In the reading for a class that I’m auditing, we came across
Junseok Hwang and Choong Hee Lee, Agent-based Modeling for Differentiated Admission in P2P systems Using Evolutionary Game Theory Focused on Ownership Reputation, Second Workshop on the Economics of Peer-to-Peer Systems, June 2004
and there were some pretty surprising conclusions from it.
Pretending to be original can be as successful or more that being original.
And so only in a system in which hacking or faking data/information/identity is strongly punished or made costly/difficult is it ever viable to actually be original.
A system that rewards the creative winds up rewarding the luck creative and punishes everyone else.
This was what really blew me away. They make reference to this phenomena referred to as “reputation condensation” in which if people receive any sort of royalty – even “units of fame” – from their work even when they aren’t personally involved in a transaction that involves their work or a derivative thereof, then there will be a tendency for the wealth (fame as I interpret it – they use a generic notion of “reputation”) to coalesce around some of those that create.
At first I thought this was good “This means everyone can just be creative, and this is a dominant strategy, right?”. And then it was mentioned that since they assumed, and evidence bears out, that the Zipf distribution seems to be in effect, that not everyone will achieve enough fame – in fact, that most won’t – and so the reputation condensation will still be in effect.
This makes me think that since The Fractal Blogosphere is on an exponential scale, to make it fit with the Zipf distribution, it would need to be on a double exponential scale. So not 10, 100, 1000, 10,000, 100,000, but rather more like 2, 4, 16, 64, 65,536, 4 billion + (which are roughly the powers of 2 of the powers of 2 – that is, 2 raised to each of the following, which is a power of two: 1, 2, 4, 8, 16, 32).
This presents a much bleaker picture, unfortunately. Let me (mildly parodying the original Fractal Blogosphere post, suggest the following interpretations for those numbers:
2: Me and my buddy.
4: Me, my buddy, and some other folks.
16: I’m the king of the party. I know it is a small one though.
64: Ah! I have a readership! I need to be responsible to them or I’ll lose them.
65,536: I’m famous enough to likely get a comment on any post, but not famous enough to expect Joe Shmoe, or even Joe Blogger, to know my name.
4 billion+: I’m famous.
Unfortunately though, when you look at the area under a Zipf distribution curve and you try to integrate it up, you find that the only way that The Long Tail can promote a Content Renaissance On The Web is when you effectively have zero-overhead. Open-source and open-processes built on standards enable automated solutions that provide effectively zero-overhead techniques. This is why it is really only the Web 2.0 people like O’Reilly that can really put themselves into a situation where the long tail can be monetized.
To support the long tail, you must support the open-source, open-process, open-business model that is core to these Web 2.0 companies, because it allows for effectively zero-transation overhead. Otherwise, reputation condensation and the Zipf distribution results in an unmonetizable long tail.