Intel Corp (INTC)

[Tenchu]

Elmer,

We can't simulate more than a few seconds of CPU operation and you put faith in climate models that project out decades/centuries?

Here's the problem. We need to put our faith in simulations that only amount to maybe 15 minutes of CPU operation before the design makes it to silicon. Hence we NEED to make sure the models are as accurate as they can be. If they aren't, then the chip will come out of the fab DOA. Goodbye schedule, goodbye millions of dollars of R&D, goodbye competitive advantage.

Climate models, on the other hand, don't need to be accurate. They just have to present a pre-determined narrative in fancy new forms. If they turn out to be wrong, no big deal. We'll just continue tweaking the model, tweaking the variables, until we get something that we feel can be useful.

It's a lot like stock market analysis. There is a ton of computing resources dedicated to predicting the markets, and yet they are no better at telling me where the S&P 500 will be a month from now than a Chinese fortune teller here in the Bay Area.

Tenchu