Suivre par Email

samedi 5 octobre 2013

Some Obvious Things About Internet Reputation Systems

Debates around the “shar­ing econ­omy” have been dri­ven by per­sonal sto­ries and broad claims. In con­trast, this is a fairly dense step-by-step look at the inter­net rep­u­ta­tion sys­tems on which the shar­ing econ­omy claims are based, with some pre­dic­tions about the neolib­eral future of the shar­ing econ­omy. Ref­er­ences, sta­tis­tics, and, yes, some per­sonal sto­ries are rel­e­gated to foot­notes. If you’re really inter­ested you can down­load a PDF.


Inter­net rep­u­ta­tion sys­tems let indi­vid­u­als rate other indi­vid­u­als over the inter­net and pro­vide rec­om­men­da­tions based on those rat­ings. A new class of enter­prise claims to use inter­net rep­u­ta­tion sys­tems to enable shar­ing of per­sonal goods and ser­vices at unprece­dented scale. Its rise has been announced by both Forbes, and by The Econ­o­mist, accord­ing to which accom­mo­da­tion rental ser­vice Airbnb…

…is the most promi­nent exam­ple of a huge new “shar­ing econ­omy”, in which peo­ple rent beds, cars, boats and other assets directly from each other, co-ordinated via the inter­net. …[T]echnology has reduced trans­ac­tion costs, mak­ing shar­ing assets cheaper and eas­ier than ever—and there­fore pos­si­ble on a much larger scale… social net­works pro­vide a way to check up on peo­ple and build trust; and online pay­ment sys­tems han­dle the billing.

The claim is that inter­net rep­u­ta­tion sys­tems solve two prob­lems. One is coor­di­na­tion (can I find some­one who has what I want, or wants what I have?) and the other is trust (can you trust the per­son on the other side of the exchange to keep their end of the bar­gain?).1 Shar­ing econ­omy advo­cates claim, and I will return to this at the end of the essay, that it is both nec­es­sary and suf­fi­cient to solve these prob­lems to unlock a large new econ­omy of resource sharing.

Trust and Coordination

To under­stand the shar­ing econ­omy it is nec­es­sary to under­stand trust.2

A truster must decide whether or not to make a loan to a poten­tial trustee; if the truster does make the loan, then the trustee must decide whether or not to repay it. We say the truster trusts the trustee if she expects him to repay, and the trustee is trust­wor­thy if he would repay a loan, should the truster make it.

Trust is a prob­lem of asym­met­ric infor­ma­tion: a truster can­not divine the trustee’s trust­wor­thi­ness directly but must look instead for signs of trustworthiness.

An oppor­tunist is some­one who is not trust­wor­thy but who seeks to mimic signs of trust­wor­thi­ness in order to deceive poten­tial trusters. Oppor­tunists cre­ate what Bacharach and Gam­betta call a “prob­lem of sec­ondary trust” which, they argue, “almost always accom­pa­nies, and is often the key to solv­ing, prob­lems of pri­mary trust” (p158). Instead of just look­ing for signs of trust­wor­thi­ness, the truster must decide whether she can trust those signs; instead of just dis­play­ing signs of trust­wor­thi­ness, the trustee must con­vince the truster that he is not mim­ic­k­ing them.

Sec­ondary trust is a sig­nalling prob­lem in the sense first spelled out by econ­o­mist Michael Spence. An effec­tive sig­nal is an action or sign that is easy for a trust­wor­thy per­son to dis­play but costly for an untrust­wor­thy per­son to dis­play. If it’s not worth the effort for an oppor­tunist to mimic the sig­nal, we say that the sig­nal sep­a­rates trust­wor­thy peo­ple from untrust­wor­thy peo­ple or dis­crim­i­nates between them. If no dis­crim­i­nat­ing sig­nal is avail­able, then there is no way to dis­tin­guish trust­wor­thy peo­ple from opportunists—an out­come that is called pool­ing—and trust can­not be estab­lished between truster and trustee. In real life, of course, we deal with prob­a­bil­i­ties rather than cer­tain­ties, but there is a spec­trum from sep­a­rat­ing to pool­ing out­comes in prob­lems of trust.

