuaddib, mr_flay, anjylle, andyravensable, pandemona, feanorfinwe, mjwin, leila_child, leathermgoddess, wintrmute, alienspacebat, hughe
• 1050 - 1100: diffrentcolours, artremis, stonecia, pirate_gypsy, mattblakk, olethros, wuluf, gusmacroy, pywacket, thingie
• 1100 - 1150: angeoverthere, brain_hurts, darkstones, essbee80, rhialto, sulis_minerva, crag_du, blankinfinity, gothbabe, x_mass, latexiron, mock2, _boing_
• 1150 - 1200: swisstone, lnr, jhaelan, ninadelenne, cheekbones3, nyecamden, mandrake91, meirion
• 1200 - 1250: clockworkwasp, cookwitch, snowberries, tylluan_dawel, decepticonboy, puzzle_, calum, fabalafae, amphioxa, liadnan, childeric, wandra, jenbird, mab, lproven
• 1250 - 1300: buca, spidermonster, scunthorpe, oilrig, parallelgirl, sneerpout, baratron, androktone, my_mundane_life, protocol_rain, foibey, mankylarry, cassieclaire, drreagan
• 1300 - 1350: redandfiery, neriedes, batswing, liz_with_hat, giolla, simonb, deathboy, penguinsteph, squiddity
• 1350 - 1400: ofiwriter, lilgreendave, taimatsu, jimthegoth, davefish, ant_girl, bi_dancer13, zoo_music_girl, ms_katonic, dmh
• 1400 - 1450: fire_sermon, sushidog, anonymouseth, tabatha_yeddap, satan_dot_com, alexa_robinson, lark_ascending, neferet, gypseymission, fruit_boy, northerncutie, interior_lulu, spacelem, mirukux, zarbi, excy, steer, purple_fayz, headinclouds, kixie, emmelinemay
• 1450 - 1500: god_dot_com, rainbowskye, princesssulky, esbat, cosmicdjester, eternaldancer, ophelia_complex, bauhaus_691
• 1500 - 1550: chillies, dr_lovely, bob_dot_com, shera619, 2eyes, voofy, poggs, djsynne, coeur_en_hiver, becksydee, purpledonna, libellum, martling, circumnavigator
• 1550 - 1600: teqkiller, mistersleepless, blahflowers, ghoti, dopedolljunkie, a_musing_amazon, davidmcn, bustiq, neilh, sarcaustick, reindeerflotila, funky_firelord, kyte, insecuregoddess, johnnydefective

TrustFlow II: Who is closest to your friends list?

Comments

[booklectica]

Tried, but it says it's too busy — will try later!

[ciphergoth.livejournal.com]

Wow, that was quick! Machine doesn't seem too heavily loaded, so I've doubled the maximum number of parallel users.

Got it

[booklectica]

Got it. Too long to post as a comment though. But looked accurate - I knew or had heard of most of the people listed, and knew fewer of them as the distance increased.

Re: Got it

[ciphergoth.livejournal.com]

If you post it to your journal, it includes a cut-tag automatically...

Re: Got it

[booklectica]

Can do if you like.

[glitchdrei.livejournal.com]

Very cool :)

[funky-firelord.livejournal.com]

And still you have not Friended me ciphergoth *Pouts*

:-)

Firelord

[drdoug.livejournal.com]

Seems to work as advertised for me - almost uncanny in picking out all the people I'm aware of through friends' journals but haven't friended, which is what I'd expect of a 'LJ friends closeness' measure. Hooray!

Interesting to compare LJ 'friends of friends' with what I know about who are real-life friends together (or at least, SFAIK) - they are clearly not the same thing but there does seem to be at least a loose correlation.

I do like your 'bucket of trust juice filled from heaven' explanation too. A lovely image, and clearly explains the algorithm in non-mathematical terms.

I'm not sure calling it trust juice helps to explain it to LJers, though. I entirely understand that the algorithm is a member of the class of trust metrics, and that this is the context in which you developed it. However, LJers don't universally understand friending someone to mean expressing any degree of trust. (Even though in the trustmetrics sense one would almost certainly describe it as such.) Perhaps you could call it 'friendedness juice' or 'LJ-friends juice' or something like that? The concepts of LJ-friending and real-life friending are sufficiently confusing already without introducing trust as a third factor.

You could always include a footnote to the effect that this is a type of algorithm called a trust metric, and that here 'friendedness' takes the role of trust. The sort of people who care about the algorithm are - handily - precisely the sort of people who might read such a footnote. *grin*

[ciphergoth.livejournal.com]

That seems a good idea - will edit...

