> Again what it means in practice should be discussed but basically any reasonable person including you and anyone with your level of expertise should be able to say honestly that, yes indeed this computer is under full control of the owner and the owner only. It has no backdoor or ability to be remotely updated without the owner intention.
That, unfortunately, is impossible for silicon platforms. Documentation can't prove the lack of a backdoor, because you can't prove that the chip you got is what was documented (even if you have a scanning electron microscope and a lot of time, that's a destructive process and you can't prove that the chip you'll actually use is the same as the one you analyzed).
This is a common fallacy espoused by those who demand ultimate freedom: that along with it comes full trust and control. It doesn't. Because physics. Chips aren't software.
Instead, you should be investing in one of these:
https://www.crowdsupply.com/sutajio-kosagi/precursor
Precursor makes the quite solid argument that general purpose FPGA backdoors are infeasible (because it's an intractable problem for arbitrary logic circuits) and therefore it is a device that can be trusted even if the silicon can't. Of course, then you'd better be happy running all your computing on a 100MHz RISC-V.
> Do they? Wouldn't you say that Apple have a full control over their Mac M1?
Given that it's sitting on my desk, it's running my own OS, there is no Apple code running on it with full system access by the time it boots into Linux, and Apple can't remotely update it (the bootloader doesn't even have USB support let alone networking, and there is no resident supervisor like there is on Intel machines, Android phones, etc), no, I'm pretty sure I have full control over it for all normal practical purposes.
In fact I'm much more sure about that than I would be with the laptops the FSF peddles as "respects your freedom"; last time I looked at the schematics for one of those, it had over a half dozen chips running secret blobs, and at least two or three of them had full access to all system RAM via a DMA capable bus. You'd have to be insane to trust that over an M1, which is designed to sandbox all coprocessors from the main CPU and RAM via IOMMUs, such that even if all firmware is backdoored it can't take over your main CPU.
Is it perfect? No, if I wipe the Flash without a backup the recovery process requires phoning home to Apple, since that's how it re-downloads things like certificates, calibration data, MAC addresses, etc (though at least we have open source tools that implement all that and you can run on Linux). But that's a repairability/longevity argument; while the thing has a proper bootloader installed, it isn't phoning home anywhere during normal operation.
Then again, for those FSF laptops, if you wipe Flash you need a soldering iron to recover them, so from that point of view the M1s are a lot more robust, since you can recover them via USB from any other random machine, no disassembly required.
This is why nuance matters. Absolutist positions like the FSF's and yours lead to less trust, because reality isn't absolute, it's nuanced. If you want absolute trust, you can pre-order a Precursor today. If you want a laptop class machine you can reasonably trust not to be backdoored, you'd do much better getting an M1 than the obsolete ThinkPads the FSF certified, which don't even have modern security features like IOMMUs and have known blobs with full control over the computer. I can't prove the M1 doesn't have any secret silicon backdoors but at least the design is clearly intended to prevent firmware ones, and there are no known bypasses, which can't be said of those ThinkPads.
> If you ask today in practical terms what is a reasonable line to draw, let's draw it where Apple have done it except the key should be in the hands of the owner, not Apple. How about that? Would it satisfy?
That is a very difficult question to answer. Would hacker me like a device where I can burn in my own iBoot signing keys? Sure. Devices where you can do that kind of thing exist; you can get SoC dev kits (e.g. I know the Nvidia Tegra X1 devkit can do it) where you can burn your own keys permanently and become the trust authority, and I wouldn't be surprised if it can be done on open designs like Novena, though it may not be documented. Is it a good idea for the general public? No. The problem is building a platform that is "fully user controlled" (whatever that means, usually "everything that runs after the Boot ROM") means delegating an immense amount of responsibility to the user, which is at odds with security paradigms that attempt to protect you from higher level attackers. Can you build a signing and key storage facility on par with the security of Apple's? Highly unlikely. So yes, you might have full control over the device, but you're getting less security against other attackers as a result. The reason Apple can build a user friendly yet secure device is because they control the early boot stages, so they can build the complex mechanisms on top that make that possible; full control, secure, user friendly: pick two.
Can Apple sell me a backdoored M1? Yes. But the entire point of their design is that nobody else can backdoor it for me. Not even you, if you get ahold of it. Their user-controlled secureboot delegation requires authenticating using your machine owner credentials to install your own kernel, after having asserted physical presence, unless you wipe the whole machine and start fresh. And even then you can't backdoor the recovery mode used to do this, so you couldn't backdoor someone using the Asahi Linux installer since it runs from that secure recovery. Their design is such that compromised third party chips can't compromise the main system. You can't backdoor the motherboard and replace the flash. Etc. You have to trust Apple to an extent, but in exchange you're much safer from threats from other parties than with other machines. You can buy a second hand M1 Mac and be very confident it's every bit as secure as one straight from the Apple store. That is something that cannot be said for the vast majority of consumer hardware.
Those are all worthy things to have. Is it worth the trade-off of letting Apple control the early boot stages? I think it is for many people. Is it possible to build a system with identical security guarantees that is simultaneously fully user controlled and doesn't require every user to have a secure lab, HSM cluster, etc to achieve the same level of threat resistance? I don't know. Maybe. I invite you to try to come up with such a design; it would be very interesting to hear about it :)
The good news is that security is composable. That means that, for example, if I use LUKS encryption from Linux on top of the native hardware keystore and encryption, I know I have top notch security against third party attackers and I know I'm safe from Apple helping some government entity to crack it if they seize it, since they won't be able to break the second encryption layer (and I'm very confident they don't have any remote access backdoors for normal usage, so I'm safe from post facto active targeting since they just have no way of doing that when the thing is running my own OS).
And the fact that I just wrote a wall of text about this is, again, evidence that nuance matters. You can't reduce any of this to "do I control the signing keys" or "does it have any blobs". I'm sure some people will read my story and decide they don't want to touch the M1 machines with a 10 foot pole, and would rather get a Novena or a Pinebook or whatever. And others will be very excited and want to get one ASAP, because it's massively more trustable than any recent Intel or AMD machine by design. And the only way you can decide is if you know the facts and how the machine works and what the threats are and who you need to trust and so on and so forth. A "Respects your Freedom" sticker tells you nothing.