The Econ­o­mist observed, above, that the inter­net has reduced the trans­ac­tion costs of col­lab­o­ra­tion, enabling what Yochai Ben­kler calls a “new modal­ity of orga­niz­ing pro­duc­tion: rad­i­cally decen­tral­ized, col­lab­o­ra­tive, and non-proprietary… ‘commons-based peer pro­duc­tion’” (The Wealth of Net­works, p60). But the prob­lem of sec­ondary trust empha­sizes that low trans­ac­tion costs do not nec­es­sar­ily improve col­lab­o­ra­tion.

Other things being equal, and in the absence of oppor­tunists, lower trans­ac­tion costs (dis­cov­ery and com­mu­ni­ca­tion) should increase the amount of col­lab­o­ra­tion, but in the pres­ence of opportunists—in what Bacharach and Gam­betta call “mimic-beset trust games”—collaboration is pos­si­ble only in the pres­ence of an effec­tive sig­nalling mech­a­nism, and low­er­ing trans­ac­tion costs can destroy trust-dependent col­lab­o­ra­tion by mak­ing it eas­ier for oppor­tunists to mimic trust­wor­thi­ness.3


In the sense used here, rep­u­ta­tion is a sign of trust­wor­thi­ness man­i­fested as tes­ti­mony by other peo­ple. When my neigh­bour says “Don’t hire John the Plumber: he came to fix my sink but it’s still blocked”, she is pro­vid­ing infor­ma­tion that lets me decide whether to trust John to fix my drains.

When it works well, rep­u­ta­tion is an effec­tive dis­crim­i­nat­ing sig­nal that pro­motes trust and col­lab­o­ra­tion based on trust. In a com­mu­nity with strong word of mouth, it is easy for a good plumber to estab­lish a rep­u­ta­tion as reli­able, punc­tual, and skilled sim­ply by being reli­able, punc­tual, and skilled; it is dif­fi­cult for an incom­pe­tent or lazy plumber to do the same.

Rep­u­ta­tion is not a per­fect dis­crim­i­nat­ing sig­nal. Much of what is com­mu­ni­cated in tes­ti­monies may be pri­vate and infor­mal (“he fixed my sink and came on time, but there was some­thing about him… I just didn’t like hav­ing him in my house”) and this pri­vacy and infor­mal­ity can have both good and bad effects. It can trans­mit jus­ti­fied but neb­u­lous sus­pi­cions, but it makes it dif­fi­cult for John to gain a good reputation—no mat­ter how trust­wor­thy he is—if he is a black man try­ing to find work in a white com­mu­nity with a his­tory of racism, or dif­fi­cult for Jane the Plumber’s skills to be taken seri­ously if the com­mu­nity has tra­di­tional norms about women’s roles. “Old boys’ clubs” and other insider groups pro­vide mem­bers with an inbuilt advan­tage when it comes to estab­lish­ing a reputation.

Rep­u­ta­tion is only one mech­a­nism for solv­ing the prob­lem of trust. Oth­ers include reci­procity in long-term rela­tion­ships, reg­u­la­tions (you can trust this restau­rant because it has passed a food safety inspec­tion), pro­fes­sional qual­i­fi­ca­tions (you can trust this per­son to fix your leg because she is a doc­tor), vol­un­tary indus­try cer­ti­fi­ca­tions (you can trust this cof­fee to be fair trade because there is a fair trade label on the pack­age), inde­pen­dent rat­ing agen­cies, indi­vid­ual firm com­mit­ments (you can trust this retailer because they have invested heav­ily in their brand, and so must act accord­ingly), the com­mon prop­erty regimes explored by Eli­nor Ostrom, and many others.

Rep­u­ta­tion, in the sense used here, is peer-to-peer, infor­mal, decen­tral­ized, community-driven, and non-commercial, and it is those alter­na­tive qual­i­ties that shar­ing econ­omy advo­cates claim can be scaled up by using inter­net rep­u­ta­tion sys­tems. Airbnb and BlaBlaCar both describe them­selves as “a trusted com­mu­nity mar­ket­place”; Lyft’s one-million rides show “the power of community”.

The effec­tive­ness of rep­u­ta­tion depends on the moti­va­tions of those giv­ing tes­ti­monies as well as on the actions of the trustee: the prob­lem of sec­ondary trust described above. Rep­u­ta­tion is effec­tive only if the tes­ti­monies are inde­pen­dent and free from the taint of col­lu­sion or retal­i­a­tion. Tes­ti­mony from John’s brother does not carry the same weight as that of some­one who has no stake in John’s suc­cess or fail­ure, and while John may not want my neigh­bour to tell me about his fail­ure to fix their sink, there’s not a lot he can do about pri­vate con­ver­sa­tions over a gar­den fence.

