Except the device doesn't conform to any standards, and apple hasn't bothered to document any of it. Given they tend to release new devices on a yearly cadence its quite possible that the "hackers" will never catch up. Sure in 4-5 years it might be possible to boot a linux distro on the M1, but by then apple will have the M5, with new GPUs, reworked interrupt, iommu, etc. So its great apple has this speedy device, but its likely that in the couple years it will take to open it up, it will be behind the curve. That is the problem with the rest of the "open" arm ecosystem, you looking at 5 year old SOCs designs. The "good" stuff might as well not exist at all.
And then its there is almost always a perf/power difference between a reverse engineered device and the native drivers.
This again has nothing to do with whether "you only get to do what Apple lets you do."
I'm not arguing the M1 is an open platform (as, say, PINE makes), only that it is not significantly more closed than the Intel-based Mac. Nothing has changed with regard to Apple's lack of documentation. Nothing has changed about the ability of the platform to execute open source code. The pace of hardware releases, and Linux's preparedness to deal with the changes to the hardware, are independent from the openness or non-openness of the platform.
The fact that my router was hacked to run DD-WRT doesn't make it an open hardware platform. The fact that a newer model can't be hacked the same way doesn't make my router "more closed" in retrospect.
I don't know what standards they should conform to but I doubt there is a vendor-agnostic GPU driver that Linux has that works with every non-M1 computer. It wasn't that long ago that Wi-Fi support on Linux all seemed to depend on reverse-engineered Atheros drivers. Hardware support has always been a problem on Linux, regardless of vendor. Day 1 availability of hardware support in Linux is not the indicator of openness.
There is more to reverse engineer on the M1 than there is on a random Intel based mac. Intel has been a reasonably good opensource steward as evidenced by all the .intel.com contributors to the linux kernel, and a goodly amount of documentation. Yes the intel has plenty of closed source firmware/etc but a lot of it exists to provide standardized boot flow/etc.
The M1 has none of that, so a random intel mac's reverse engineering is limited to only the parts Apple changed from what is mostly a set of platform standards that have been built up over decades. The core system IP (interrupts, iommu, virtualization, pci, memory controllers, USB, etc) already have linux drivers. The M1 OTOH isn't even fully compliant with the ARM instruction sets because they apparently have extended even that.
So, yes the hardware may not technically be locked down, but for all intents it might as well be, since Apple could have picked up the ARM system specifications and conformed to them but they didn't. The whole thing is vaguely reminiscent of SGI's failed attempt to create a new "PC" standard by dumping all the legacy, designing their own chipset/etc and running their own firmware. Yes it was an x86, but it didn't run anything except their blessed version of windows NT (IIRC). It was a dead end, because it turned out it didn't really offer any advantages, over a boring old PC, cost twice as much and removed the ability to run a bunch of software.
The M1 is much the same, it loses out in a lot of ways not only on the software front, but the hardware front as well. If it weren't for its fairly outrageous single threaded perf, which is at least partially enabled by being fully two process generations ahead of intel it wouldn't be noteworthy at all.
And then its there is almost always a perf/power difference between a reverse engineered device and the native drivers.
Consider Nouvea as an example here.