Lone Gunman

The small sam­ple size of Stan­ley Milgram’s small world exper­i­ment means that the the­ory of ‘six degrees of sep­a­ra­tion’ and the con­clu­sion drawn from it–primarily, the Influential’s the­ory pop­u­larised by Mal­colm Glad­well in The Tip­ping Point–could be deeply flawed. That was the start­ing point for Dun­can Watts’ research that led him to say “the Tip­ping Point is toast”.

So to research how ideas and trends spread virally, Watts (who is author of Every­thing is Obvi­ous, prin­ci­pal research sci­en­tist at Yahoo! Research (he directs their Human Social Dynam­ics group), and found­ing direc­tor of Colum­bia University’s Col­lec­tive Dynam­ics Group) ran large-scale repro­duc­tions of the small world exper­i­ment and hun­dreds of com­puter sim­u­la­tions that brought for­ward two con­clu­sions: the six degrees of sep­a­ra­tion the­ory is cor­rect, but there is no evi­dence for super-connected ‘trend gate­keep­ers’ (such as Gladwell’s ‘Con­nec­tors’):

But Watts, for one, didn’t think the gate­keeper model was true. It cer­tainly didn’t match what he’d found study­ing net­works. So he decided to test it in the real world by remount­ing the Mil­gram exper­i­ment on a mas­sive scale. In 2001, Watts used a Web site to recruit about 61,000 peo­ple, then asked them to ferry mes­sages to 18 tar­gets world­wide. Sure enough, he found that Mil­gram was right: The aver­age length of the chain was roughly six links. But when he exam­ined these path­ways, he found that “hubs”–highly con­nected people–weren’t cru­cial. Sure, they existed. But only 5% of the email mes­sages passed through one of these super­con­nec­tors. The rest of the mes­sages moved through soci­ety in much more demo­c­ra­tic paths, zip­ping from one weakly con­nected indi­vid­ual to another, until they arrived at the target. […]

[His com­puter sim­u­la­tion] results were deeply coun­ter­in­tu­itive. The exper­i­ment did pro­duce sev­eral hun­dred soci­ety­wide infec­tions. But in the large major­ity of cases, the cas­cade began with an aver­age Joe (although in cases where an Influ­en­tial touched off the trend, it spread much fur­ther). To stack the deck in favor of Influ­en­tials, Watts changed the sim­u­la­tion, mak­ing them 10 times more con­nected. Now they could infect 40 times more peo­ple than the aver­age cit­i­zen (and again, when they kicked off a cas­cade, it was sub­stan­tially larger). But the rank-and-file cit­i­zen was still far more likely to start a contagion.

I can’t help but find it some­what ironic that, writ­ten almost four years ago, this argu­ment hasn’t really gained much trac­tion and Gladwell’s ideas are still dis­cussed ad nau­seam.

User inter­face design­ers (and par­tic­u­larly icon design­ers) could learn a lot from comics and the the­ory behind them.

Tak­ing his cue from Scott McCloud’s excel­lent Under­stand­ing Comics, Lukas Mathis looks at how for opti­mum recog­ni­tion and in order to aid under­stand­ing, user inter­face ele­ments must find the sweet spot between uni­ver­sal­ity and real­ism. Like when draw­ing cer­tain comics, it’s about find­ing the opti­mum com­prim­ise between too lit­tle detail and too much.

Peo­ple are con­fused by sym­bols if they have too many or too few details. They will rec­og­nize UI ele­ments which are some­where in the middle.

The trick is to fig­ure out which details help users iden­tify the UI ele­ment, and which details dis­tract from its intended mean­ing. Some details help users fig­ure out what they’re look­ing at and how they can inter­act with it; other details dis­tract from the idea you’re try­ing to con­vey. They turn your inter­face ele­ment from a con­cept into a spe­cific thing. Thus, if an inter­face ele­ment is too dis­tinct from its real-life coun­ter­part, it becomes too hard to rec­og­nize. On the other hand, if it is too real­is­tic, peo­ple are unable to fig­ure out that you’re try­ing to com­mu­ni­cate an idea, and what idea that might be.

To keep the peace on the ever-expanding Stack Exchange Net­work of online com­mu­ni­ties, own­ers Joel Spol­sky and Jeff Atwood intro­duced the timed sus­pen­sion of dis­rup­tive users’ accounts. Over time the trans­parency of the timed sus­pen­sion process proved to be occa­sion­ally inef­fi­cient when dis­cus­sions arose regard­ing the mer­its of cer­tain sus­pen­sions. This led the admin­is­tra­tors of the com­mu­ni­ties to inves­ti­gate other ways of mod­er­at­ing prob­lem­atic users.

What they found were three fan­tas­ti­cally devi­ous secret ways to effec­tively con­trol trolls and other abu­sive users on online com­mu­ni­ties: the hell­ban, slow­ban, and errorban:

A hell­banned user is invis­i­ble to all other users, but cru­cially, not him­self. From their per­spec­tive, they are par­tic­i­pat­ing nor­mally in the com­mu­nity but nobody ever responds to them. They can no longer dis­rupt the com­mu­nity because they are effec­tively a ghost. It’s a clever way of enforc­ing the “don’t feed the troll” rule in the com­mu­nity. When noth­ing they post ever gets a response, a hell­banned user is likely to get bored or frus­trated and leave. I believe it, too; if I learned any­thing from read­ing The Great Brain as a child, it’s that the silent treat­ment is the cru­elest pun­ish­ment of them all. […]

(There is one addi­tional form of hell­ban­ning that I feel com­pelled to men­tion because it is par­tic­u­larly cruel – when hell­banned users can see only them­selves and other hell­banned users. Brrr. I’m pretty sure Dante wrote a chap­ter about that, somewhere.)