Market-based incen­tives erode the effec­tive­ness of rep­u­ta­tion, and in this respect rep­u­ta­tion is a cul­tural com­mons (see here, and see also me here). In her TED talk, influ­en­tial author Rachel Bots­man says that in the new econ­omy “rep­u­ta­tion will be your most valu­able asset”, but as rep­u­ta­tion becomes an impor­tant asset, mar­kets will grow around it and inter­me­di­aries will claim to help you boost your rep­u­ta­tion, but these market-based incen­tives destroy the value of rep­u­ta­tion as a mech­a­nism for estab­lish­ing trust. Mech­a­nisms for buy­ing and sell­ing tes­ti­monies, for exam­ple, cause tes­ti­monies to lose their abil­ity to dis­crim­i­nate between trust­wor­thi­ness and oppor­tunism because an oppor­tunist with money could buy them­selves a good reputation.

Inter­net Rep­u­ta­tion Systems

Inter­net rep­u­ta­tion sys­tems promise to cre­ate a global vil­lage by scal­ing up infor­mal word-of-mouth rep­u­ta­tion mech­a­nisms for shar­ing and for cre­at­ing trust, and so solve both the coor­di­na­tion and the trust prob­lem for a vari­ety of ser­vices which could not pre­vi­ously be exchanged. For shar­ing econ­omy advo­cates, rep­u­ta­tion is an alter­na­tive to reg­u­la­tion: in the recent book The Rep­u­ta­tion Econ­omy, law pro­fes­sor Lior Strahile­vitz asks us to “imag­ine if every plumber, man­u­fac­tured prod­uct, cell phone provider, home builder, pro­fes­sor, hair styl­ist, accoun­tant, attor­ney, golf pro, and taxi dri­ver were rated… In such a world, there would be dimin­ished need for reg­u­la­tory over­sight and legal reme­dies because con­sumers would police mis­con­duct themselves.”

Do inter­net rep­u­ta­tion sys­tems act as an effec­tive sig­nal of trustworthiness?

Fig­ure 1 is the dis­tri­b­u­tion of rat­ings for the Net­flix Prize data set. Net­flix rat­ings are not a rep­u­ta­tion sys­tem in the sense used here, in that they are not tes­ti­mo­ni­als about peo­ple: the data set con­sists of rat­ings of movies and TV shows by Net­flix cus­tomers. There is every rea­son to believe that the rat­ings are inde­pen­dent and hon­est: the rater can offer an opin­ion freely, hav­ing no rea­son to expect expect reward or pun­ish­ment for any par­tic­u­lar rat­ing. The rater also has an incen­tive to give a rat­ing that matches their actual opin­ion, as it enables Net­flix to rec­om­mend movies that bet­ter match their tastes. So Fig­ure 1 can take this as a rea­son­able dis­tri­b­u­tion of inde­pen­dent ratings.

Fig­ure 1.

BlaBlaCar, a French shar­ing econ­omy com­pany that con­nects “dri­vers with peo­ple trav­el­ling the same way” through­out Europe, has over a mil­lion reg­is­tered dri­vers, trans­ports over half a mil­lion pas­sen­gers every month, and is expand­ing rapidly. Also, it makes testimonial-based rat­ings avail­able on its web site. Fig­ure 2 is the dis­tri­b­u­tion of a set of 190,000 rat­ings from the site.4

Fig­ure 2.

Of 190129 dis­tinct rat­ings, 2152 were one-star, there was not a sin­gle two-star rat­ing, there was one three-star rat­ing, five four-star rat­ings, and 187971 five-star rat­ings. A BlaBlaCar rat­ing means some­thing dif­fer­ent from a Net­flix movie rating.

