I wouodn’t let them advance to positive times BP, until that can correctly get the CE, and BCE data calculated from Peter Humbug;’s model.

See http://seattletimes.nwsource.com/html/boeingaerospace/2009565319_boeing30.html and http://blogs.crikey.com.au/planetalking/2009/07/07/how-the-joint-strike-fighter-and-dreamliner-787-programs-are-compromised-by-similar-project-management-failures/ for some of the triumphs of computer modelling. “”"

As I recall, the 787 has yet to get off the ground (commercially) so don’t count that as a success. I think the 737 was a success; the 787 may never be.

And the JSF if they ever build any, will just be displacing a far superior aircraft.

But if you want to buy a product that resulted from computer modelling, you can now get the Logitech “Everywhere MX” mouse that will track (almost) everywhere; even on glass. I can pretty much guarantee that it will NOT work on the Hubble Telescope Primary mirror, even though they won’t let anyone try that; but it works most everywhere else.

George

Is this tacit acknowledgement that current climate models are not too good in their simulations?

“Nice one Roy, may I post your comparison steps elsewhere?”

Be my guest. Permission granted. ;^)

Reminds me of an “Idea for Design” in a famous (old) electronics industry magazine. Someone donated an hour of computer time on an IBM 1100 or some such to this “engineer” so he decided to do a Monte Carlo analysis, of the two transistor amplifier circuit he had “Worst case” designed to amplify with a gain of 10 +/-1 .

The results of the analysis said that the lead resistor of the main gain stage was the most critical item in the circuit. It also said that the spread of amplifier gains was skewed outside the worst case design limits (how could that be if it is a worst case design). So it suggested a 5% change in the value of that load resistor, to resore the gain to its desired value.

Whoop de do; the :engineer didn’t do the WCD properly; but more importantly it was a totally lousy circuit to use in the first place; and he could have rearranged the exact same number of circuit components into a much better negative feedback amplifier whose gain would be the ratio of just two resistors (easy to make 10:1 with standard values); and quite independent almost of the properties of either of the two transistors; a much more robust design.

So a fast computer can perhaps optimise a lousy design; or climate theory; but don’t expect it to come up with a better design or climate theory that a better engineer (climate scientist) could have figured out without the faster computer.

NASA isn’t going to statistical mathematicate its way to a better climate science explanation; so why the bigger computer.

The observations will be wrong, of course, and the model right. :-)

I find it “interesting” (other words that came to mind were far too coarse to post) that they plan to make model runs back a millennium, and forward to 2100. I need only ask: ON WHAT BLOODY DATA?!?!?!!?
I certainly hope someone without an axe to grind can access such high-res simulations so that we can get an independent assessment of how well the simulations replicate reality. I’m betting not very well. Current models do not necessarily do very well for hurricanes, so why should the GCMs.
It’s also nice to see that they have admitted never seeing a well-formed eye wall in prior simulations.
The key would seem to be for the modelers to work with the people who do field research and add computations for “natural factors” and see what happens under this new computational capacity, but I rather have doubts that will occur before I’m pushing up daisies … or the next ice age arrives.