In terms of the ability to develop a technical understanding of computers and to use computers as a medium for certain types of creative expression, much ground has been lost over the last 25 years.
Mr. Torvalds may never have got Linux off the ground if the state of the world had been such that, as today, a desktop operating system that didn't support WebKit were not useful to anyone.
Even if you don't care about connecting to the internet and rendering modern DOMs, try building a desktop operating system from scratch for the raspberry pi and as soon as you want to interface with a keyboard you'll run into the fact that you have to do it through USB, the protocol for which will cost you months of your life to understand and implement.
This is not to say the USB and the modern web are unequivocally bad things, but they were designed/evolved without any consideration of certain kinds of concerns (both pedagogical and engineering concerns) and the benefits that they provide could probably have been had without imposing these particular costs on the inhabitants of the modern computing ecosystem.
Now, admittedly, most people (and even most people learning about computing) won't want want to build an operating system from scratch. But that's not the point.
The problem is that (my personal, and yours, and society's at large) innovation is stifled by the contingency of a project's survival or success on its ability to provide feature parity and backward compatibility with incumbent juggernauts. No serious compiler author is going to use your code generation framework unless it does everything that LLVM does. Nobody is going to join your social network unless their posts will reach everyone that their posts reach on twitter.
Is it a mistake to romanticize the idea that a person with less than a billion dollars could totally rethink a particular facet of computing and then execute on that vision and not have to give up, in the process, ostracize themselves from the rest of the computing ecosystem?
10-20 years from now, maybe 90% of everyone's interaction with computers will be via speech recognition and AR glasses, using devices and software infrastructure provided by Google, Apple, and Microsoft. These systems may be provided in a very closed way where the usefulness of the devices is dependent on connectivity to the proprietary cloud infrastructure of the megacorps and through which everything you say and everything you see is recorded. Maybe you'll have an idea for a similarly-useful personal computing system that works offline and doesn't invade the user's privacy. Or maybe their invasion of privacy won't be so egregious but you'll have some ideas about how to improve the UX in some way that's outside the range of your Apple device's configurability.
In either case, the sophistication/complexity of both the hardware and the software involved will be such that you'll need a TON of capital in order to make a competing product. There are basically two ways you'll be able to get that capital
1. Crowdfunding and voluntary, gratis, open-sourcy contributions. This requires wide interest in what you're doing from the public at large (and waning public interest in privacy is one of the things that the OP is concerned about).
2. Venture Capital, which means you need a plan to generate return on the investment, which means you either need to plan to sell your system at a high retail price, in which case it won't win any market share from the cheaper established systems from the existing megacorps, or else you need a plan to monetize the data of your users by sucking it all up into your own proprietary cloud.
I don't have enough time remaining in my day to make this rant shorter or more coherent, and I've spent too much time writing it to just leave it unposted. So I'm now going to post it in its current state.
In terms of the ability to develop a technical understanding of computers and to use computers as a medium for certain types of creative expression, much ground has been lost over the last 25 years.
Mr. Torvalds may never have got Linux off the ground if the state of the world had been such that, as today, a desktop operating system that didn't support WebKit were not useful to anyone.
Even if you don't care about connecting to the internet and rendering modern DOMs, try building a desktop operating system from scratch for the raspberry pi and as soon as you want to interface with a keyboard you'll run into the fact that you have to do it through USB, the protocol for which will cost you months of your life to understand and implement.
This is not to say the USB and the modern web are unequivocally bad things, but they were designed/evolved without any consideration of certain kinds of concerns (both pedagogical and engineering concerns) and the benefits that they provide could probably have been had without imposing these particular costs on the inhabitants of the modern computing ecosystem.
Now, admittedly, most people (and even most people learning about computing) won't want want to build an operating system from scratch. But that's not the point.
The problem is that (my personal, and yours, and society's at large) innovation is stifled by the contingency of a project's survival or success on its ability to provide feature parity and backward compatibility with incumbent juggernauts. No serious compiler author is going to use your code generation framework unless it does everything that LLVM does. Nobody is going to join your social network unless their posts will reach everyone that their posts reach on twitter.
Is it a mistake to romanticize the idea that a person with less than a billion dollars could totally rethink a particular facet of computing and then execute on that vision and not have to give up, in the process, ostracize themselves from the rest of the computing ecosystem?
10-20 years from now, maybe 90% of everyone's interaction with computers will be via speech recognition and AR glasses, using devices and software infrastructure provided by Google, Apple, and Microsoft. These systems may be provided in a very closed way where the usefulness of the devices is dependent on connectivity to the proprietary cloud infrastructure of the megacorps and through which everything you say and everything you see is recorded. Maybe you'll have an idea for a similarly-useful personal computing system that works offline and doesn't invade the user's privacy. Or maybe their invasion of privacy won't be so egregious but you'll have some ideas about how to improve the UX in some way that's outside the range of your Apple device's configurability.
In either case, the sophistication/complexity of both the hardware and the software involved will be such that you'll need a TON of capital in order to make a competing product. There are basically two ways you'll be able to get that capital
I don't have enough time remaining in my day to make this rant shorter or more coherent, and I've spent too much time writing it to just leave it unposted. So I'm now going to post it in its current state.