With over 98% of rat­ings being five stars, the rep­u­ta­tion sys­tem does not mean­ing­fully dis­crim­i­nate among dri­vers or rid­ers. A rep­u­ta­tion sys­tem that does not dis­crim­i­nate fails as a rep­u­ta­tion sys­tem: it fails to solve the prob­lem of trust.5

Col­lu­sion and fear of retal­i­a­tion are the rea­sons why there are essen­tially no reviews less than five stars for rides that take place. If you give a less-than-five star review then, unlike in the case of offline community-based tes­ti­mo­ni­als, it is vis­i­ble to the revie­wee, who can give you a harsh review in return and so affect your chance of get­ting future rides. Do you want to defend your opin­ion that the dri­ver was a bit close to the car in front, or that the car was a bit dirty, or do you just want to give a five-star review and make a note to your­self not to ride with them again? Col­lu­sion is the other side of the retal­i­a­tion coin: I know I turned up late and was eat­ing smelly food in your car and you didn’t like it, but so long as you give me five stars I’ll give you a good pos­i­tive rat­ing and we’re both bet­ter off. Nei­ther of these fac­tors need to be explicit or even to be very impor­tant to pro­duce large effects, because it makes no dif­fer­ence to me how I rate you. One seem­ingly tiny dif­fer­ence between word-of-mouth and the inter­net rat­ing sys­tem makes all the dif­fer­ence, that tes­ti­mo­ni­als are vis­i­ble to every­one includ­ing the revie­wee instead of every­one except the reviewee.

The prob­lem is not unique to BlaBlaCar. Reci­procity and col­lu­sion in the eBay rep­u­ta­tion sys­tem has been stud­ied here and the authors also pro­vide an esti­mate of how many dis­sat­is­fied peo­ple are not rat­ing their trustee:

The fact that from 742,829 eBay users… who received at least one feed­back, 67% have a per­cent­age pos­i­tive of 100%, and 80.5% have a per­cent­age pos­i­tive of greater than 99%, pro­vides sug­ges­tive sup­port for the bias. The obser­va­tion is in line with Del­laro­cas and Wood (2008) who exam­ine the infor­ma­tion hid­den in the cases where feed­back is not given. They esti­mate, under some aux­il­iary assump­tions, that buy­ers are at least mildly dis­sat­is­fied in about 21% of all eBay trans­ac­tions, far higher than the lev­els sug­gested by the reported feed­back. They argue that many buy­ers do not sub­mit feed­back at all because of the poten­tial risk of retaliation.

Finally, on Airbnb, review­ing of hosts by guests and guests by hosts also hap­pens in pub­lic and is rec­i­p­ro­cal. The Airbnb web site does not dis­play indi­vid­ual numer­i­cal reviews, although it does dis­play indi­vid­ual text reviews; instead it dis­plays the aver­age rat­ing that a room has received in each of sev­eral cat­e­gories (clean­li­ness, loca­tion, com­mu­ni­ca­tion,…) together with an over­all aver­age, rounded off to the near­est 0.5 out of five. The web site is less easy to tra­verse pro­gram­mat­i­cally, but out of well over a hun­dred offer­ings in New York, Syd­ney, Berlin and Paris I have yet to see a sin­gle one that is not rated 4.5 or 5.6

So even in the absence of explicit gam­ing, peer-to-peer inter­net rep­u­ta­tion sys­tems do not solve the prob­lem of trust. The BlaBlaCar site fails the basic test of dis­crim­i­nat­ing among almost any of the 190,000 dri­ves that took place—it fails to deliver any use­ful infor­ma­tion beyond giv­ing the occa­sional sign that a dri­ver or rider may not turn up.

The “Growth” of the “Shar­ing Economy”

Rachel Bots­man claims that “Even four years ago, let­ting strangers stay in your home seemed like a crazy idea”, and she describes the mete­oric growth of the shar­ing econ­omy. The pic­ture she paints would seem to be incom­pat­i­ble with the idea that inter­net rep­u­ta­tion sys­tems fail to solve the prob­lem of trust. What’s going on?

Some per­spec­tive is in order. The rapid growth of indi­vid­ual shar­ing econ­omy com­pa­nies does not rep­re­sent the appear­ance of new social prac­tices of shar­ing. The growth of shar­ing econ­omy com­pa­nies is, at least in part, a move­ment of already-existing social prac­tices to online forums.

