AirBnB Where are you?

Fig. 1 - Professionally bound neighborhoods

Web e-commerce has long been seen as a threat to traditional middlemen, but the threat it now poses businesses like cabs and hotels seem to raise the odds.  Uber takes the phone out of Louie the dispatchers hands, and AirBnB puts Web automation into the door-to-door process once known as 'can I crash at your pad.' Representing a new style of broker, the upstarts pose yet another threat level to those whose hegemony would be disrupted.

The role of the broker can rub at least two ways, maybe that is why it has been seen as a favorable position. History has seen brokers tend to favor one side over another in a transaction or two. Uber, for now, sets itself as arbiter of the cost of the ride – tho they might point to the black box if you ask them who decides. Uber could change its modus – it's still early.

How things could change may be seen in an AirBnB algorithm that has received some recent coverage. To hear tell, AirBnB saw what it could benefit if its customers on the providing housing side of the equation could come to a better estimate of the probable value of a night in their abode. While there are precedents in price advice systems from eBay and elsewhere, AirBnB claims somewhat convincingly that there approach is unique.  Still, they like some help with the estimator, so they have made it open source.

As depicted in an August 2015 IEEE Spectrum story, AirBnB set its secret sauce to percolating when the data science team began to calculating. Author Dan Hills writes that it converted the questions people ask when looking for a place to stay into machine learning algorithms. eBays' problem is different. With eBay, location is not really a factor. The timing is now, not 3 nights in October.  There's not a whole lot of difference between one good copy of Big Brother & the Holding Co.'s first Columbia LP and another.

AirBnB looked to create a tool that was dynamic,  that considered the unusual characteristics of a listing, and left room for human intuition when necessary.  I will classify that last bit as Surprise 1. Surprise 2 is that they use focus groups (no blind machine learning patriots, they) – and Surprise 3 is that they hired a professional  cartographer to hand draw accurate neighborhood boundaries for important world cities for travellers. [See Fig. 1]

I don’t know if this is a surprise.. but it is a bit amusing that author Hills previous gig was with a company called "Crash Padder" (itself bought by AirBnB). Amusing to this writer as he experience the crash pad experience first hand, coming late in the cycle when a sufficient number of people had been burned by thieving guests to put a serious lid on peace, love and  why not?

In Boston, and some other towns, Uber has already come to unwanted prominence for its capacity to bring on the wrong help. They and AirBnB will probably keep a team of data wonks busy for some time creating filters that bar the occasional felon flotsam from gumming up their march to ever higher evaluations. - Jack Vaughan, Boston

Related
http://spectrum.ieee.org/computing/software/the-secret-of-airbnbs-pricing-algorithm