Friday, December 5, 2008

Bad Models?

Further review of yesterday's MSL press conference surfaces even more disturbing revelations. As the Emperor pleaded the minions' latest $400M overrun case, he philosophized on why no blame should be assigned, at least not to him and his accountants.

So development programs have unknowns, we take that as given. The Emperor says you can't predict them, and NASA is in the business of doing cutting edge stuff, so Congress expects overruns. But, do they really?

You may ask, why not put reserves into the budgets of this projects? The real answer is that there is a pot of reserve, held at multiple levels within the Empire. But you can not individually put encompassing reserves in each program or there would be no resources to actually do programs. Some will do better, some will do worse, and hopefully what is kept in the side pot will cover.

Unfortunately, under the Emperor, working with inexperienced, or worse, incompetent, program managers, the balance tips in the wrong direction and reserves are called for throughout the portfolio. The misbegotten things called ARES and ORION have drained the pool and only the turds are left sitting on the bottom. Within its own little hot tub of Mars programs, the steroidally sized MSL and its Rube Goldbergian SkyCrane delivery system, has also left the planetary reserve pot bone dry. Europa's whales will be quite safe for some time to come.

But those unknowns of which the Emperor spoke have some statistics associated with them or he would not be directing the cost probability analysis that was mentioned. Apparently, the models aren't giving out good answers for uncertainties, resulting in the present situation. And this is why we should be afraid. Very afraid.

Engineers build models of systems and then test them for accuracy with real world examples. If errors are found, the models are recalibrated and, hopefully, will generate better answers the next time they are used. So, too, should financial models have been updated over the years. If the models are not attended to, things like MSL result.

Kind of makes you wonder how those ARES-1 models are coming along, don't it?


Anonymous said...

I continue to not understand why the sky crane concept is superior to that of the regular parachute-then-retrorocket system. Shit, the Phoenix did that - worked fine!

And how many kg of hardware is being flown away and dropped to support the sky crane? How many of those kg could be instrumentation?


R2K said...

Retro rockets really make me nervous. A large parachute to airbag system would be more reliable I have to think. But there is the extra mass plus tanks to inflate.