For people reading this stuff, always search for something called the Odds Ratio (OR), which is a gold standard way of comparing how clinically significant a thing is. For example, smoking has an OR of 60 for lung cancer. A well covered paper on birth control suggested its OR for developing depression is 1.26, and the massive aspirin on heart health study had 0.95 OR of cardiovascular disease in one of the surveyed counties (ie aspirin improved clinical outcome).

If error bars on measured OR contain 1.0 you can’t say if it is helpful or harmful. OR accounts for baseline incidence in a way that RR (relative risk) does not. So it lets you actually choose among behaviors instead of within diseases or outcomes. It informs why it’s possible to tell everyone they shouldn’t smoke (OR is very high), keep taking birth control (because depression outcome is better than unwanted or complicated pregnancy outcomes), and aspirin for cardiovascular disease prophylaxis (very limited side effects, very cheap so even 0.95 is worth it).

I don’t know if there’s a consensus that electromagnetic radiation from electronics has clinical significance. If there is clinical significance it will be a small OR for most outcomes. It won’t be comparable at all to e.g. smoking.

However this is still an imperfect measure. In Japan OR for lung disease from smoking is 15, 4 times better than the US. The researchers speculated this has to do with a specific filter on common Japanese cigarettes but it could be genetics or other things that they didn’t control for (ie BMI and gender matching but not Japanese ancestry). So technology can have huge effects on clinical outcomes but 4x 15 is a much bigger deal than 4x 1.015. My point is that it has its limitations but OR definitely measures something real.

This is a robust framework for evaluating how this research matters to you in a way that does not require conspiracies or investigating the researcher’s backgrounds. Additionally it is widely accepted by legitimate medical practitioners across the world.

Under this framework, the reviews posted here do not document clinical significance, although individual papers cited in them might.

A big limitation is that mice are tested with orders of magnitude higher incidence rate of the disease to quickly find RR (high RR in mice vs placebo is strong research evidence just not strong clinical evidence). So mouse OR isn’t often predicative of human OR. But this applies to chart reviews too. Still a pretty durable framework.

It's an interesting comparison with the vaping thread last week or the week before, where a similar number of comments came to the conclusion that it should be heavily restricted because we don't have enough data.

The article you linked second, is interesting. He makes some bad claims (The electric fields produced by muscles are strongly polarized contrary to his claim that they are unpolarized) and there is some questionable math, but what I really want to highlight is how he uses the pulse repetition interval in his section on forced oscillation. By his own math, a CW signal at mm-wave frequencies produces no biological effect being to low of intensity. The formula given is proportional to 1/(2pif) so it follows, but he then uses the PRF to set the frequency and displacement and finds that at these low frequencies, the effect is strong enough to be dangerous.

The problem with this though, is that the PRF is how often you activate the transmitter and not the frequency of the transmitted RF signal which leaves us with the perverse conclusion that, mm-wave is fine under normal circumstances but becomes more dangerous the less you are exposed to it.