Comment by adsfqwop | Hacker News Reader

Comment by adsfqwop | Hacker News Reader

Parent Full thread

Unfortunately the human factor is exactly the problem with these types of systems. I don't think humans can ever be considered to be safe stewards of anything where a seemingly remote cascade event can wipe out life on the planet.

Look at the disaster of Long-Term Capital Management from the financial industry as an example, where a presumed ten sigma event almost wiped out the whole western financial system.

http://epicureandealmaker.blogspot.fi/2007/05/11000000000000...

These nuclear reactors are built with the same underlying principles and risk-management profiles. There's always the element of human error; you can't over-engineer something you did not foresee. In addition to that you have the enormous profit incentives from running nuclear power. These incentives can overwhelm the incentives for safety, or lead to disregard of warnings where profit potential would be compromised if warnings were attended to.

It's human, therefore it's hard to ever sleep completely sound knowing that these constructions are running on the planet.

Fukushima didn't just happen because one mistake was made either. It happened because a long string of mistakes and disregard of safety happened. Want to bet it doesn't happen again? It's hard to bet against human nature...

http://carnegieendowment.org/files/fukushima.pdf

"cascade event can wipe out life on the planet"

That is like saying that exploding all the ice cream factories would lead to continents being covered in continental ice sheets.

Yes, there are some risks - but they are very local, and do not last a long time. I would claim a lone oil tanker is an equivalent environmental risk to a nuclear plant. And look how much oil production we have operating in the oceans.

I think it's fair to say, we've accepted certain level of risk to get our power.

Please don't overstate risks. We are living on a planet whose oceans are dying, whose atmosphere is being clogged by CO2 ... nuclear power is mostly a benevolent technology. It creates very little waste, whose side-effects we can control. Unlike, say ,CO2. Or an oil spill.

Yes, there are sometimes mistakes - but please understand - no matter how the media spins it (because disaster new sell) they really are not that horrific when compared to other things that happen constantly.

The world is already radioactive: People are constantly bombarded by radiation. Like a poster below already mentioned https://en.wikipedia.org/wiki/Banana_equivalent_dose

No single plant failure can wipe out life on the planet. Not even if all plants fail simultaneously and dump their entire cores into the atmosphere you'd wipe out all life on the planet. So please don't overstate the risks.

Now, obviously no catastrophic event in a nuclear power plant (or any other system engineered for safety) happens because of a single mistake. If it did that would be a huge error in the systems design. For a catastrophe to happen, several layers of safety systems need to fail at the same time.

I think its possible to engineer systems that are safe enough even in the presence of human factors. Modern reactor designs for example are a lot safer than what we had in Fukushima. But convincing you of this would probably take a lot more text than fits in HN discussion, so let's just disagree on this point.

> anything where a seemingly remote cascade event can wipe out life on the planet.

Seriously? This ludicrous hyperbole does not belong in a thread where others are discussing the issue in good faith.

>presumed ten sigma event

How are financial people even calculating that kind of rarity?

I remember seeing a recent article here on HN about a guy that lost a ton of (someone else's) money playing with oil contracts. He went on about how he diligently planned for "huge" failures, such as a drop of up to 30 percent in the price of oil.

I don't think it's fair to bring in finance. Natural disasters and nuclear reactors are far more well behaved and understood.

> How are financial people even calculating that kind of rarity?

They don't properly, and that is the problem. A lot of it had to do with creating models that do not reflect reality. They worked fine until extraordinary circumstances occur under which they break down. A well known mistake is assuming a normal distribution for financial returns. In reality this greatly underestimates the likelihood and magnitude of extreme moves that can cause catastrophic losses. Nassim Nicholas Taleb's book "The Black Swan" discusses this in great detail. The bottom line is essentially that a move more extreme than what you anticipated is going to happen eventually.

LTCM made the mistake of neglecting the impact of changing correlations. The thought they had outsmarted the market by eliminating risk from their portfolio. But in times of financial turmoil, correlations of risky assets tend to increase. If your carefully constructed and hedged portfolio does not take this into account, you may be exposed to much more risk than you intended. This is precisely what happend to them. The fund was bailed out by a group of large banks, and their bets did eventually pay off, but the temporary losses were far beyond what their models had anticipated.

> a seemingly remote cascade event can wipe out life on the planet

What is this cascade event?

Reply on news.ycombinator.com