“Let­ting strangers stay in your home” has long been a com­mon prac­tice. Mil­lions of peo­ple let strangers stay in their homes with­out the ben­e­fit of inter­net rep­u­ta­tion sys­tems: the over­all vaca­tion rental mar­ket, which includes cot­tages, apart­ments, sec­ond homes and other per­sonal rentals, is much larger than Airbnb.7 , 8

Sim­i­larly, BlaBlaCar is a small frac­tion of the over­all car­pool­ing prac­tice, which pre­cedes inter­net rep­u­ta­tion sys­tems. Car­pool­ing has been a wide­spread prac­tice in many urban cen­tres for decades, encour­aged by local gov­ern­ments and tran­sit author­i­ties with “Park-and-Ride” facil­i­ties, espe­cially in Europe where fuel costs are higher than North Amer­ica. Stu­dent union notice­boards have long been a way to coor­di­nate rides home for the week­end.9 , 10

The dis­persed nature of tourism boards, local travel author­i­ties, book­ing agen­cies, uni­ver­sity notice boards, and so on make count­ing trips and vis­its dif­fi­cult, while the cen­tral­ized nature of shar­ing econ­omy sites makes data col­lec­tion triv­ial for those with the infra­struc­ture, so it is easy to underestimate—or entirely neglect—the pre-existing economy.

Also, not all the activ­ity on shar­ing econ­omy web­sites is of the per­sonal, infor­mal kind that the site own­ers por­tray, so the growth of these com­pa­nies over­states the growth of shar­ing. Mul­ti­ple rentals are com­mon, sug­gest­ing rentals of prop­er­ties other than the host’s pri­mary res­i­dence, such as invest­ment prop­er­ties.11 , 12

The Prob­lem of Trust in the Shar­ing Economy

Still, shar­ing econ­omy web sites are grow­ing fast. How are they suc­ceed­ing if the peer-to-peer rep­u­ta­tion sys­tems fail to solve the prob­lem of trust?

One rea­son is that coor­di­na­tion is use­ful in itself. Clas­si­fied ads, whether local news­pa­pers or sites like Craigslist and Kijiji, solve coor­di­na­tion prob­lems but do not even try to solve prob­lems of trust beyond the most basic ver­i­fi­ca­tion. It’s left to indi­vid­u­als to con­tact each other, make arrange­ments, decide on the terms of a sale, and com­plete the deal. In trad­ing second-hand lawn­mow­ers for cheap prices, the worst that can hap­pen to a pur­chaser is that they over­pay by a few dol­lars, and that’s a risk that many are pre­pared to take. One option for shar­ing econ­omy com­pa­nies would be to accept that they are solv­ing only the sim­pler coor­di­na­tion prob­lem, and adopt a busi­ness model that has no involve­ment in the trans­ac­tion itself and which charges small list­ing fees.13 But such a busi­ness model will not pro­vide the returns that ven­ture cap­i­tal is expect­ing from this indus­try. Shar­ing econ­omy com­pa­nies funded by ven­ture cap­i­tal have no option but to solve the prob­lem of trust.

In some cases, com­mu­nity mem­ber­ship itself has pro­vided an ade­quate sig­nal of trust­wor­thi­ness, par­tic­u­larly in com­mu­ni­ties that are pre­pared to accept some level of risk. Oppor­tunists are screened out, to the extent that they need to be, by an implicit com­mu­nity selec­tion process, so that match­ing within the com­mu­nity is reduced to a coor­di­na­tion prob­lem. For exam­ple, travel site Couch­surf­ing built itself largely by word-of-mouth among young trav­ellers. Couch­surf­ing mem­bers pre-selected them­selves to be adven­tur­ous (so not look­ing for a high degree of assur­ance from the orga­ni­za­tion) and community-minded indi­vid­u­als with a com­mon inter­est in travel. Sim­ply being part of the Couch­surf­ing com­mu­nity was a sign that cor­re­lated with trust­wor­thi­ness, and com­mu­nity mem­bers vol­un­tar­ily under­took the addi­tional risk that came with the program.

Unfor­tu­nately, com­mu­nity mem­ber­ship as a sign of trust­wor­thi­ness does not sur­vive large scale growth, for two reasons.

As a com­mu­nity grows, it attracts oppor­tunists. In a small com­mu­nity, the ben­e­fit to an oppor­tunist of mim­ic­k­ing a sign of trust­wor­thi­ness is small, but as the com­mu­nity scales, the poten­tial ben­e­fits for oppor­tunists are larger, and the incen­tive to mimic trust­wor­thi­ness is greater. In evo­lu­tion­ary terms, Bacharach and Gam­betta describe the phe­nom­e­non as “model pre­cedes mimic”. Shar­ing econ­omy sites have ben­e­fited from com­mu­nity mem­ber­ship as a screen­ing process, but as they become larger they will need new solutions.

