[benjaminjosephw]

> The reality is that software is extremely expensive, especially polished software with a good user experience that's usable by non-experts.

I get your point here and I think this has been the case for so long that it seems like an immutable law by now. But, is it inevitable that software should be so expensive to produce?

How we write software today is largely based on ways of working and technical limitations that are not much different to what Fred Brooks wrote about in the Mythical Man Month. Brooks had some hope that programming languages would raise the level of abstraction we work at and that software design would shed much of its "accidental complexity".

Better programming languages could enable authors to work on problem solving rather than generating artefacts for machine computation.

The promise of better languages has been with us for a while but I'm not convinced that this avenue is as well explored as some believe it to be. The scope for these kinds of new abstractions isn't just drop in replacements for the programs you might write in C - it extends to other flavours of programming.

An example of progress here is how component libraries are used in web UI development. Mature component libraries require very little work to use and massively speed up development of "polished software".

Declarative end-user programming isn't a lucrative problem domain, but innovation in this space is still possible and could change the face of free software (both definitions) for everyone.

Progress like this could enable user communities to build, maintain and run their own platforms without the level of expense that currently prevents these kind digital commons from forming.

[api]

There was a push in that direction in the 1980s and 1990s in the form of Smalltalk, JITed super-portable and somewhat Smalltalk-inspired languages like Java and C#, WYSIWYG GUI design tools like old school Visual Basic (terrible language, but UI the designer is still unequaled), highly productive low-code systems like Hypercard, and so on.

Then we threw all that in the trash and went back to bespoke architectures, brittle un-portable OS-specific (and even OS-version-specific) compiled binaries, and of course the gigantic pyramid of hacks that is the web.

It's a classic case of "worse is better."

My own view on "worse is better" is that it's a result of the same phenomenon I'm alluding to in my parent posts: people want free-as-in-beer stuff. When people invest tons of time and deep thought into a platform they generally want (or need) to get compensated for that. The vast majority of the stuff I listed above was commercial or closely linked to commercial efforts and had commercial, "source available," or at least less liberal sorts of open source licenses. Meanwhile the pile of crap was free, unencumbered, and could thus be copied and cloned at will.

It's not just cost either. It's also friction. Having to pay for things and juggle licensing is a pain in the rear. You don't have to think about free. You just get it and run it. Low friction results in faster viral spread and speed wins.

You get what you incentivize, and you don't get what you don't incentivize. We do not incentivize quality.

Edit:

The same phenomenon is now taking hold in the news media. Quality news and fact-checked information is starting to cost money. Bullshit and propaganda is and will always remain free.

[CRConrad]

> WYSIWYG GUI design tools like old school Visual Basic (terrible language, but UI the designer is still unequaled)

It was equalled and surpassed only a few years later, by Borland Delphi.