I guess it's official............the British have nuked USA's west coast.

I'm not surprised that the USSR and the USA did so many tests... but why the hell did France need to conduct 210 tests?

Up until computer modeling replaced actual testing every nuclear power had to periodically test new weapons and old weapons to make sure they actually worked.
They also tended to do tests to show off to everyone else that their nukes still worked.

But computer models never work. Right?

Well, because of the data collected in the several hundred controlled tests there's adequate data to construct computer models that can be reliably shown to demonstrate the performance of nuclear devices. And, still, there may be a need for actual tests in the future to determine the reliability of the computer models.

A nuclear explosion takes a few seconds, at most, to complete.

Modeling an eons-old and extant global climate with fewer sensors than were used to chart any given controlled nuclear detonation and then expecting to be able to accurately model said climate is absurd.

Were we to have the data from several hundred planetary climates to develop a climate modeling system then you'd be comparing apples to apples. As it is, you're comparing nuclear weapons to climatology and, surely, you see a logical disconnect here.

No, I see the word "computer model" and I realize that any computer model is a summary of the knowledge and biases of the person who wrote it - that's because part of my JOB is working with computer models that *try* to forecast financial trends.

Over the time I've been doing this I've observed that optimists write models that tend to be bullish and pessimists write models that tend to be bearish. In tracking the actual discrepancies in model performance from actual market performance I've proven this to management over and over again. I am a BIG enemy of automated buy/sell triggers because the criteria for those sometimes huge transactions are invariably the summary of someones biases.

You haven't lived until you've seen a market trend that was started by one of your own firms automated systems that forecast a trend and then initiated the trend by acting on the forecast information and then compounded the error by seeing the resulting market impact as a trend independent of the original action. Then the damn fool thing tries to order MORE costly stupidity.

So, yeah, I cringe when some asswipe uses a computer model to assert a conclusion about damn near anything.

And the orbits of satellites and planets were being calculated with astounding accuracy by the Chinese and the Maya long before the advent of computers so all a computer model does with these subjects is parse known information to produce known conclusions.

It's not hard to get a computer to agree with you.

It is another thing to get it to gin up prophecies.

If this guys computer model can conclude that human-caused global warming is responsible for temperature increases then I want to get my hands on that model so I can use it to forecast next weeks stock picks.

And, still, there may be a need for actual tests in the future to determine the reliability of the computer models.