For the particular problem that comment is discussing (automatic code generation), I suspect that the CS people were describing about a general automatic code generation problem, and the engineers solved a relaxation to that problem which was not in fact intractable.
In general, I don’t know how much I like the P-NP distinction. I hear from people who have been in the metaheuristics field for a while that until that became common knowledge, it was basically impossible to get a heuristic published (because you couldn’t provably find the optimal solution). But it seems like that distinction leads to an uncanny valley of ignorance, where a lot of people avoid problems that are NP hard instead of looking in their neighborhood for problems that admit polynomial-time algorithms. (For example, instead of “find a tour that is not inferior to any other tour” use “find a good tour” for the TSP.)
Right, I wanted to mention in the original comment that good-enough solutions to NP-hard problems are not, in fact, NP-hard to find. This is, of course, well known. But it detracts from the impact of the quote, so I left it out.
For the particular problem that comment is discussing (automatic code generation), I suspect that the CS people were describing about a general automatic code generation problem, and the engineers solved a relaxation to that problem which was not in fact intractable.
In general, I don’t know how much I like the P-NP distinction. I hear from people who have been in the metaheuristics field for a while that until that became common knowledge, it was basically impossible to get a heuristic published (because you couldn’t provably find the optimal solution). But it seems like that distinction leads to an uncanny valley of ignorance, where a lot of people avoid problems that are NP hard instead of looking in their neighborhood for problems that admit polynomial-time algorithms. (For example, instead of “find a tour that is not inferior to any other tour” use “find a good tour” for the TSP.)
Right, I wanted to mention in the original comment that good-enough solutions to NP-hard problems are not, in fact, NP-hard to find. This is, of course, well known. But it detracts from the impact of the quote, so I left it out.