Sessions Top Ten Insights - Five

One of the insights from our Sessions event was from Yochai Benkler who said the following about peer production:
we do have very good research on how adding money undermines social motivations depending on contents
I am not sure how to intepret that last part - "depending on contents" - but Yochai's point that "adding money undermines social motivations" is something that ought to be watched carefully.
One place we can watch it carefully is Amazon's recently announced Mechanical Turk service, mturk.com.
Mechanical Turk is about peer producing HITs.
According to Amazon, a HIT is a Human Intelligent Task, ie something that humans do better than computers. Like identifying photos or filling out captchas.
If you need a human to do something, you send a HIT request (via the mturk API) to Amazon.
The HIT is displayed to the masses, who then complete them, and get paid for doing so.
This is an attempt to automate peer production and add a payment system on top of it.
So we'll see if this service "undermined" by the money involved.
In the meantime, I've suggested to a couple of our portfolio companies who have mundane operational taks that they'd love to outsource to check mechanical turk out. Let's see what kind of results they get.
Note: You can find the research Fred references on pages 321-328 of Yochai’s paper Sharing Nicely http://benkler.org/SharingNicely.html where he references the extensive work done by Bruno S. Frey among others.
November 5, 2005 12:24 PM, By Fred Wilson
Tags: amazon api peerproduction webservices
Comments (3)
Money has to be added somewhere for it to be business, right?
It's how it's extracted that corrupts.
Are you implying that to avoid "undermining social motivations", that only the company that fascilitates the commons should extract value from it?
Surely this is exactly wrong, not only lookign at mturk, but also judging by the eBay example (and goog (adsense), amzn (associates pre-mturk).
I think if you plan to trade in the product of the community, not just offer it a service, then it makes sense to be upfront about it - not doing so, and extracting value without sharing it, is probably the fastest way to undermine peer's contributions.
Posted by David Gibbons , November 8, 2005 01:45 PM
Hi David, I totally agree.
Clearly attaching $-signs to a community can quickly generate a lot of interest; but anything less than transparant financial dealing is asking for trouble.
Also the more cash (and kudos) that is fed into the community the greater the opportunity to build a strong and vital community of interest - even if that interest is driven by purely commercial motives.
It is interesting to take a snapshot of community interchange rates. eBay take between 2 and 5.25%, Visa and Mastercard take between 1 and 5%. These services clearly represent good value to both sides of the transaction within a highly competitive market place. mTurk have an interchange rate of 10% - which seems high, but they have quite a unique proposition at the moment. It will be interesting to see how it develops.
I think that we (Yellowikis) should aim to offer a zero percent interchange rate (where 100% of the revenue goes to the person doing the work.) - with a focus on transactions at a much higher value than mTurk (say $20 rather than $0.65 on mTurk). I'm sure that Yellowikis can make money from value added extras offered to both sides of the transactions.
So(in conclusion): Facilitating value exchange builds strong community; Greed in the community is the road to doom.
Posted by Paul Youlten , November 11, 2005 07:58 AM



I think what Yochai means is that while eBay and Amazon have a clear commercial imperative that contributors understand (and can benefit from) from the get-go; Other projects like Wikipedia or Flickr would lose a lot of support if they started paying for some contributions and not others.
'Plan A' must be to act as the intermediary - like Amazon does with mTurk. Putting requests for help and helpers together and taking a small % of the transaction.
Yellowikis will be testing mTurk as soon as we are able to.
Posted by Paul Youlten , November 6, 2005 04:06 PM