The premise of the study is that light is a flawed but easily obtainable metric that correlates with GDP growth. There are no doubt lots of other metrics that go into estimating economic growth when self-reported numbers can't be trusted. But those take money and expertise to collect, and are probably mainly available to intelligence agencies.

I agree with your skepticism of the method and it's good to explicitly list these things. But I think the authors of the research would also probably also agree that the method is far from perfect.

> when people move in to dense modern housing and shift to white collar work the model breaks down....more modern lighting is more efficient

These should apply equally to dictatorships and democracies right? Or at least it shouldn't correlate with the dishonesty of the regime so the model can factor it out.

> people increasingly socialize through phones

You still need light for most forms of economic development. I've been to a few places where it's almost completely dark at night and people communicate on phones. But the economic centers, for example where people congregate for night life, have lights on.