You're right. Especially with the rise of agentic AI. You could have hundreds of contributors, all using agents, working on different modules, according to existing spec and tests, create a new OS, or Web Browser or anything. It's the end of monopolistic control of software.
But, I think the giants already know and accept this. The moat now is compute. A centralization of power back to the server, the rise of thin clients, and fat services.
So, it is a revolution but there's also counterbalancing forces. Still, we should ride that wave :)
> You could have hundreds of contributors, all using agents, working on different modules, according to existing spec and tests
The current problem with "Linux on phones" is the locked down nature of the hardware. For example, looking at PostmarketOS's support device list [0], sensors, Wifi, even phone calls don't work. Would what you're saying enable faster implementation of those support modules? (This would be really cool if possible).
[0] https://en.wikipedia.org/wiki/PostmarketOS#Supported_device_...
If it's just about building software against suites of tests and spec that already exist, then definitely what I'm saying would make it faster. But if it's a hardware control issue, then no.
In that case (ie, if in order to be free we need to free the hardware, too), we need to create a hardware company that builds a phone from the modem/radio on up and owns every layer.
Obviously non trivial hahahahaha :)
AI is letting the world of bits move faster than before by exponentially reducing rework and sharing around the benefit of network effects from collective human knowledge. It's not touching hardware in the same way, and doesn't give us the same superpower.
edit: I guess the "easier" play is to convince an existing full stack phone hardware company to make us an OpenPhone that we can hack on because they believe in the inevitabilities of trends and consequences from AI and want to invest in that future. That would be cool? Any takers? Reach out cris@dosaygo.com