Then you don't understand risk.

There is no case in which the failure of a wind farm could cause death or injury to millions of people at once.

The whole "safety" argument of solar is buried in flat-out bullshit analysis about how people fall off of roofs while installing panels. There's no safety concern to speak of with solar/wind.

The issue of falls is enough to put the death toll from solar above the death toll from nuclear. It's just the solar deaths are spread out thin enough they aren't even going to make the local news, the nuclear deaths are concentrated and make worldwide news. (And note that neither Three Mile Island nor Fukushima involved any nuclear deaths.)

I didn't claim to be any kind of expert. But I don't understand how the scenario you are describing is necessary for solar/wind to have a higher risk per MWh. What does the presence of catastrophic failure modes have to do with past versus future results?

Let's put it in concrete terms.

I'm going to toss a coin, you call it. If you win, you get $5. If you lose, you pay me $1. You probably do it, right?

Next scenario. I'm going to roll a 100-sided die. If it comes up one, I'm going to shoot you in the head. Anything else, you get $100. Now, would you agree to that? Of course not. Even though you're 99% likely to get $100. It's that 1% chance of dying that stops you.

That's because "risk" isn't just one number. Even at a very basic level, we distinguish between the likelihood of the risk, and the consequences of the risk. You're only considering the likelihood, not the consequences. That's where you're going wrong.

> You're only considering the likelihood, not the consequences.

I'm considering the likelihood of the same consequence: people dying. You are describing a totally different scenario with your coin toss example. You are making it sound like I'm weighing profits against lives and that is not correct. I'm weighing lives against lives.

No, I understand what you're weighing, but you're still weighing it wrong. You're looking at individual deaths. What is the maximum number of deaths/injuries that can be caused by a solar power accident? Two, maybe? Now, what is the maximum number that can be caused by a nuclear accident? Millions.

Someone stumbling off a roof will not cause the evacuation of a city. Period.

The problem is you seem to think the bigger incident is automatically worse.

Mr. Evil is sitting in a tower. He has two weapons, a rifle and a rocket launcher.

Scenario A: He rolls a d100. If it comes up 1 he picks up the rifle and shoots at the first person he sees. He's competent, his target dies.

Scenario B: He rolls two d100s. If both come up 1 he picks up the rocket launcher and fires at the first group of people he sees. The shrapnel from the rocket will kill 10 people.

You live in the city he's in. Do you want him to roll one die or two?

Okay, let's imagine an alternative universe where home-sized nuclear reactor somehow got popular and people got used to it.

As a result, a typical city has millions of small nuclear reactors. Even though the government tries hard, it is impossible to regulate all of them to absolute safety, and every month there's a nuclear leak somewhere, in every city. As a result pretty much all mankind (well, at least the urbanites) enjoy(?) a permanently raised level of radiation and elevated rate of cancer.

Experts say millions die per year worldwide, but since nuclear is a part of our life, we have accepted it as a given and decided it's the price we're willing to pay. On the bright side, because each reactor is so small, each leak doesn't cause evacuation of a city: at worst the unit is locked down for a few months, and (because we've been so good at it) usually nobody dies as a result. At least no one's death can be conclusively linked to any particular accident.

So, in your metric, do you think this world is safer than our own Earth with large-scale nuclear plants?

...Now, note that this isn't just an idle fantasy: it's pretty much an exact mirror image of what we do with vehicular emission.

Right, that's the whole pro-nuclear argument: individual deaths are what matter, not individual incidents. What relevance do the individual incidents have when you only need 1 nuclear plant for every 500000 solar panels?[1] If we are talking about overall utility to the human race, then why shouldn't we be looking at the overall death count? I just don't see how the potential magnitude of individual incidents factors into that when the overall deaths have already been taken into account.

[1]: Based on an extremely generous estimate of 1000W per solar panel, and a fairly conservative estimate of 500MW per nuclear plant