Hidden Code: Why Do Companies Open Source Code and Publish Papers?

6 Dec

Hardly a day passes without a major company announcing the release of a new scientific paper or code around a powerful technique. But why do so many companies open source (via papers and code) so many impactful technologies almost as soon as they are invented? The traditional answers—to attract talent, and to generate hype—are not compelling. Let’s start with the size of the pie. Stability AI, based solely on an open-source model quickly raised money at a valuation of $1B. Assuming valuations bake in competitors, lots of money was left on the table in this one case. Next, come to the credit side — literally. What is the value of headlines (credit) during a news cycle, which usually lasts less than a day? As for talent, the price for the pain of not publishing ought not to be that high. And the peculiar thing is that not all companies seem to ooze valuable IP. For instance, prominent technology companies like Apple, Amazon, Netflix, etc. don’t ooze much at all. All that suggests that this is a consequence of poor management. But let’s assume for a second that the tendency was ubiquitous. There could be three reasons for it. First, it could be the case that companies are open-sourcing things they know others will release tomorrow to undercut others or to call dibs on the hype cycle. Another reason could be that they release things for the developer ecosystem on their platform. Except, this just happens not to be true. Another plausible answer is that when technology moves at a really fast pace — what is hard today is easy tomorrow— the window for monetization is small and companies forfeit these small benefits and just skim the hype. (But then, why invest in it in the first place?)