A slow­banned user has delays forcibly intro­duced into every page they visit. From their per­spec­tive, your site has just got­ten ter­ri­bly, hor­ri­bly slow. And stays that way. They can hardly dis­rupt the com­mu­nity when they’re strug­gling to get web pages to load. There’s also sci­ence behind this one, because per research from Google and Ama­zon, every page load delay directly reduces par­tic­i­pa­tion. Get slow enough, for long enough, and a slow­banned user is likely to seek out greener and speed­ier pas­tures else­where on the internet.

An error­banned user has errors inserted at ran­dom into pages they visit. You might con­sider this a more severe exten­sion of slow­ban­ning – instead of pages load­ing slowly, they might not load at all, return cryp­tic HTTP errors, return the wrong page alto­gether, fail to load key depen­den­cies like JavaScript and images and CSS, and so forth. I’m sure your devi­ous lit­tle brains can imag­ine dozens of ways things could go “wrong” for an error­banned user. This one is a bit more eso­teric, but it isn’t the­o­ret­i­cal; an exist­ing imple­men­ta­tion exists in the form of the Dru­pal Mis­ery mod­ule.

Chil­dren learn a lot from imi­tat­ing the actions of adults, with recent research sug­gest­ing that infants as young as 14 months are selec­tive imi­ta­tors — tak­ing cues from our behav­iour in order to decide which of us adults to learn from and which to ignore.

In a study where researchers expressed delight before either pre­sent­ing an infant with a toy (the reli­able con­di­tion) or not pre­sent­ing the infant with a toy (the unre­li­able con­di­tion), they dis­cov­ered that infants detect “unre­li­able” peo­ple and choose not to learn from then, opt­ing instead for adults that appear con­fi­dent and knowl­edge­able — the reli­able group.

“Infants seem to per­ceive reli­able adults as capa­ble of ratio­nal action, whose novel, unfa­mil­iar behav­iour is worth imi­tat­ing,” the researchers said. “In con­trast, the same behav­iour per­formed by a pre­vi­ously unre­li­able adult is inter­preted as irra­tional or inef­fi­cient, thus not wor­thy of imitating.” […]

The new find­ing adds to a grow­ing body of research show­ing children’s selec­tiv­ity in who they choose to learn from. For exam­ple, chil­dren pre­fer to learn from adults as opposed to their peers, and they pre­fer to learn from peo­ple they are famil­iar with and who appear more cer­tain, con­fi­dent and knowledgeable.

Recent research sug­gests that domes­tic dogs seem capa­ble of dis­play­ing a rudi­men­tary “the­ory of mind” — a very human char­ac­ter­is­tic whereby you are able to attribute men­tal states to oth­ers that do not nec­es­sar­ily coin­cide with your own (in a nut­shell). Stray domes­tic dogs, mean­while, do not dis­play this trait, sug­gest­ing that such men­tal attrib­utes are devel­oped through close con­tact with humans. That’s inter­est­ing, but not the main rea­son I’m shar­ing this infor­ma­tion with you.

This cog­ni­tive dif­fer­ence between stray domes­tic dogs and their house­bound brethren was uncov­ered by test­ing whether or not they under­stood the very human action of point­ing (y’know, with your index fin­ger). What struck me most in this dis­cus­sion was this brief the­ory of how the action of point­ing evolved:

Go ahead, let your wrist go limp and look at your hand from the side, or if you’re too inse­cure in your own sex­u­al­ity, just pic­ture Adam’s limp wrist at the moment of cre­ation in Michelangelo’s mas­ter­piece on the Sis­tine Chapel’s ceil­ing. See how even in this relaxed state the index fin­ger is slightly extended? By con­trast, when chimps do this […] their index fin­ger falls nat­u­rally in line with their other fin­gers. Povinelli and Davis rea­son that this sub­tle evo­lu­tion­ary change in the mor­phol­ogy of our hands, which occurred after humans and chim­panzees last shared a com­mon ances­tor five mil­lion to seven mil­lion years ago, is at least par­tially respon­si­ble for the fact that human point­ing with the index fin­ger is so cul­tur­ally ubiq­ui­tous today.

The argu­ment goes some­thing like this. When young infants begin reach­ing for objects just out of their range, adults are most likely to respond to those reach­ing attempts and to retrieve the item for the baby when the latter’s index fin­ger is more promi­nently extended. That is to say, ini­tially, the adult mis­tak­enly reads into the child’s reach­ing attempt as a com­mu­nica­tive ges­ture on the part of the child. Over time, this dynamic between the child and adult serves to fur­ther “pull out” the index fin­ger because the child implic­itly learns the behav­ioral asso­ci­a­tion, so that it slowly becomes a gen­uine point­ing gesture.