Projects that push aside the envelope of what was previously possible and reach entirely new levels of performance tend to have not that predictable schedules and costs. I'm pretty sure the planners involved know that, but on the political level fixed ranges and estimates are required to get funding - so you get cost overruns and delays.

Also, they probably wouldn't have gotten funding at the start if they said it'd take 25 years and 10 billion. Starting with a low-ball and gradually ramping up has a greater probability of (funding) success.

And the fact that everyone knows this leads to some interesting game theory.

Every HN thread brings up the cost overruns, and I just…OK, smart guy, you come up with a reasonable estimate of shooting a telescope into a very distant orbit and then cooling it to 12 degrees above absolute zero.

To <7 degrees!

> … down to its final operating temperature of less than 7 kelvins

I guess the GP means US customary degrees rather than standard ones (also technically kelvins are not "degrees", the unit is simply K rather than °K, but as they share the magnitude with °C the difference gets muddled when talking about relative temperatures…)

No, the bureaucracy is not why it didn’t fail. Cost overruns and delays due to reworking stuff I can buy but I’ve seen tons of projects fail with plenty of bureaucracy (see the initial healthcare.gov launch).

No, cost overruns are generally a sign of things going bad, not 'just more money for double checking'.

The Gov. of Canada Payroll system is now up to $1B with no end in sight.

--> Payroll. System. $1B.

I'm stoked this worked though can't wait to see the photos.

Sure, but when we’re talking about NASA the context is quite different from random software “engineering” projects, or even random traditional engineering projects. For NASA cost overruns don’t usually seem to lead to cases of sunk cost fallacy, and their infamous late-90s attempt at “better, faster, cheaper” rather ended up demonstrating how you can have any two of those three, and arguably, in the case of “never done before” flagship projects, only one of them.

It is still true though that many times bad planning and management has been at least a partial cause of cost/schedule overrund at NASA, and they’ve received well-justified criticism for it from both internal and external auditors.

The phoenix project is going entirely according to plan, from the point of view of IBM/PeopleSoft.