We could perhaps get better software, but at what cost?
One of the nice things about engineering disciplines is that costs can be reasonably forecast. There are spectacular counterexamples (e.g. Big Dig), but bear in mind those projects were much more than "just" engineering projects.
Software costs seem hard to forecast. Software integrates with other software. Software is created via a whole pantheon of algorithms. The "better" algorithm is often context dependent. That context depends on what else gets integrated. Did I mention we probably won't know at the design phase how things will be integrated?
Other engineering disciplines integrate with things that are well-known: the ground, water flows, the atmosphere. These things don't change. This is nice.
This article makes it sound like "better" is just a matter of spending more. My read is different: spend more and be correct. Correctness _isn't_ just something we know but aren't allowed to implement. In many cases, we don't _know_ what correct is. We have to _find_ correct via exploration and experimentation. Even when we do, we might severely misjudge the effort required to get there.
This complexity isn't just accidental. The world is a complex place. Our human minds act on a small portion of that complexity, based on all sorts of heuristics we aren't even conscious of. They work most of the time. The rest of the time? We're compelled to shoehorn things into our heuristics. This ends increasingly poorly the more complex the underlying mechanism is.
This is a hard argument to make to someone who will not accept that they are not correct. Correct in their criticism. Correct in their diagnosis. Correct in the the behaviors that lead to our condition. Correct in asserting that avoiding those behaviors would result in no unanticipated consequences. Resentful about it because they were correct, and we didn't accept that.
One of the nice things about engineering disciplines is that costs can be reasonably forecast. There are spectacular counterexamples (e.g. Big Dig), but bear in mind those projects were much more than "just" engineering projects.
Software costs seem hard to forecast. Software integrates with other software. Software is created via a whole pantheon of algorithms. The "better" algorithm is often context dependent. That context depends on what else gets integrated. Did I mention we probably won't know at the design phase how things will be integrated?
Other engineering disciplines integrate with things that are well-known: the ground, water flows, the atmosphere. These things don't change. This is nice.
This article makes it sound like "better" is just a matter of spending more. My read is different: spend more and be correct. Correctness _isn't_ just something we know but aren't allowed to implement. In many cases, we don't _know_ what correct is. We have to _find_ correct via exploration and experimentation. Even when we do, we might severely misjudge the effort required to get there.
This complexity isn't just accidental. The world is a complex place. Our human minds act on a small portion of that complexity, based on all sorts of heuristics we aren't even conscious of. They work most of the time. The rest of the time? We're compelled to shoehorn things into our heuristics. This ends increasingly poorly the more complex the underlying mechanism is.
This is a hard argument to make to someone who will not accept that they are not correct. Correct in their criticism. Correct in their diagnosis. Correct in the the behaviors that lead to our condition. Correct in asserting that avoiding those behaviors would result in no unanticipated consequences. Resentful about it because they were correct, and we didn't accept that.