degamad

How do they ensure that?

Reading the description, looks like a stricter chroot for purely-network applications, maybe?

Is there a good example of a private equity buyout which wasn't based on that strategy?

As an element of backwards rationalisation, my brain interprets "ENOTTY" as "E-NO-TTY" (Error: NO TeleTYpe available) which is the rough current meaning, rather than "E-NOT-TY" (Error: NOT a TYpewriter) which is the original meaning.

> everyone in the LLM space

The problem is that the language spills outside of the LLM space, so we should use terminology which is more generally understood in a way which reflects reality.

From long-lost memory, I think it's implicitly closed tags (e.g. img).

Ditto - another long-time K9mail user on Android.

Tried a couple of other clients from time to time (mostly when OAUTH logins for GMail and Outlook365 weren't implemented yet), but haven't been convinced to switch.

> You could have one "master recovery key", the public part loaded on your device (to be registered during account creation) and the private part sitting in your safe.

This is true, it's possible to have just the verification information present without the key itself.

However, a part of almost every credential enrollment process is verification that the credential is usable and controlled by the enrolling party.

For passwords it's simply "please type the password again", for TOTP it's generating the next number, for email/SMS it's entering a code sent to that address, so for adding a public key, it should include some way of verifying that you control the private key associated with that public key. It's hard to do that without having the private key available.

> ... just "put this recovery key/password in a safe and forget about it until needed"...

Surely that is also impacted by:

> Important accounts are not properly backed up until you drive to your safe. That can be many days for a bank deposit box, or even longer if you're traveling.

Exactly. There's many a page on random which explains why taking the modulus of rand is the wrong choice, but it's also the easiest one, which makes it all to common.

That's the ambiguity in the way we use some words like "server". If I say I'm going to "build a web server", you don't know if I'm developing software to compete with nginx, or building an x86 box and installing nginx on it.

In addition, while the net elevation change is bounded by the delta, given the choices might be "go up this steep section" vs "go along this gentle incline", one route can be "flatter" between the same endpoints.

Billing caps that Google notoriously doesn't enforce?

> the situation would have to be very particular

The situation is very particular. When the system makes an error, I make a bigger profit. If the system made fewer errors, it would approve more things that cost me money.

Why do you think I'm motivated to make fewer errors under that situation?