John Carmack's Arguments Against Building a Custom XR OS at Meta

Unless its for love or devotion, there's no compelling reason to create a new OS in 2025. Certainly that could change in the future but I think his observation (if I understand him correctly) is correct.

Just google/youtube the history of temple OS and its creator. It's fun, sad and tragic at the same time.

Spoiler alert: a single person coded it in his own programming language, but the person suffered from severe mental illnesses and ended up taking their own life.

Just walk up to the gate of your nearest Concent next Apert and they will take you right in!

Actually, I don't know how you join the Ita now that you mention it.

Yeah, the Linux kernel has ~12m lines of code. <1m are the core, the rest are drivers.

Android built a new, giant moat for Linux (or "Linux" depending on your opinions about Android) in the embedded application processor space - now the "standard" board support package target for new embedded AP hardware is almost always some random point-in-time snapshot of Android. Running "mainline" Linux is hard (because the GPU and media peripheral drivers are usually half-userspace Android flavored stuff and rely on ION and other Androidisms) and bare-metal is even worse (where previously, you'd get register-level documentation, now, you get some Android libXYZ.so library).

With my experience of being written up for constructive criticism the reasoning was that I didn’t give constructive criticism to others and they felt singled out. I only give such criticism in private so of course they were not there to see the others. Apparently that wasn’t a sufficient explanation.

This.

Being "reported to HR" doesn't mean "almost got fired". It likely meant a meeting where someone explained "hey, the way you communicated that caused some upset, let's discuss better ways to handle that situation next time." Very often in larger companies, complaints about things like "this bigwig from this other group jumped all over us" are automatically sent through HR because HR has staff whose job just is resolving conflicts between people and keeping things peaceful.

Having worked in the valley, I've seen what critical feedback meant in many companies there, and it removes all usefulness of the info because there is a ceiling of what is socially acceptable to say; therefore, you can't know how bad or urgent things are.

Everything is ASAP. They are super excited about everything. And nothing you do is wrong, it just could be improved or they like it but don't love it.

You don't know if something is important, basically.

Just like Louis CK said, "if you used 'amazing' on chicken nuggets, what are you going to say when your first child is born?". But in reverse.

Personally, I'd rather work with someone who would tell me my work is terrible if it is.

In Germany, you can't even legally say somebody did a bad job at your company in a recommendation letter. Companies created a whole subtext to workaround that, it's crazy.

Some things are just bad. You should be able to say it is. Not by saying it could be better. Not by using euphemism. It's just something that needs to go to the trash.

In fact, I don't trust people who can't receive this information, even if not packaged with tact (which you should attempt to, but life happens). If you can't handle people not being perfectly polite every time, I can't help but feel I won't be able to count on you when things get hard.

That must be the French in me talking.

This is not that though. This is just developers being unable to handle constructive criticism, and when they can't win the argument on merits, went for the HR option. It happens.

I've had it happen to me too, but my response was to resign on the spot (I was already not satisfied with the company).

The "toxic behaviour" I had done? I reverted a commit on the master branch that didn't compile, and sent a slack to the Dev who had committed it saying "hi! There appears to have been a mistake in your latest commit, could you please check it out and fix it? I've reverted it in the meantime since I need to deploy this other feature"

The dev responded by force pushing the code that did not compile to master and contacted HR.

I decided there was greener grass on other pastures. I was right.

From what you know of Carmack, does "can't give critical feedback in a professional setting without being rude or belittling others" sound like him to you? It does not to me, though granted maybe he's different in his non public persona than what you can see in presentations and talks.

> unsolicited advice

He was the CTO of Oculus. Surely it is appropriate for the CTO to give advice on any big technical decisions, if not outright have veto power.

> If you're in the middle of trying to write a new operating system, then it's probably not helpful to have John Carmack standing over you repeatedly telling you that you shouldn't be doing it. In this case Carmack gets the last laugh. Then again, it is easy to get the last laugh by predicting that a project will fail, given that most projects do.

I mean, if you're working on a project that is likely to fail, wouldn't it be nice if someone gave you cover to stop working on it, and then you could figure out something else to do that might not fail? Can't get any impact if your OS will never ship.

When a veteran tells you something and is passionate about it, maybe it is worth listening or at least dealing with internally. At the end, he left anyway even if the project didn't fail and Meta remains wealthy but largely mediocre in terms of the products it delivers while relying heavily on startup acquisition and large spending. Pretty sure most people who work there only do so for premium rent-seeking.

None of it surprising if this is a signal of how they operate.

The only reason you want me to "cool off", is because you feel bad just interacting with somebody expressing a polite, strong opinion. Online. On the other side of the world. With text.

