How we choose what to work on

We can run about four projects at a time. Choosing them well matters more than working faster.

Three questions at the start

Would the answer change what someone else does? Can we measure it with the instruments we have? Is there a version that takes three months rather than three years?

One question at six months

If we knew then what we know now, would we start this? Twice in seven years the answer was no, and stopping was the right call both times.