**Take an idea immediately 30 years out to evaluate** If “it would be ridiculous if the idea weren’t possible 30 years out”, only then is it worth thinking about how to do it. The process is to “bring the idea back from the future”, with the first stage being about 10-15 years out. What could be done then? What would the software be like? Could we build a supercomputer to simulate what could be done by then? The answer is often “yes”. And this leads to short term plans for both SW and HW. If the software is thought about first—and partly simulated on existing machines—then the supercomputers can be optimized for the software (thus eliminating much viscosity and errors from busywork).
~
DE
Der Prozess besteht darin, "die Idee aus der Zukunft zurückzuholen", wobei die erste Phase etwa 10-15 Jahre in der Zukunft liegt. Was könnte dann gemacht werden? Wie würde die Software aussehen? Könnten wir einen Supercomputer bauen, um zu simulieren, was bis dahin möglich ist? Die Antwort lautet häufig "Ja". Und das führt zu kurzfristigen Plänen sowohl für die Software als auch für die Hardware.
See Ends Planning.