Sec­ond, peo­ple who have a com­mit­ment to a com­mu­nity may be pre­pared to take on addi­tional per­sonal risk, either because of the nature of the com­mu­nity itself (a com­mu­nity of adven­tur­ers is not look­ing for a high level of secu­rity) or because they are pre­pared to over­look lapses to sup­port the com­mu­nity. There is a reci­procity, not just between indi­vid­u­als, but between the mem­bers and the community-as-a-whole. How­ever, the rev­enue and growth mod­els of venture-capital funded com­pa­nies are based on pro­vid­ing a ser­vice that can scale to peo­ple with no par­tic­u­lar com­mit­ment to the com­mu­nity itself. As AllTh­ingsD reporter Liz Gannes writes: “main­tain­ing cus­tomer trust is para­mount because, at any given moment, they are all one bad inci­dent away from users turn­ing back to more tra­di­tional arrangements”

The Future of the Shar­ing Economy

Ven­ture cap­i­tal demands for scale will pro­duce changes in the nature of the shar­ing econ­omy sites, changes that erode any com­mu­nity focus they have, and which turn them into far more tra­di­tional mod­els. Such changes are already under­way at the largest, most heav­ily funded sites.

As Gannes reports, a sin­gle bad inci­dent has forced Airbnb to hire a 50-person “trust and safety team” headed by a for­mer US Army intel­li­gence office and a for­mer gov­ern­ment inves­ti­ga­tor. The use of a human team clearly doesn’t scale, so Airbnb is now turn­ing to cen­tral­ized analy­sis to solve its prob­lems, say­ing “We want to apply data to every deci­sion. We want to be a very data-driven com­pany.” On April 30 2013, assert­ing that “Trust is the key to our com­mu­nity”, Airbnb intro­duced a “Ver­i­fied ID pro­gram” which demands that you pro­vide government-verified iden­ti­fi­ca­tion and per­mit the com­pany to ana­lyze your social net­work­ing pres­ence or pro­vide it with a video profile.

There is also a drive for more pro­fes­sion­al­ism among hosts. Airbnb now lets hosts sell tours and activ­i­ties, and here is Chip Con­ley, the new “Head of Global Hos­pi­tal­ity” for Airbnb, hired from the hotel indus­try, in a Sep­tem­ber 2013 inter­view:

We’ll be intro­duc­ing nine min­i­mum stan­dards around what we expect an Airbnb expe­ri­ence to be, whether it’s related to clean­li­ness or the basic ameni­ties you expect, which is not cur­rently the case. The idea that we cre­ate some ameni­ties that you should expect—clean tow­els, clean sheets—that’s important.

In short, Airbnb is aban­don­ing the idea that peer-to-peer rep­u­ta­tion sys­tems can solve the prob­lem of trust, is mov­ing away from the casual “air bed” men­tal­ity that gave it its name, and is resort­ing to tra­di­tional cen­tral­ized sys­tems of enforced min­i­mum stan­dards, doc­u­men­tary ver­i­fi­ca­tion, and so on.

There is, how­ever, one remain­ing dif­fer­ence between Airbnb and a tra­di­tional hos­pi­tal­ity busi­ness. To go back to the begin­ning of this essay, shar­ing econ­omy com­pa­nies claim that it is both nec­es­sary and suf­fi­cient to solve prob­lems of trust and coor­di­na­tion to unlock a large new econ­omy of resource shar­ing. The “suf­fi­cient” part of this is valid only if there are no spill-over effects from the oper­a­tions of the shar­ing econ­omy, so shar­ing economies will cam­paign for free­dom from those con­straints that pre­vent them max­i­miz­ing their returns: health and safety stan­dards, employ­ment stan­dards, licens­ing laws, and so on.14

To be suc­cess­ful, the venture-capital-funded “shar­ing econ­omy” will be forced to lose all those aspects of infor­mal shar­ing that makes “shar­ing” attrac­tive, and to keep those aspects that erode neigh­bour­hoods, erode employ­ment rights, and remove basic stan­dards. And if they suc­ceed, they will have used the lan­guage of shar­ing to bring about an unreg­u­lated, free-market, neolib­eral economy.

