> Glad I won't have to give my Chromebook a screwdriver enema again!
If you’re willing to do that, you can disable this key sequence.
Edit: let me elaborate on what I think is the reason. They are selling the hardware at a loss or close to it. When you put Linux on there you eat their revenue. Instead of being honest about this, they dress it up in a bunch of phony security talk. They don't want people reselling them as useful Linux workstations.
The point of the Chromebook’s security approach (with the developer mode beep + notification, etc.) is to have a “Trusted Computing Base”, in the original, non-dystopian meaning of the phrase: you’re supposed to be able to tell, as a regular user, just by looking at/booting up a Chromebook, that it’s running Google-signed firmware/software.
A Trusted Computing Base is precisely for the case where you do trust some particular third-party to be your IT admin (because, for example, they’re your school or employer, or—more likely—a subcontractor of one of those that MDM is being delegated to.) In such cases, you want to know that your device is in exactly the state that that party wants it to be in, and that no other party has been able to modify it.
Google aren’t trying to make it impossible for you to use your Chromebook for whatever you want; that’s just a side-effect. Their goal is to make it impossible for someone who intercepts your Chromebook in shipment to install a rootkit on it, in a way where the device still appears to be (and act like) a Chromebook, with no sign that there’s anything wrong.
Yes, Google is the IT admin of the device. Someone’s gotta be; in a Trusted Computing Base scenario, you remove the possibility of handing the thing off to a repair shop (because what is a repair shop—from the TCB perspective—but another attacker?)
Essentially, a Chromebook has the same goal as a high-security lock: nobody should be able to pick it. Not even a locksmith. Not even the locksmith you hired because you lost your keys to the lock you own. Just like with encryption, you can’t leave a backdoor that only works for people with good intentions. You have to leave no backdoor at all—only a “front door.” In the case of a high-security lock, the “front door” is ordering new keys from the manufacturer, supplying your proof of purchase of the lock. In the case of a Chromebook, the “front door” is the signed update system, which allows one party (in this case the manufacturer, but in TCB Windows/macOS PCs this is your enterprise MDM admin) to manage everything. And also spy, if they want. But the two things are one and the same, really. The manufacturer of your high-security lock could break into your building, too. That’s why you pick a lock manufacturer you trust.
Sadly, as the author proves, it’s not impossible to install a rootkit (in this case, a Linux distribution, but it could just as well be a keylogger-injected version of ChromeOS) onto a Chromebook, in a way that’d be invisible to the end-user (unless the end-user X-rays their devices upon receipt and uses ML to compare them to a standard blueprint, ala the brouhaha around spy chips surface-mounted onto the bus of Supermicro servers.)
> you’re supposed to be able to tell, as a regular user, just by looking at/booting up a Chromebook, that it’s running Google-signed firmware/software.
Google did do that. But they went a step further - making it extremely easy to accidentally wipe your device and reset it to factory-defaults: https://www.chromium.org/chromium-os/chromiumos-design-docs/...
If you make it easy to persistently use a rooted device, then regular users would use rooted devices for regular tasks. (Because there are advantages of doing so, like side-loading native apps.)
If regular users use rooted devices, some people’s only experience of a device would be of a rooted device.
If some people’s only experience of a device is of a rooted device, then they’ll have no idea what an actually tampered device looks like, because every device they’ve seen is “tampered.”
This is exactly the state of Android, as it happens. Tons of people use Android in its developer mode; and so people don’t take a security mindset toward using a rooted phone.
The solution to this is to use technical mechanisms to achieve social effects (a.k.a. “incentive design” — the type of thing that tax credits are.) By making it irritating and risky to use developer mode in a persistent way, people won’t do it for anything other than tasks that truly require using developer mode for what it’s for (that’s “using the device as a debug build target”, by the way, not “writing code on the device.”)
Thus, nobody uses Chromebooks in developer mode.
Thus, the average person’s experience of seeing a Chromebook in use is, 99.9% of the time, of a non-rooted Chromebook.
Thus, the boot beep is surprising, and will scare the average user off of using the device.
