I think never. Between the "ship early" ethos and SaaS becoming increasingly popular, software quality has been going downhill for quite a while. Most parties seem to be OK with it, so I expect it to get even worse.
Most parties seem to be OK with it, so I expect it to get even worse.
Are they OK with it, or do they just not see that they have a viable alternative at present?
I can think of numerous cases where from a personal and/or professional point of view I would have happily spent real money on upgraded/alternative software to what I've got if it fixed bugs that waste my time or make the results I get worse than they should be. Obviously some people will just rip off software whatever you do, but for paying customers, I'd be very surprised if quality alone couldn't drive a significant movement in a market, other things being equal.
I think it's all the unrelated things that aren't equal that are holding back that kind of competition. An interesting question is therefore at what point the willingness to use good software to develop more good software could cost less than putting up with poor quality incumbents, given that the real cost of both strategies is high. Even as a glass-half-full kind of guy, believing that a relatively small part of the industry could begin to pull that off without requiring the entire mainstream to shift, it would still have to involve far, far more people than are involved at the moment if we're going to create a sufficiently comprehensive foundation of development tools and essential libraries to bootstrap a whole quality-first ecosystem. The good news is that if you can establish that foundation, everything you do afterwards is easier in quality-first world has that advantage over the quick and dirty status quo, so momentum is on your side.
These two concerns aren't separate. You can publish agile software frameworks like rails with assurance guarantees based on proofs. If programmers play within the bounds, they can retain various proof-grade guarantees about the software they write.
I view "ship early" (and iterate!) as a superior verification strategy than formal methods for most typical software. Most errors occur much earlier in the pipeline than the authoring of the first line of code.
> Between the "ship early" ethos and SaaS becoming increasingly popular, software quality has been going downhill for quite a while
Do you have a citation for this? Having spent plenty of time working around bugs in the older era of software development which took years to fix, I think your statement reflects nostalgia or inexperience. Those long, deliberative development cycles weren't some bygone golden age – and any codebase from that era has the thicket of "#ifdef BROKEN_SUNOS_FEATURE" blocks to prove it.
