The metrics don't include failures so infrequent that they aren't expected to happen in our lifetimes. That sort of infrequent failure was precisely what I was speaking of.

I think the thing that bugs me here, as usual with this topic is the adversarial and somewhat ignorant nature of the debate. The original research that mdsolar spoke of, may well be accurate. But I don't see his condemnation of that research (as undermining the credibility of the Nuclear Regulatory Commission) based on the actual content. That's more a problem with mdsolar's point of view than with the NRC.

As to your current arguments, I find them a lot better quality.

To elaborate, my view on this is that moving bytes just doesn't take that much energy. A heavily loaded network doesn't consume much more energy than a lightly loaded network.

In hindsight there are other issues. For example, you have to set up what you have on that SD. If you're downloading music on demand from the network, you can just change your demands in order to change what you hear. For an SD, you have to manage it, adding and removing stuff. Sure, you get more control over what you hear, but similarly, you get more work.

Finally, there's the matter of what happens when the SD gets destroyed. A server farm with backup and redundancy takes a lot less work to set up per person it serves than individual people managing SDs with redundancy.

And the "I" is India.

The obvious counter is that Germany was aggressive (both with a vast military buildup and multiple invasions and forced annexations before the Second World War officially began) and Japan was not keeping to itself (such as their invasions of Korea and China and their well-telegraphed war with the US and the UK).

And a military push by France in 1936 to reverse the remilitarization of the Rhine, would have been pretty easy. But having said that, it would be easy to underestimate Nazi Germany in 1936. They had over the course of a few years rebuilt a world-class military from the stunted post-Versailles remnants. I believe the Second World War would have been easy to prevent in 1936 or 1937, but I also believe that it wasn't that clear what course to take.

Still when one looks at the bizarre things that France did, such as building a vast and expensive defensive network of fortresses (the Maginot line), but not actually defending against the invasion path taken in the previous war (and which Nazi Germany used again), it's painfully clear that they weren't making good decisions even given the uncertainty of the times.

And do what? They have no power base outside of the Ukraine. And a few thousand "battle-hardened fighters" won't make much of a difference politically or militarily.

I'd look rather at the people surrounding Putin or domestic unrest.

Actually, it is a case of "in anger". The term has nothing to do with anyone's emotional state or any particular defense/offense scenario under which they might find themselves. "Used in anger" with respect to a weapon means that the weapon is being used as designed with intent to kill someone.

Since the already beaten enemy was still fighting.

It means that those Silicon Valley workers can fund their drug habit. The high stress part is motive. The high pay part is opportunity.

mdsolar indicates that he is referring to some estimate rate of 1 such accident in several million. But that sounds like a theoretical rate for a limited class of failure modes under ideal maintenance and regulation conditions. Can't say any more about that until I find out what he's speaking of.

But that is the case whether one downloads the music or buys a SD. The latter requires actual movement of physical objects, which I think is more costly relatively speaking.

Ok, where's the reference then?

I see when I googled, an estimate for "large" "loss of coolant accidents" around 5*10^-6 per year per plant. That sounds like your number. It's worth noting that the accident category in question hasn't happened yet since they're speaking of loss of coolant from pipe corrosion and mechanical failure in a plant with proper maintenance and the following of procedures, not the many other sorts of loss of coolant accidents that can happen to a nuclear plant (such as the real world examples caused by earthquakes, incompetence, and poor maintenance).

mdsolar stated that the NRC estimated the odds of a "nuclear accident" at 1 in 10,000 and then claimed that such accidents occurred at a frequency of 1 every 18 years. What I noted is that there are 435 reactors currently (according to Wikipedia) and that accident rate he claims corresponds to one such accident per 8,000 years of operation of the nuclear reactor. That is very much in line with the estimate.

This has nothing to do with "nuclear FUD websites". This is just rudimentary statistics.

So you're so concerned by my behavior that you post an unsubstantiated ad hominem attack anonymously? Do tell.

18 years*435 current reactors=7830 reactor years. Which is close to the claimed 10,000 reactor years of your original post.

How did the music get on that SD?