Comment by cauch | Hacker News Reader

Comment by cauch | Hacker News Reader

Parent Full thread

That's not at all the point.

Imagine the following game: you throw three dices. If the three dices are not ending on 1, you win 1 dollar. If they are ending on three 1, you lose 100'000 dollars.

If you throw the dice ten times, you will, with high probability, not have three 1 (the probability of appearing is 0.5% per throw). You will probably need to throw more then 10 times before loosing. But when loosing, you will lose a lot too, which makes the bet intrinsically a bad idea (the win does not justify the risk if unlucky).

France has 56 nuclear plants. 56! It's totally unreasonnable to draw conclusion on intrinsic safety of the nuclear plants on such a small number. The fact that you cover the full population with a small number of plants does not change that: it does not change the fact that you cannot conclude scientifically. The same way it is incorrect to say "I only need 3 throws, so it means that the 1-1-1 bet is intrinsically a good idea".

(PS: please do not answer about "loosing 100'000 dollars, but with nuclear disaster, it's not that bad", it's not the point. The point is that statistics is a large number game, and that we just don't have a large number of nuclear power plants to do statistics with them)

There are hundreds of nuclear power plants that have been in operation for decades. Most operating nuclear power plants run with dramatically higher uptimes than what they were originally conservatively designed to meet. Nuclear power plants are operating in an active war zone right now! Six reactors were the site of a military battle and ended up sustaining damage beyond anything foreseeable by the designers. Every one of them shut down successfully and at no point was there any evidence that there was ever a safety risk to the public unless a malevolent actor wanted to cause such an accident.

This body of evidence strongly supports the view that nuclear is not only safe but incredibly safe. We have incredible knowledge about how to run nuclear plants safely.

Again, you miss my point.

"hundreds of reactor in operation for decades" is NOTHING. These plants have a large number of differences, and are located and operated often very differently. At the end, you have something like "10 plants operated for ~10 years under conditions A", "10 plants operated for ~15 years under conditions B", ...

SIX! Six reactors were the site of a military battle and ended up sustaining damage beyond anything foreseeable by the designers. Six is NOTHING. You cannot say "I trust that nuclear is intrinsically well-design to resist to military conflict" based on SIX cases. The damage may have been lucky. Or even may have been unlucky (I'm not saying that nuclear is unsafe, I'm just saying the stats says you cannot conclude).

The only people who pretend we have incredible knowledge about how to run nuclear plants safely are either people who also demonstrate they don't have any idea of what they need to check to "demonstrate safety", or they are experts that are on the left side of the Gaussian curve: they are as numerous as as-much-qualified experts who pretend nuclear plants are proven very unsafe (in other word: they are cherry-picked experts, and do not represent the consensus which is: nuclear plants are too complex, too diversified, too sensitive to surrounding circumstances and not yet used enough so that we can scientifically conclude)

How do you reconcile this kind of thinking with driving a car?

Are you seriously thinking that someone is incapable to drive a car if they are not incorrectly convinced that an incorrect math computation is correct?

The large majority of things in life is "impossible to statistically prove". I had no proof that I will not burn my toast for breakfast this morning. Yet I had breakfast without problem AND I did not pretend that "since I've used this toaster 5 times and did not burn a toast, it implies that burning a toast will never occur" (this sentence is incorrect: the observations are compatible with the fact that my setting is very good and reduce the risk of burning the toast very low, but also compatible with another realistic hypothesis: my toaster burn a toast about 1 time over 10, and so far I was just "lucky").

But why would I not make a toast or take my car? I just don't know if it will work or not, but I also don't know if it will not work or yes. You cannot say "if you are not sure of X, you should act as if non-X is sure", because if you set Y=non-X, you will say "if you are not sure of Y, you should act as if non-Y is sure", and you end up saying you should both act as if both X and non-X are sure.

Nobody here is saying nuclear is not a solution (I mean: if you think I'm saying that, you are dead wrong). But it is tiring to see pro-nuclear people making pseudo-scientific claims to entertain their own belief (belief that may turn out to be right). If you are so confident in your belief, just say so: "I know we cannot properly tell that nuclear is 'safer' than X or Y, it does not make any sense to do so, but yet I believe it is still a good option. You are free to disagree with me, but you also have no ground to pretend it is 'less safe' than X or Y, so your position is as legitimate as mine".

Note that powering the world (all uses of energy, not just the grid) with nuclear will require ~6000 3 GW(th) nuclear reactors, to provide the 18 TW of primary energy the world currently consumes. About two orders of magnitude more than France has. At the historical rate of world nuclear accidents this would yield > 1 meltdown a year.

Global warming is 100% snake eyes when you get down to the bottom line of "why nuclear?"

Reply on news.ycombinator.com