« June 2006 | Main | August 2006 »

July 20, 2006

Incentives the old fashioned way: hard cash for user-contributed content

TechCrunch posted a story that AOL is coming on strong with incentives for user-contributed content:

A little known Digg-fact is that a relatively small group of users submit a large percentage of the stories that end up on the Digg home page. Netscape, which recently relaunched as a Digg-clone, wants to pay those top users to switch over to them. Jason Calacanis, who runs the Netscape property, wrote a post earlier today offering to pay top Digg users $1,000 a month or more to switch to Netscape and submit news there instead.

I don't find this surprising, and I expect we'll see more of it. We already know that many "information portals" pay people to provide ripped off or unreliable content to draw viewers for their ads. And of course, paying writers for content which earns ad bucks is the standard business model for magazines and newspapers. Little surprise that "new media" newspapers (like the new Netscape) are trying the same incentive design.

Also interesting, but again probably not too surprising, that the majority of successful contributions to digg.com come from a small group of users: 20% of stories from the top 20 contributors; a bit over half from the top 100 contributors.

Posted by jmm at 05:33 PM

July 17, 2006

ICD to establish trust for transactions

Huberman, Wu and Zhang just published an article in Netnomics called "Ensuring trust in one-time exchanges: the QoS problem". (Folks at UM can access through the library's "Electronic Journals" access page.) They are interested specifically in the problem of purchasing from an IT service provider who does not have a verifiable reputation; this is a generic problem in transactions, but their modeling makes specific assumptions for this particular situation.

Summary from the paper:
In our model, a quality of service contract describes the likelihood that the service provider delivers the promised service. We have designed a mechanism that forces the provider to reveal his true assessment of the probability that he will be delivering a given service in a single interaction with a user/customer. We also solved the complementary truthtelling reservation problem of obtaining from the user his assessment of the true probability that a given level of resources will be required at the time of their delivery. In both cases, our mechanisms use a contingent contract to elicit true revelation of both QoS and likelihood of use through a pricing structure that forces the parties to make accurate assessments of their ability to do what they commit to.

They also apply the problem to situations in which service providers might overbook.

Posted by jmm at 11:10 AM | Comments (0)