"We might want to zip our car with their car..." We do or we don't. There is no "might". Spending money on "might" has been the death of many projects. If we

"We might want to zip our car with their car..." We do or we don't. There is no "might". Spending money on "might" has been the death of many projects. If we didn't, and now we do, we could write a fn to map the car to parts, or we could define the car struct in terms of its parts, or we could just do away with the car altogether. But far more valuable would be an analysis of what changed about the requirements that the model no longer works. Now, don't get me wrong: I'd love a better language, and by better I mean "as fast as assembly but 'dynamic'". The problem is that, at the end of the day, all compilers are just "premature optimizations" or perhaps "willing premature optimizations". We could all be happily programming in smalltalk or build a runtime using predicate logic, but a) the number of people who could program in it is vanishingly small and b) it would be fucking slow. These languages don't solve a problem that I have, or rather they don't solve a problem that I don't already have a far better solution for. They solve a problem that academics have.