evolutionary economics and open source

a closed world

We start out with a simple closed-source closed-world model. Firms' sales are driven by their technology advantage. Innovation is proportional to a firms' size (bigger firms create more innovation), then multiplied by a random factor (innovation is unpredictable).

After a few hundred runs, this model tends to be dominated by one big firm. You can almost hear the managers of the succesfull firm congratulating themselves in the board room on their genius and cunning. However, after a while another firm claims market leadership.

Just looking at a running model for a few minutes is strangely sobering. Fortunes wax and wane, people strive and succeed and fail. And it's all just random. Lovely!

use the source, Luke!

Next, some guys start sharing innovation. Introducing just a handful of firms that share a small percentage of their innovation, is sufficient to turn the model results on it's head. Crucially, the open-sourcers share innovation only with eachother, i.e. with firms that reciprocate the sharing.

Because the open source firms share innovation, innovation accelerates. Or, from a closed-source competitor's viewpoint: technology obsoletes faster: closed-source firms lose technology market share.

Though the open source firms also compete amongst each other, as a collective they outpower the closed-source competition. It may take a few hundred world iterations, but then one of the open source firms hits the innovation jackpot and from there on the open source firms as a group gain market dominance.

hey, but that's cheating!

No, it's called changing the rules of the game These open source firms not only innovate in technology: they also innovate the business rules by sharing innovation credits that no hard-nosed "normal" firm would think of sharing.

Put differently, the open sourcers discover the value of community.

try this at home

Here's the model in a python script for download. You can play around with the key parameters and inspect the algorithms. Results are quite robust across a range of parameter settings. Have fun!