1 For exam­ple, econ­o­mist Arun Sun­darara­jan says that in peer-to-peer mar­ket­places “Rep­u­ta­tion sys­tems and active sup­plier screen­ing main­tain qual­ity” and that “These rep­u­ta­tion sys­tems take com­mu­nity enforce­ment up a few notches from the time of the Maghribis, com­bin­ing numer­i­cal scores and tex­tual feed­back with reviews, pic­tures, and peer ref­er­ences that are instantly vis­i­ble to any poten­tial mar­ket par­tic­i­pant. By mak­ing both prod­uct and trader qual­ity instantly trans­par­ent, this approach reduces the risks that often lead to mar­ket fail­ure”; PBS New­shour says that “Users trust each other accord­ing to a person’s accu­mu­lated social credit; user rat­ings thus form a cur­rency to increase the odds of find­ing a will­ing dri­ver”; and in a Jan­u­ary 2013 inter­view Airbnb CEO Brian Chesky said “Well it turns out that cities can’t screen as well as tech­nolo­gies can screen. Com­pa­nies have these mag­i­cal things called rep­u­ta­tion sys­tems”; Forbes writes that “Ebay’s much-duplicated rat­ing sys­tem bestows com­mer­cial cred­i­bil­ity on individuals”.
2 The descrip­tion draws from the work of soci­ol­o­gist Diego Gam­betta, who has spent years writ­ing about trust, and in par­tic­u­lar from a 2001 arti­cle writ­ten with Michael Bacharach.
3 The need for trust in col­lab­o­ra­tion is essen­tial to oppo­si­tion move­ments in author­i­tar­ian states, which is one rea­son I don’t believe that the “low trans­ac­tion costs” of social media were key to the Arab Spring upris­ings of 2011. I have made trust-based argu­ments here and more for­mally in this work­ing paper.
4 Method­ol­ogy: In a BlaBlaCar rat­ing, a reviewer gives a rat­ing to a revie­wee. The first step of the algo­rithm is to record all the rat­ings of an indi­vid­ual user. Each sub­se­quent step chose a new revie­wee by select­ing at ran­dom from the lists of review­ers, and then records all the rat­ings for that user. There are cases where a sin­gle reviewer has reviewed a sin­gle revie­wee mul­ti­ple times; these were dis­carded. Only the numer­i­cal rat­ing was recorded, and if a reviewer has not been reviewed by any­one a sin­gle “null” rat­ing was recorded. The pro­ce­dure is far from sci­en­tific, but at 190,000 rat­ings is way bet­ter than a few anec­dotes, which is the point.
5 Almost all the one-star rat­ings are given when the rider and dri­ver failed to meet with­out can­celling ahead of time. Per­haps the rider didn’t turn up, or per­haps the dri­ver didn’t, but one way or another the ride didn’t hap­pen. If a ride takes place, it is almost uni­ver­sal that each of the pair will give the other a five-star rating.
6 See also this ques­tion on Quora for impres­sion­is­tic responses.
7 In 2013, Airbnb claims over 300,000 list­ings, but a 2010 study showed that “more than 6 mil­lion vaca­tion prop­er­ties in the U.S. and Europe were being rented out to trav­el­ers for a fee at least two weeks of every year”. Ad Age reports that Airbnb’s 2012 rev­enue was $150 mil­lion, which a lit­tle arith­metic based on 3 per­cent host fees and at least 6 per­cent guest fees sug­gests it is part of $1.5B in trans­ac­tions. The same sur­vey esti­mated the global vaca­tion rental mar­ket at $85 bil­lion in 2010, at an aver­age of $14,000 per prop­erty. Sup­port­ing these esti­mates, a report by mar­ket research firm Pho­cusWright esti­mates the size of the over­all vacation-rentals mar­ket in the US alone to be $23 bil­lion, and vaca­tion rentals in Europe are much more pop­u­lar.
8 Per­sonal story: Dur­ing my child­hood my mother booked almost all our fam­ily hol­i­days from the Farm Hol­i­day Guide, which listed farm­houses that pro­vided either bed-and-breakfast or self-contained accom­mo­da­tions through­out the UK.This prac­tice was not unusual. For four years, we stayed with Miss Whit­taker in the Dud­don Val­ley in the Lake Dis­trict. She was a din­ner lady at the local school, and rented out accom­mo­da­tions to sup­ple­ment her income. She had a sheep­dog called Jan, which we played with in her back gar­den. When it rained, which was often, we would read books from her book­shelves: I first encoun­tered Miss Marple and Her­cule Poirot at Miss Whittaker’s house when I was about ten. My younger brother would some­times get up early in the morn­ing and sit in the kitchen with her, colour­ing in a colour­ing book while Miss Whit­taker started her day. Fam­i­lies I know in Ontario have rented out cot­tages for decades, through per­sonal con­tacts and scat­tered local tourism orga­ni­za­tions like this one.
