I think it's natural for new tools to be Propietary, while Free alternatives of the basics work their way up the chain:
Most of the cost of software development isn't in writing software — it's in the exploration of the solution space. Once you have settled on a good design, re-implementation is vastly cheaper and more streamlined than the original fumbling around in the dark. And sometimes it's better because you can jettison the legacy that comes from all that exploration.
So IBM employed Ted Codd and an army of engineers and salespeople, but now we all get to use Postgres.
What is being built today commercially that will be distilled into architectural principles and re-implemented as Free in the future? It's hard to know, but in hindsight it may seem obvious.
The article points out categories that were once proprietary — OSes, compilers, runtimes, clients, data stores. For example: DB2, System V, PCC, Internet Explorer. They were built at great cost — and remember they all had proprietary siblings that have since been abandoned, that also had to be paid for: OS/2, Itanium, Hypercard, DBase, various compilers, IDEs.
And then the few survivors were copied by open equivalents. System V gave way to Linux, DB2 to Postgres, IE to Chrome. A few never had proprietary equivalents AFAIK — I don't think there is a closed ancestor of Redis.
And sometimes the Free version hasn't fully arrived yet (x86), or it's just free (Google Docs), or it's doomed to remain not nearly as good as the proprietary tools (GIMP, desktop Linux). Or it's rocky (Linux vs SCO) or chaotic (BitKeeper to git).
(And it's no surprise that tools used mainly by programmers are more likely to have high quality Free equivalents than tools with a wider audience — after all, their users are capable of improving them directly.)
This is true - proprietary tends to lead - albeit it doesn't address the societal concern here: Are these really the only roles a software developer can have?
On one end, acting as a mercenary for platform monopolies doing the new stuff, and on the other, reproducing those designs without the same kind of paycheck?
I guess there is the third option of bilking investors by saying you'll definitely be a monopoly any day now and then open sourcing the whole thing.
My sense of it is that this particular phenomenon - the entire "hacker" arc from Unix in 1970 through the formation of the FSF to the present - was of an era, and the era is finishing up. Before that, there wasn't a software business to speak of, and after, the era of individual programs and proprietors is likely superseded by the needs of specific communication networks, which, like the Internet generally, everyone ends up standardized on, but no-one owns.
Part of why software is in a cynical state now is because the convergent network goal is ethically desirable, but the only way in which we seem to be capable of framing it societally is "someone owns this", so we have proceeded down a path of toxic corporate ownership, while everything else is a weird thing deserving of mockery.
Most of the cost of software development isn't in writing software — it's in the exploration of the solution space. Once you have settled on a good design, re-implementation is vastly cheaper and more streamlined than the original fumbling around in the dark. And sometimes it's better because you can jettison the legacy that comes from all that exploration.
So IBM employed Ted Codd and an army of engineers and salespeople, but now we all get to use Postgres.
What is being built today commercially that will be distilled into architectural principles and re-implemented as Free in the future? It's hard to know, but in hindsight it may seem obvious.
The article points out categories that were once proprietary — OSes, compilers, runtimes, clients, data stores. For example: DB2, System V, PCC, Internet Explorer. They were built at great cost — and remember they all had proprietary siblings that have since been abandoned, that also had to be paid for: OS/2, Itanium, Hypercard, DBase, various compilers, IDEs.
And then the few survivors were copied by open equivalents. System V gave way to Linux, DB2 to Postgres, IE to Chrome. A few never had proprietary equivalents AFAIK — I don't think there is a closed ancestor of Redis.
And sometimes the Free version hasn't fully arrived yet (x86), or it's just free (Google Docs), or it's doomed to remain not nearly as good as the proprietary tools (GIMP, desktop Linux). Or it's rocky (Linux vs SCO) or chaotic (BitKeeper to git).
(And it's no surprise that tools used mainly by programmers are more likely to have high quality Free equivalents than tools with a wider audience — after all, their users are capable of improving them directly.)