Friday, March 30, 2007

Be Hospitable: Waiting for Answers

The last post I made dealt with the chief Christian value of hospitality and how places like CouchSurfing could possibly revive this spiritual imperative because they curb fear. On some mornings my preparation for the day includes watching part of a morning episode of ESPN. The other morning when ESPN shifted to commercial, Hilton ran an advertisement twice in the span of 5 minutes in which they referred people to their website titled behospitable.com. On the website people offer testimonials of times when hospitality was heaped upon them and how grateful they were for it. This really underscores the power of hospitality. It's a feel good site. So hospitality is good, now, let's look at the economic side of CouchSurfing.

(Approx.)Number of Hotels Owned by Top 10 Hotel Owning Companies in 2005: 27,000
(Approx.)Total Hotel Revenues in 2006: 100 Billion Dollars

Should hotels be scared of such a network as CouchSurfing? Probably, not. Mark and I anticipate that it would drive down the demand for hostels to the demographic of school aged backpackers but will have zero effect on business travelers that stay at the Marriott or Hilton. (As an aside, I wonder if the movie Hostel had any damaging effects on the demand for hostel stay?) While those statistics had nothing to do with that argument here is a definition that will be important to risk reduction analysis:

Trust Metric: A trust metric is a measure of how a member of a group is trusted by other members. Examples include: references that you would include on a resume (in this case it would be a potential member) or feedback that you give on a buyer or seller on internet purchases.

There were two questions that I submitted to the folks at CouchSurfing, when they post a reply I will post on the blog. Here are the two questions:

1. If somebody uses a voucher system in which one person vouches for another's credibility and the "vouchee" becomes a problem for hosts by breaking some universal code of behavior such as stealing, would the voucher be penalized?

2. Are you recording the traits of all of the failures between host and guest? What are the dynamics of such failures? Do you use that information to perform regression analysis to isolate the contributing factors? Regression analysis is a tool economists use to control variables and test any variable's individual impact on a correlation. That is what econometricians do, an example of who thinks that is important: American Express. Econometricians are in big supply out at their Phoenix office because they can analyze the characteristics of credit card theft and throw up red flags more effectively.

So that was really more than two questions, but I hope they get back to me. CouchSurfing wants to streamline all of its current trust metrics into one overall better metric. This is kind of like the NCAA BCS rankings in which many factors (strength of schedule, coach's poll, AP poll, margin of victory, stuff like that) contribute to one numerical ranking. But much like the BCS there is the potential for controversy.

Skim to the Next Paragraph if you don't like the Technical Side
Anytime you use ordinal rankings, like the BCS or VH1's Top 100 Most Shocking Rock Moments Ever, you run into problems of violating some reasonable principles for "what is best". Say for instance that there are three traits they were looking at for our guests: The Guest Cleans Up After Himself (We'll call this choice A), The Guest is Conversational (B), The Guest is Culturally Different (C). Then say that there are three people that vote on the characteristics they find desirable in a guest.

Characteristic Preferences of the Voters
1. ABC (Clean, Conversational, Culturally Diverse)
2. BCA (Conversational, Culturally Diverse, Clean)
3. CAB (Culturally Diverse, Clean, Conversational)

Because it violates the transitive property in mathematics where if A=B and B=C, then A=C we know this is not what we want. But there are all kinds of flaws within ranking systems like this. If you want to investigate further look at Arrow's Impossibility Theorem: http://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem
The same problems exist with scalar rankings like putting your guest on a rank 1 to 5, unless you really believe that the thing you rate as "5" is in fact "5" times better than "1". So all of this is to say, they have their hands full trying to find one overall metric. Maybe they should consider sticking with the metrics they already have but including regression analysis to isolate potential red flags, maybe they already are, but they haven't written me back yet.

If You Left, Welcome Back: The Bottom Line
CouchSurfing desires a steady scale they can point to in order to reassure someone that bad things will not happen. This is a move in the right direction but we must realize that there is no trust metric system that will eliminate everyone's fears. You can't account for psychopaths but you can account for many other things. Here is the the truth, to practice hospitality you are going to have to go out on a limb. The question will be whether the limb is the size of a twig or a trunk.

1 comment:

Unknown said...

What is sad, and is my biggest fear about the couchsurfing project, is that while it has literally taken a joint effort of hundreds of thousands of people to get moving toward the mainstream, it will only take one psychopathic killer to bring the entire thing crashing back down. The moment one person is killed for letting another into their home through this method, wide-spread irrational fear will quickly end all of the good will they built up. I pray this does not happen.