In a more civilized era, Ross and I would settle our differences privately, but as this is the era of the weblog, we can now squabble in public.

Of my post about LinkedIn, Ross says:

Clay observed a power-law pattern emerging within LinkedIn. This is a temporary phenomenon, as the system's constraints work against preferential attachment.

I don't believe this for a second. I think we will always see a power law distribution on LinkedIn, because preferential attachment works in two directions. The smartest guy I ever knew in the ad business (like being the tallest dwarf, I know...) said, of managing people, "Whatever chart you put on the wall goes up." People will, even if unconciously, optimize their behavior to maximize whatever numbers they see being broadcast in public (c.f. slashdot moving its karma system from numbers to labels - Terrible, Neutral, Excellent, etc - to discourage karma whoring.)

LinkedIn sorts searches by number of connections, so if Joi Ito, currently the best connected user, wants to stay on top, he has to bring lots of people into the system, and those people will come in connected to fewer people than Joi is -- his position as the head grows by extending the tail. Even if these new users grow in connectivity, they are so far behind they will never even make the top 10. Its Joi's preference for connectivity that's driving the imbalance.

It's counter-intuitive to believe that as a social system grows, its connection distribution becomes more imbalanced rather than less, but that's what happens, at least in free systems. So, to make this sporting, I propose the following bet:

If, in three months time, LinkedIn is still sorting searches by degree of connectivity, I predict that a search on 'Internet' will yield a wildly unequal distribution, with at least the following characteristics:

1. The person in the #10 position (currently you, Ross, but whoever it turns out to be then) will have less than 15% of the number of connections of the person in the #1 position (currently Joi).
2. The top 20% of the result set will account for ~80% (+/-5%) of the links.
3. The average number of connections will be at least double the median number of connections. (For the record, I think this is the riskiest part of the bet.)
4. Neither the median nor mode number of connections will be in the double digits.

If I am wrong about any of these 4 particulars, I will buy you dinner at the Silicon Valley restaurant of your choice. If you accept, and if I am right about these conditions, you will buy me dinner at Blue Ribbon in Brooklyn. (I hope you are not a vegetarian -- they serve a wonderful paella.) Deal?