The radius of the observable universe is about 14Gly, not 14Gpc; only off by a factor of Pi (not exact, but a handy mnemonic), but still, like the old saying goes "Off by a factor of Pi is still wrong."

--MarkusQ

Only if there is no scrubber at the power plant. Modern coal power plants are quite clean.

Wrong. Scrubbers don't stop the release of radioactivity from a coal fired power plant. You'd essentially have to go to CO2 sequestration for that.

--MarkusQ

soft skills are perceived as more valuable in a manager than technical expertise. To me, that's something that's stupendously obvious.

I agree. Soft skills are perceived as more valuable than technical expertise. Further, your arguments have convinced me that you not only share this perception but do indeed think it is stupendously obvious. If we were having this chat in person I would offer to buy you a drink and suggest we play a diverting little game of chance I happen to know in which soft skills are more valuable that technical expertise.

-- MarkusQ

*sigh*

Let me walk you through this:

• Google made a major point of ensuring that managers had technical expertise
• If we assume that they (Google) were honest in reporting this priority, competent in executing it, etc., we can conclude that given an individual who was a manager at Google it's highly likely that they had technical expertise; that is, to a good first approximation, HasTechnicalExpertise(X) is true for all X for which IsManagerAtGoogle(X) is true.
• Google then took a survey of the people being managed, and asked them what was important to them about their manager.
• The resulting list of features was presumably finite, as they completed the survey in a finite amount of time.
• This might at first seem surprising, since there are an infinite number of things that might be said about a manager. However, a little thought shows that the most probable cause is that predicates that were true of (almost) all or (almost) none of the managers did not make a serious contribution to the data. Note that this filtering could have occurred at any part of the process (if it was a "pick the most important" list, neither "drinks water" or "can fly" were likely to be included; if by chance they were, they would be unlikely to be chosen; likewise, if it was a free-form question most respondents would be unlikely to volunteer such observations).
• Therefore we should not expect to see common traits shared by all the managers as a strong component of the data.
• Specifically, we should not expect "has technical expertise" to be a strong component of the data.
• It was not. No story here.

-- MarkusQ

Older does not always equal less safe.

Chernobyl was built at a time when countries outside the soviet block who cared more about safety already had better designs. The problem wasn't that it was old, the problem was that it was badly designed. If you built a new Chernobyl style reactor today it would still suck even though it was brand new.

-- MarkusQ

most of the time I wish this wasn't true.

You're in luck. This is another case of #statisticsfail.

If all of their managers are selected to have deep technical expertise, it isn't going to correlate with success any more than "having two ears" will. This is a well known phenomenon called "sample bias" and is dearly beloved by everyone who wants to lie with statistics.

-- MarkusQ