9 BlaBlaCar hosts half a mil­lion pas­sen­gers a month, which sounds like a huge num­ber, but in the year 2000 12% of the 128.3 mil­lion work­ers in the US car­pooled to work (US Cen­sus) which, assum­ing two peo­ple per car, is around 7 mil­lion pas­sen­gers a day, exclud­ing non-work trips.
10 Per­sonal story: My own pre-digital rideshar­ing expe­ri­ences include a year of com­mut­ing between Hamil­ton and Water­loo, dri­ven by four different—and all wonderful—people, as well as sev­eral years of inter­mit­tent and luck­ily incident-free hitch-hiking in the UK, mainly between Leeds and London.Quiz for the reader: how would you rate the fol­low­ing drivers?
  • The Ger­man ex-prisoner of war who stayed in the UK and who gave me a ride in a small lorry, spend­ing much of the ride explain­ing in a friendly man­ner how Hitler had been misunderstood.
  • The dri­ver who spent 30 min­utes com­plain­ing about how dark-skinned immi­grants were wreck­ing the country.
  • The dri­ver of the empty lorry who explained that he was pick­ing up my friend Lawrence and myself on a windy day just to pro­vide some extra ballast.
  • The dri­ver who had cere­bral palsy and whose hand shook increas­ingly as he reached for the gear lever, explain­ing that he drove bet­ter when he’d had a cou­ple of drinks because he didn’t trem­ble as much.
  • Women may expe­ri­ence many kinds of prob­lem that I have never faced. What is the accept­able level of flir­ta­tious­ness and/or overt propo­si­tion­ing from a dri­ver? And would you report over-the-line behav­iour if the dri­ver had your phone number?
11 Per­sonal story: Rachel Bots­man also makes a point that the shar­ing econ­omy now means you can stay in unusual places, not just reg­u­lar hotels. When I was three years old my par­ents rented a con­verted rail­way car­riage near Aber­dovey in Wales. The jel­ly­fish on the beach were dis­gust­ing, but oth­er­wise it was fun.
12 Per­sonal story: my sis­ter and I tried to book an apart­ment in Rome in spring of 2013 through The owner said that apart­ment wasn’t avail­able, but another one listed on was free. We never met the owner, but we did meet a cleaner who had a key. The rental was clearly a busi­ness trans­ac­tion, and the apart­ment was one of six owned by the same per­son. The expe­ri­ence was very sim­i­lar to book­ing a tra­di­tional bed and break­fast or hol­i­day apart­ment through emails, web­sites, or even older mechanisms.
13 Thanks to Eric Far­rar for insights into the dif­fer­ences between sites like Kijiji and sites like eBay.
14 One of the legal chal­lenges faced by Airbnb is that local gov­ern­ments demand licens­ing of rental accom­mo­da­tion for rea­sons that go beyond the inter­ests of the landlord/host and tenant/guest to include the inter­ests of neigh­bours and wider com­mu­nity. Airbnb must cam­paign against such rules if it is to main­tain its growth. It is no sur­prise that the first two actions of the new shar­ing econ­omy advo­cacy group Peers are in sup­port of two of the best funded com­pa­nies: Airbnb and Lyft. The very first action was to campaign in the Sil­ver Lake neigh­bour­hood of Los Ange­les against restric­tions on short-term rentals. San Fran­cisco is the home base for the venture-capital wing of the shar­ing econ­omy, and claims of unnec­es­sary bureau­cracy are now run­ning into the claims of neigh­bours. In the Mis­sion dis­trict, local news­pa­per El Tecolote reports on how peo­ple are mak­ing large amount of money by “using a unit that’s rent con­trolled and they’re tak­ing the space away from peo­ple who really need a place to stay in San Fran­cisco” and gen­tri­fy­ing the Mis­sion neigh­bour­hood, push­ing native res­i­dents out.

September 29, 2013

Aucun commentaire:

Enregistrer un commentaire