Tuesday, October 11, 2005
Me vs The World and the Problems with Odlyzko's Law
So apparently a major financial institution has sent out a note about eBay and Skype using network econ to do some basic valuation. Funnily enough, these guys also show up on my referrer logs.
1) Look, I like the fact that my ideas have an impact. But *please*, at least credit me. I don't wanna sound whingey, but I can't even afford a new laptop. I'm just a young guy trying to make it. Don't rip me off; my ideas are my currency.
Next time, I'm going to start naming and shaming people.
2) They didn't get it right. Here's why:
Odlyzko's law uses a log term to dilute the value of possible network connections. There's a simple reason why this is not great economics. The value of nodes of a network is expected value. It's contingent; it's not today's value.
So clearly, I may not derive any value now from the Germans that were randomly calling me on Skype last summer; but I may in the future. That value has to be factored in; because it is, with some probability greater than zero, expected.
So how do we assign probabilities to these expected values? Odlyzko law essentially argues that they're assigned logarithmically. You could also argue that they're assigned linearly, whatever.
But the point to note is that in real-world valuation sitatuations, these probabilities are already assigned at something less than 1 in the usual venture/M&A discount rates that heavily discount the future value of resources like networks. So if you use Odlyzko's law, you're essentially discounting the value of a network twice.
I think it's far more accurate to assume the full value of a network based on it's ideal economics, just like we do with other resources in valuation, and then discount it at a rate that reflects the opportunity cost of capital.
These 'laws' are much better suited to analyze returns to scale - not to do valuation.
... if you name them ...
... we will flame them ...
Web2 justice is swift!!!
"Next time, I'm going to start naming and shaming people."
Next time? Screw next time. If they are taking your ideas and not crediting you then you need to get into their email boxes (preferably the VP of Research) and politely explain that you would appreciate either credit or compensation if your ideas are being used for the profit of others.
I see this as a good thing...you've just found a prospective customer. If you their contact info then send them an offer on future research reports that you can sell them, with more detail, and, of course, a license for them to redistribute or whatever terms you see fit. Just one way to get you close to buying your laptop =)
Although the comment above from gen kanai is good advice, and probably a good way to build some cred and relationships with financial institutions, I would send them a brief note indicating the "rules" here. Then I would offer to write a "thought piece" on your area of specialty. It makes everyone look good.