,
8 tweets,
2 min read
Read on Twitter
As scale increases, alignment decreases. Eventually a sufficiently disaligned organization degenerates into infighting.
We can turn this into a mathematical model that will predict the optimal size of a network, and the point beyond which network defects dominate.
We can turn this into a mathematical model that will predict the optimal size of a network, and the point beyond which network defects dominate.
Metcalfe’s law is based on the idea that every node gains value from every other node. Hence N^2 utility.
Odlyzko argues that it’s more like N log(N), as most nodes don’t contribute value to other nodes.
But neither model predicts a network size beyond which utility declines.
Odlyzko argues that it’s more like N log(N), as most nodes don’t contribute value to other nodes.
But neither model predicts a network size beyond which utility declines.
However, if we think of network participants as playing a game, we can forecast an optimal network size.
Specifically we look for the point at which a coalition can achieve a better payoff from defecting than by collaborating with all other players. en.wikipedia.org/wiki/Cooperati…
Specifically we look for the point at which a coalition can achieve a better payoff from defecting than by collaborating with all other players. en.wikipedia.org/wiki/Cooperati…
Suppose we have N employees, each of which receive both salary & equity.
At small scale, salary is low while equity stakes & upside are both high. At large scale, these flip.
At a certain point coalitions may form to gain more salary at expense of others, vs playing for upside.
At small scale, salary is low while equity stakes & upside are both high. At large scale, these flip.
At a certain point coalitions may form to gain more salary at expense of others, vs playing for upside.
Need to do math to confirm, but at large N the positive sum game may become a zero sum game.
Idea is that if there isn’t much shared upside remaining because the entity is mature, it may be economically rational for 51% to vote themselves more salary at the expense of 49%.
Idea is that if there isn’t much shared upside remaining because the entity is mature, it may be economically rational for 51% to vote themselves more salary at the expense of 49%.
One high level distinction is between communication networks (telephone, email) & cooperation networks (social networks, cryptocurrencies).
The presence of global shared state and statistics (likes, followers, BTC) is what turns a tool into a game. You can now win or lose.
The presence of global shared state and statistics (likes, followers, BTC) is what turns a tool into a game. You can now win or lose.
A consequence of this line of reasoning is that leaders should focus on creating, quantifying, and communicating alignment as much as possible.
It’s complementary to daily management. Alignment is why people do things even without assigned todos.
It’s complementary to daily management. Alignment is why people do things even without assigned todos.
Alignment isn’t just money. There are at least three macroscopic ways to coordinate human beings.
1) Incentives: shared economics
2) Inheritance: shared genes
3) Ideology: shared beliefs
Respectively quantified by game theory, Hamilton’s rule, and ideal point estimation.
1) Incentives: shared economics
2) Inheritance: shared genes
3) Ideology: shared beliefs
Respectively quantified by game theory, Hamilton’s rule, and ideal point estimation.
