Comment by michaelt | Hacker News Reader

Comment by michaelt | Hacker News Reader

Parent Full thread

  Yet the patent possession/success correlation is hard
  to doubt.

How do you tell apart correlation and causality in studies like this?

I mean, it seems to me patents are at their most useful when you have to make large upfront research investments, which is mandatory to get your foot in the door in some parts of the tech industry (IC fabrication) - but in other parts of the tech industry people would prefer to test the market with an MVP, and may be able to avoid a large upfront investment at all, so investors would tolerate higher risks of failure (iphone apps).

So it's believable there could be confounding factors - how do studies control for them?

As far as I recall, most of these studies, especially [2], mostly just present their findings of a correlation, along with the data and any variables that may be relevant, without trying to prove causation. They may posit some tentative theories in their analyses, but are careful not to make any conclusions on those.

However, your observation about many software startups would prefer to release an MVP and iterate rather than invest a lot of upfront work and patent is also borne out by data in [1] and in other studies, such as some of Bessen and Maskin's papers.

I think Malandrew's sibling comment is also a very salient point, and combined with my hunch that a vast majority of startups (or, heck, software companies in general) are not doing particularly technically innovative work, it goes a long way in explaining the data in those papers.

Reply on news.ycombinator.com