This is exactly the madness I'm talking about.

Case in point.

It got mentioned for a reason. And obviously escalating with HR is a big deal as it comes with career risks for the person you are reporting. Risking someone else's career should be a last resort but seems to be more commonly a knee-jerk reaction with HR becoming weaponised.

The drawback of this is you lose good talent and keep rent-seekers instead.

And, let's be real here: engineering prestige.

Everyone wants to make an OS because that's super cool and technical and hard. I mean, that's just resume gold.

Using Linux is boring and easy. Yawwwwn. But nobody makes an OS from scratch, only crazy greybeard developers do that!

The problem is, you're not crazy greybeard developers working out of your basement for the advancement of humanity. No. Youre paid employees of a mega corporation. You have no principles, no vision. You're not Linus Trovalds.

Wouldn't that be an application (or at most system library) concern though? The OS is just there to sling pixels, it wouldn't have any idea whether those pixels are blurry… well for VR it would all be OpenGL or equivalent so the OS just did hardware access permissions.

Just overclock (more) the system that’s already in a severe struggle to meet power, thermal and fidelity budgets?

You can do that on Linux today with vfork.

Hold on a sec, is that the same on PS5? I am pretty sure that wasn't the case two generations ago. Is that the norm now, running on hypervisor ?

See? This is why the Doom man got paid the big bucks. The Doom man has a literacy distortion field.

According to Wikipedia, looks like only the Nest Nub.

Google Nest Hub

That was also frustrating in that sel4 was right there. Why not invest efforts in the existing thing?

Yup it’s because they had to rewrite the drivers I presume. Always going to be the biggest issue with any new OS.

While I agree with the sentiment given my bias towards safe systems languages, Genode OS is pretty much mostly C++, although they added some Ada/SPARK as well, which is relatively recent research OS.

Sounds like super fun if I could be paid a bit for it.

What is an easy gate task to get into “reverse engineering some drivers for some OS”?

Second thought: I don’t even know how to write a driver or a kernel, so I better start from there.

But is that a good use of Meta's money? Compared to making a few patches to Linux to fix any performance problems they find.

(And I feel bad saying this since Meta obviously did waste eleventy billion on their ridiculous Second Life recreation project ...)

Writing drivers is easy, getting vendors to write *correct* drivers is difficult. At work right now we are working with a Chinese OEM with a custom Wifi board with a chipset with firmware and drivers supplied by the vendor. It's actually not a new wifi chipset, they've used it in other products for years without issues. In conditions that are difficult to reproduce sometimes the chipset gets "stuck" and basically stops responding or doing any wifi things. This appears to be a firmware problem because unloading and reloading the kernel module doesn't fix the issue. We've supplied loads of pcap dumps to the vendor, but they're kind of useless to the vendor because (a) pcap can only capture what the kernel sees, not what the wifi chipset sees, (b) it's infeasible for the wifi chipset to log all its internal state and whatnot, and (c) even if this was all possible trying to debug the driver just from looking at gigabytes of low level protocol dumps would be impossible.

Realistically for the OEM to debug the issue they're going to need a way to reliably repro which we don't have for them, so we're kind of stuck.

This type of problem generalizes to the development of drivers and firmware for many complex pieces of modern hardware.

    > custom Wifi board

Things you can’t buy: vendor who cares enough to replicate your exact use cases in their lab

XROS had a completely new and rapidly evolving system call surface. No vendor would've been able to even start working on a driver for their device, let alone hand off a stable, complete result. It wasn't a case of "just rename a few symbols in a FreeBSD implementation and run a bunch of tests".

Vendors might say that they don't have the resources (man hours) and don't want to hand over documentation to external developers.

The VMM on GCP has only really been tested with Linux. You are kinda wasting your time, the only way to make it work is to make the hobby OS Linux.

To clarify, are you having trouble getting the signal to reboot from the gcp console into your OS? Or are you having trouble rebooting on gcp?

I also think that they have access to more helpful resources than people outside the field do, e.g. being able to contact people working on the lower layers to get the missing info. These channels exist in the professional world, but they are hard to access.

> "Oops, sorry, didn't get around to documenting it".

That's obviously the wrong message. They should say "Go ask the engineering VP to get us off any other projects for another cycle while we're writing 'satisfying' documentation".

Extensive documentation comes at a price few companies are willing to pay (and that's not just a matter of resources. Look at Apple's documentation)

> If the system is complex, there's more need to document

It’s not first party documentation that’s the problem. The problem is that they don’t share that documentation, so in order to get documentation for an “unsupported” OS a 3rd party needs to reverse engineer it.