Adding a switch in the OS that disables this, would be akin to having a switch in your OS (on a device MDM-managed by someone else, including the certificate store) that disables web browsers from doing any certificate checking. You could flip it; but so could viruses, and so could the NSA half-way through the laptop’s delivery process. Why would you want that switch to exist?
——
What people with complaints like yours really want, I think, is not to disable the TCB. You want to be able to be in control of the TCB. You want something that’s like a Chromebook, but where the MDM controller is an enterprise (e.g. you individually, if you’re clever enough) rather than Google.
Well, I mean, first: that’s what a regular PC with Secure Boot is.
But if the particulars of Chromebooks appeal to you (and I suppose they could appeal to somebody): the cheap hardware, the keyboard layout, the sandboxing, etc. Then what you might want is a variant Chromebook.
Such a device would have to look different than a regular Chromebook, somehow, to show regular users that it’s not for them. Sort of like how developer units of game console hardware are clearly not the same as production units. But if that were done, I think it’d be okay for those units to have a user-controllable TCB.
This is false. All but the most technically inept user will understand "WARNING! YOUR COMPUTER IS NOT RUNNING GOOGLE CHROME OS/HAS BEEN ALTERED FROM FACTORY DEFAULTS!" shown at boot time. Which is what Chromebooks do. All that's needed to make it into a usable device is to get rid of "Press space to wipe your device."
> Tons of people use Android in its developer mode; and so people don’t take a security mindset toward using a rooted phone.
Android security is catastrophic on factory settings already, so how you managed to blame the tiny proportion of users that root their phones to get rid of vendor bloatware for it is beyond me.
No. Not if half† the Chromebooks people own emit that message.
Users do not read. I know, it's hard to believe, but a large portion of average (not "most technically inept", average) users have never read a warning message on a screen in their lives. They see a warning message—and then they ask an authority figure what it means. They then classify whatever technobabble response they get into two executive-summary categories: either the warning means they should stop and wait for assistance; or the warning means nothing, and they should confirm whatever it's asking them to confirm and go on as per usual. After that, every time they see a prompt that looks like the one they asked about, they'll remember which of the two categories they surmised it to fall into, and that's what they'll do. Even if it's a prompt with different text. It looks the same—it gets the same treatment.
The next time you see a non-IT person using Windows and they get a UAC prompt, observe what they do. Ask them why it came up, and why they reacted the way they did. They won't know. All they'll know is that someone once told them it's fine to ignore those and click Accept.
If the ChromeOS boot screen is as common as Windows UAC, then it fails at its mission. That is why it makes it easy to un-root the device: because, if you have a boot screen that can "trick" average users into un-rooting a Chromebook, then you can't trust the average user around your rooted Chromebook to not accidentally wipe it, so you can't ever give the average user a rooted Chromebook as some kind of cheap kiosk (i.e. exactly what the author of the article was trying to do.)
And so rooted Chromebooks "in the wild" are rare—rare enough that if any user ever sees that boot screen, it'll be a new error that no authority figure ever told them "the trick" to bypassing and getting on with their day. New enough that they'll either "fall for" the prompt and un-root the device; or they'll call someone over to ask what the prompt "means."
----
† Sure, far less than half of Android devices are rooted. But consider the demographic difference. Every human being has a potential reason to own an Android device. Right now, very few average human beings have a reason to own their own Chromebook. (Be assigned a Chromebook by an enterprise MDM system, sure. But own their own? Quite rare.)
Ignoring the enterprise users, and taking the small slice of people who own their own Chromebook as the base population, any random niche use that requires rooting the Chromebook, if it got popular enough, could blow that population out of the water. If rooted Chromebooks turned out to be, say, good for running TAILS, or game emulators, or any other stupid thing you can think of, suddenly half the Chromebooks "in the wild" that a non-enterprise user ever sees would be doing that, and so would be rooted. They'd all beep and show the warning. And, like UAC, it'd be just another one of those things you ignore.
I personally find this useful, I've got two ChromeOS devices running other OSes, and I like the minimalist hardware. Frankly, I also get enjoyment out of using things in ways not totally intended.