Space Review article

I think you both have a point.

Whilst most of those operations are probably damn near 99% chance of
success, even 30 lots of 99% is only 73%, and some of them are a lot
lower than 99% and these dominate the probability.

If the chances of success get much below 73%, what do you think the
chances are that the people with the money will sign off on it?
"Here's a hundred billion pounds; but what's that?" Man whispers in
his ear. "There's a 30% chance it's all going to fail? Oh carry on
then."

No, not going to happen. When a program gets that big, it has *got* to work.

On Wed, 09 Feb 2005 08:21:50 -0000, Xenophile wrote:
>
> Well. Let's see: 30 essential techs, and all must succeed, and if
> even one fails, then IT'S ALL OVER. And even if each has a 90%
> chance of succeeding, that's only 4% overall.
>
> OK, look, this is, strictly speaking, true. But it's just a number
> game. First of all, some things on that list have a much better
> chance of success than 90%. Such as # 14: Recruitment of the 10,000
> people. Does anybody think that there is only a 90% chance of
> finding 10,000 people for the job? Really. And how about # 16:
> Invention of the new technologies, equipment and processes required.
> But right after that we find # 17: Lunar operations, and # 18: Lunar
> prospecting, and so on. Don't both of these (and some of the others,
> too) count as "invention of the new technologies, equipment and
> processes required"? So some of these 30 essential techs are, in
> fact, the *same* techs, only phrased differently. That changes
> the 'thirty nineties equals four' equation considerably. Except that
> the whole 'thirty nineties equals four' thing is bogus, because
> engineering isn't a crap shoot with d10's.
>
> Look at it this way: I have thirty dice. These are ten-sided dice,
> of the type used to play such games as Dungeons & Dragons. Except
> that, in this case, the dice are not numbered. Nope, they are each
> green on nine faces, with the remaining face on each die colored
> red. In order to achieve "success," I have to cast them, and get
> them all to come up green. Now of course the chance of *any one die*
> coming up green is 90%, but the chance that they will *all* come up
> green is only 4%. Guess you'd have to be a fool to bet on me, eh?
>
> Ah, but if this game is a little more like engineering, and a little
> less like a casino, then I get to do something that would get me
> kicked out of Vegas. Suppose that, after casting my thirty dice, I
> look at them closely, pick up any that have come up red, and cast
> them again. Then I repeat the process again. And again. And
> again. How many turns do I have to be allowed before you'd have to
> be a fool to bet *against* me? That's how engineering works: you get
> to keep trying until you get it right... until you get it ALL right.
>
> Not to mention the fact that I'm loading the dice a little each time
> I make a new cast. Messing with the variables (in this case, dice)
> in order to bump the odds in your favor is unethical in gambling, in
> fact it's cheating. But it's unethical *not* to do exactly that in
> engineering. If I'm designing a new roof for a hospital, I'd
> *better* try to "load the dice" as much as I can, to make sure that
> roof doesn't fall in.
>
> Again, Sam Dinkin's math is good (I assume; I wouldn't be able to
> tell if it weren't), but what he does with it is silly. Engineering
> isn't a crap shoot played with d10's.
>
> Xenophile (not an engineer, not a mathematician, not a gambler... but
> come on)

--
-Ian Woollard

"You can't have really bright objects and really dim objects show up
in the same photo.
...Unless you're taking a photo of Bush shaking hands with Stephen Hawking"
- uberdave on why you can't see stars on the moon.