Hey, thanks for the comment. I am aware that "correlation is not causation" and did not stated otherwise. The correlations would not answer the original question about the causation and it seems like this analysis cannot cope with it. Swift example suggests that causes of popularity of a given technology could be found outside Stack Overflow. It is, however, not impossible for Stack Overflow itself to be one of the causes of popularity of posts on Hacker News.

I disagree with this analysis.. from a non-statistician's point of view, a "popularity" is not the same as significance, and neither term has a time-aware component.

Said differently, the number of interested questions, over certain periods of time, concurrently with adoption and use of the platforms being measured, are being measured here.

Not as simple as "correlation is not causation"

Modern statistical theory has adopted techniques of causal inference from statisical data. There are some questions about the theoretical foundations but the stats folk by and large have run with it and feel perfectly comfortable talking about causality in certain situations.

http://www.michaelnielsen.org/ddi/if-correlation-doesnt-impl...

Yes, but all of them require a complete model of all possible influences and hidden factors. It's still only as good as the model, which in the OP is just "HN and SO".

If everyone ties their ego/identity/career/intellect to upvotes then what is voted up the most is accepted as most valid, most correct, or best - even if the individual does not understand all of the language or background material mentioned.

This can create a causal feedback cycle that establishes general trends in technology, because it's rooted in self identity and the appearance of intelligence (doesn't have to actually be intelligence).

Correlation is not causation but correlation combined with a system that intertwines (belief, personal motives/self identity/self image, crowd validation) can cause correlation to become causal.

That's why the author didn't simply look at correlation, but used Granger causality tests.

And then the author remarked that Granger causality doesn't indicate causation. It is a measure of how well one variable predicts upcoming changes in another. It doesn't mean that direct manipulation of the first variable will necessarily influence the second one.

Causation is an ill-defined concept. Already Aristotle had to define several kinds of causation, and philosophers still struggle with it.

It's even possible, per Hume, that causation doesn't actually exist.

Granger causality is basically just a time-shifted correlation. It doesn't prove causation any more than a correlation test (although it's directional in a way regular correlations aren't).