Photo by Amanda Gyllenhaal

Photo by Amanda Gyllenhaal

There’s a bit of dis­cus­sion right now about a work­ing paper com­ing from Ser­guei Netes­sine and Tom F. Tan at Whar­ton that’s won­der­ing how solid the Long Tail effect really is.  A lot the crit­i­cism seems to come down to some definitions:

Ander­son is also author of The Long Tail: Why the Future of Busi­ness Is Sell­ing Less of More. The key dif­fer­ence between the opin­ion of the book and the study by Whar­ton researchers is how they define “hits” and “niches.” In the book, Ander­son focuses on the def­i­n­i­tion of hits in absolute terms such as the top 10 or top 1,000 prod­ucts, while Netes­sine and Tan argue that, to take grow­ing prod­uct vari­ety into account, one has to define pop­u­lar­ity in rel­a­tive terms, such as the top 1% or top 10% of prod­ucts, to prop­erly assess the pres­ence or absence of the Long Tail.

The ques­tion of absolute v. rel­a­tive def­i­n­i­tions can obvi­ously be looked at either way, but it seems to me that the real ques­tion is not how many total prod­ucts are avail­able (rel­a­tive) but how many prod­ucts are avail­able that would not be were Net­flix not shoot­ing for the niches.  That is, if we define a hit as the top 1% and 3000 movies are stocked by a stan­dard brick and mor­tar com­pany that isn’t capa­ble of the logis­tics of being a Long Tail busi­ness, then the top 30 movies are the hits across the entire indus­try.  For there to be a mean­ing­ful com­par­i­son between stan­dard and Long Tail you’d have to con­sider that Long Tail is based on the premise that inven­to­ries are expand­ing and that is one of the things it is look­ing at, not try to cal­cu­late the expand­ing inven­to­ries into the def­i­n­i­tion of hits and niches.  So I guess I have to agree with Ander­son on that one.

Of course, this def­i­n­i­tional ques­tion doesn’t change some of the very good points that the paper brings up about how the Long Tail effect is being used now.  The most impor­tant one to me is the crit­i­cal­ity of rec­om­men­da­tion sys­tems in a Long Tail busi­ness.  All those niche prod­ucts are just over­head if con­sumers don’t know they’re there.  Net­flix is obvi­ously aware of the prob­lem, given that the data used in this study was released by Net­flix as part of a mil­lion dol­lar con­test to improve their rec­om­men­da­tion sys­tem.  Based on my own expe­ri­ence as a Net­flix cus­tomer, I have to say improve­ment is sorely needed–though I might ques­tion whether the rec­om­men­da­tion sys­tem itself is the issue or the hor­ri­bly non-browsable inter­face Net­flix uses.  (Well, really inter­faces plural, since a large part of the prob­lem is how they bounce back and forth between dif­fer­ent looks depend­ing on how you get to the data…but that’s a dif­fer­ent discussion.)

It makes me won­der how much social rec­om­men­da­tions are actu­ally use­ful for Net­flix.  I don’t use that sys­tem myself, and it wouldn’t be vis­i­ble in the data used in this study which was just of rat­ings data, but it seems like improve­ments to the social tools used by Net­flix would pro­vide a far supe­rior rec­om­men­da­tion sys­tem to the algo­rithms devel­oped in the com­pe­ti­tion.  For me, the issue is the lack of con­trol that Net­flix gives its cus­tomers.  For instance, I don’t have any abil­ity to choose which movies I’ve rated or rented will be vis­i­ble to which friends in any sort of gran­u­lar way.  There’s no offi­cial inte­gra­tion between the closed “Net­flix friends” com­mu­nity and other social net­works, at least that I can find on Netflix’s site.  That alone would be incred­i­bly valu­able; the idea of social net­work­ing is to make the per­son the cen­ter of knowl­edge, not the net­work, and Netflix’s friends don’t allow that.

via Rethink­ing the Long Tail The­ory: How to Define ‘Hits’ and ‘Niches’ — Knowledge@Wharton.