[aidan-skinner.livejournal.com]

your 'fill in the lj username from (presumably) the http referrer header' doesn't like _

[ciphergoth.livejournal.com]

ah! Fixed - thanks!

[aidan-skinner.livejournal.com]

Also, some of the results seem a little... odd.. inverted_tree has only 2 people i know listed, only one of whom has friended them back. There a couple of similar odditys, maybe give less weight to people with low numbers of friends, since 1 or 2 mutuals seems to skew the results when they only have a handful of people friended?

[purplerabbits.livejournal.com]

Yeah, I get odd results like that. My top two are gerald_duck and nunfetishist who I have never even heard of, and who only have a couple of friends in common with me, but one of those is Xhosa, who only has those two asfriends... After that it sort of makes sense.

[vvalkyri.livejournal.com]

ahh. my list makes some more sense now.

[vvalkyri.livejournal.com]

pretty cool.
cripes, I wish that my ex's newest girlfriend, who only even got an lj some month's back, weren't quite so high in the trust ratings. Unsurprising; 2/3 of her flist are close meatspace friends of mine she met through him.

[mskala.livejournal.com]

If someone is closely connected to me, and to my friends, but is not on my friends list, then that may be someone I know pretty well, but have had a falling-out with and deliberately not included on my list. This system seems inclined to give such people low numbers (more trust?), and that would be a bad thing if it were used as a basis for automatic decisions about whom I want to trust. It's a measurement of connection but not such a good measurement of trust, because trust is sometimes quite strongly non-transitive.

There are very few people I explicitly distrust, but Elsie seems to be encountering this effect (http://elsibeth.livejournal.com/164410.html).

Do the numbers mean anything really significant? Most people's posted results seem to start at about 500, but there are about 50 people in my result in the below-500 categories. I'm part of several closely-knit groups (near-cliques in the graph sense); maybe that has an effect.

[ciphergoth.livejournal.com]

The numbers mean roughly this: at most 500 people can score below 500. In practice many fewer will. Since you and all of your friends list have lower score than the first person listed, if your friends list is longer than the first score you see will be higher.

Nitpicking

[jbadger.livejournal.com]

Seems to have a bug in missing end tags (not proper XML)
that is <lj user="someone" >
should be
<lj user="someone" />

Thank you for a great program!

Re: Nitpicking

[ciphergoth.livejournal.com]

Oops! Thanks.

BTW please leave further comments on the entry in trustmetrics - thanks!

don't think I am on yours, but you are on mine

[badriya.livejournal.com]

It's made a mistake though. It says pengshui_master is not a friend, but he is. Hang on, I friended you so you shouldn't be on mine either.


* leathermgoddess (100 - 150)
* orddu, tsebaoth, remmie, khaylock, buffer, crispygoth, pengshui_master, listy, willmash (150 - 200)
* bluebirf, padfoot_uk, plankman, sultrylady (200 - 250)
* skibbley, crazyscot, sugsna, met24, minusbat, techiebabe, cujosmurf, arealhighlander, meltie, neilh, cardboardnewman, alexmc, nolley, noidd, tenebrosity, test_freak, the_glyph, shnells (250 - 300)
* medwyn, lproven, fluke_holden, madkatsjournal, jaymez007, aperrott, nitoda, kafffeeeeeeeee (300 - 350)
* leathereagle, msdemmie, purplepants, sweetiejar, tanais, tallis_ryhope, f_l_i_r_t, confesstome, scopo, dougs, potatoprint, twinkle_lfs, nadiajane, wandra, leatherdykeuk (350 - 400)
* some_fox, envoy, soupdragon123, wood_rose, meta_zombi, bikerbabe, rosalee, hddod, kochanski, hollytg2003, cerebralpig, barakta (400 - 450)
* aeglefinus, purple_peril, adelheid, cassieclaire, ciphergoth, shinyshimmer, memevector, jaggery, bilba, rope_puddle, flickgc, katyha, burgi (450 - 500)
* theperkygoth, drpete, babysimon, darkpsychosis, d_floorlandmine, ingaborg, riocaz, 2eyes, karen2205, tictactoepony, sneerpout (500 - 550)
* fuzzygoth, incy, vmslj, imposterer, mmcpoland, princess_leanna, latexiron, cookwitch, purplerabbits (550 - 600)

Post this to your LiveJournal

Can you take a look at another trust metrics?

[krotty.livejournal.com]

Hi!

I'm working on an open source project, and we need a kind of trust-metric, but not quite. Something that would capture respect more than trust or reputation. Can you take a look at my lj (http://krotty.livejournal.com) and see if the algorithm and appoach described there makes sense? The project's web site is http://bitchun.org

Thanks,
Joe