Peter Lind­berg writes:

The ques­tion that was asked in 1968 [at a NATO sci­en­tific com­mit­tee con­fer­ence] is just as valid now: “How can we pre­dictably cre­ate bet­ter soft­ware systems?”

This is a recur­ring ques­tion. But isn’t it back­wards? Isn’t the prob­lem rather how to dis­cover which soft­ware sys­tem would be the best in a par­tic­u­lar prob­lem domain, and within a par­tic­u­lar bud­get and timeframe?

And: just software?