It’s a user-friendly wrapper for existing fake quantum. It’s not a “physics shortcut” and it doesn’t “tackle quantum problems.”

Also no quantum problems have ever been “reserved for AI.” Some quantum solutions borrow optimization techniques from machine learning, but classical machine learning algorithms aren’t designed to leverage (or even consider) quantum effects.

I’m putting this out there because there’s a tendency to lump together all the buzzwords, like AI and quantum, into one big category of powerful-technologies-I-don’t-understand that results in hyperbolic projections and magical thinking that thwarts progress.

Edit: I wasn’t actually disagreeing with the comment above. You should downvote me to too.

Board of directors

Correct. The board defines the company, not the CEO.

CEOs are usually puppets. Whatever role they play, you can bet they were hired specifically to play it, and were incentivized to stick to the script.

Their job (legally, their fiduciary obligation) is to maximize shareholder value, to take the credit or blame, and fuck off.

The board (typically key stakeholders) are so pleased when the public focuses on their CEOs, even if it’s for their shitty opinions, behavior, or obnoxious salaries.

Because the worst thing that could happen to them would be for the public eye to actually follow the money, and it’s easy to see why.

If the rabble truly fathomed just how many of those “golden parachutes” stakeholders stockpile with every disgraced CEO, however ceremoniously disavowed…

Accountability would shift to more permanent targets yes but, more importantly, it would quickly become common knowledge that, all this time, there were in fact more than enough golden parachutes to go around.

New York or Disney World

Got me

For example the tools for the really tedious stuff, like large codebase refactoring for style keeping, naming convention adherence, all kinds of code smells, whatever. Lots of those tools have gotten ML upgrades and are a lot smarter and more powerful than what I remember from a decade ago (intellisense, jetbrains helper functions, various opinionated linter toolchains, and so forth).

While I’ve only experimented a little with some the more explicitly generative LLM-based coding assistant plugins, I’ve been impressed (and a little spooked) at how good they often were at guessing what I’m doing way before I finished doing it.

I haven’t used the prompt-based LLMs at all, because I’m just not used to it, but I’ve watched nearby devs use them for stuff like manipulating a bunch of files in a repeated pattern, breaking up a spaghetti method into reusable functions, or giving a descriptive overview of some gnarly undocumented legacy code. They seem pretty damn useful.

I’ll integrate the prompt-based tools once I can host them locally.

I’ll admit, some tools and automation are hugely improved with new ML smarts, but nothing feels dumber than hunting for problems to fit the boss’s pet solution.

It seems like the US patent system today is rarely anything but a solution to its own problem. In most cases a patent is little more than an expensive troll ward or a way to demonstrate due diligence to investors. What’s taken its place is time to market. If that’s true, the patent system should either be replaced with something that serves its intended purpose or that office should stop accepting applications.

FWIW this is a common post-regime debate. Visit Berlin to see a number of creative solutions.

Yeah I have a few of those for the most secure stuff. Hard to beat! The USB-C one is the newest and I debated the choice but damn these days it’s great how it works with everything.

If we cut and run every time a big corporation “embraces” a new standard, just to lessen the pain of the day it’s inevitably “extinguished,“ we’d miss out on quite a lot.

This standard was open from the start. It was ours. Big corps sprinted ahead with commercial development, as they do, but just because they’re first to implement doesn’t mean we throw in the towel.

Also:

1. Bio auth isn’t necessary. It’s just how Google/Apple do things on their phones. It’s not part of the FIDO2 standard.
2. It works with arbitrary password managers including FLOSS and lots of hardware options.
3. Passkeys can sync to arbitrary devices, browsers, device bound sessions, whatever.

Yeah the moods in this thread, like

“[I don’t understand this]!”

“[I don’t trust this]!”

“[It doesn’t fix everything]!”

“[This doesn’t benefit me]!”

“[What’s wrong with old way]!?”

And like, all valid feelings… just the reactions are a bit… intense? Especially considering it’s a beta stage auth option that amounts to a fancy version of the old sec key industry standard, not the mark of the beast.

Yeah the counter-interoperability of proprietary expansions on FIDO standards sounds a lot like embrace extend extinguish to me. I know engineering standards generally require field revisions but these big corps have a track record of this behavior.

I can see how the FIDO standard’s dID requirement might be an issue at the org level, but even in the case of a fully custom/unknown rooted device they have provisions for using traditional security keys attached to one or more associated devices via USB/BT/NFC. Megacorp platforms might be first to facilitate adoption but the spec absolutely accommodates open provider integration.

I need to experiment with personal security passkey registration and authentication workflows to know how difficult it actually is in practice, but it looks like the equivalent of self-signed certificates are possible anywhere the user controls the stack like self-hosted intranetwork suites that are popular around here.

Thanks again for the write up!

I could see that. I’ve only found a few in the wild (mostly just enterprise, niche tech-related, and big platform web apps) but there’s probably some clunky implementations out there I haven’t suffered through yet.

For one, there seems to be this idea that if you lose your passkey you get locked out of your account forever.

True, plenty in this thread even. IIRC there’s usually a recovery key process same as a typical authenticator MFA, sometimes other routes in addition like combining multiple other MFAs or recovery contact assignment. Regardless, completely losing PW manager access across devices would presumably be the more immediate crisis for most.

Thanks for the great article! I had a question re: the top disadvantage you mention (lock-in).

Background: Although the on-device integration for Apple, Google, etc. use their cloud for E2E sync between devices, it appears KeePassXC using their passkey interception, discovery, and import procedures accomplish the same cross-device passkey implementation without needing a particular vendor cloud lock-in. As best I can tell, this meets the original standard’s sync fabric requirements (whether or not the big providers like it) and relies on platform-specific APIs mostly for interoperability.

Question: If KeePass has been able to implement their own sync this way, and the FIDO standard accommodates non-OS providers (e.g. browsers or PW managers), what is currently the biggest technical hurdle remaining for FOSS-based passkey providers?

KeePassXC has begun rollout of their own implementation, and I’m pretty sure they’re considered FOSS.

From a quick scan of the white paper, it appears they’re currently using on-device passkey discovery and otherwise “intercepting” passkey registration workflows, which I take to mean they aren’t originating the request as a passkey registrar. This may be the easiest method to satisfy FIDO’s dID requirements.

This is a big one. Lock-in and the threat of provider blacklisting means it will remain a shortcut like SSO (“sign in with ____”) until we’ve established federated providers.

On further reading, this may not be as far off as I thought. Passkey registration providers can be OS-level but browser and password manager based solutions were intended (overview from FIDO alliance). And it looks like KeePassXC has begun rollout of their own. If I’m reading correctly they currently “piggyback” off of an OS-based provider in various ways, so it’s not yet an end-to-end implementation, but these are early days.

The passkey options I’ve come across so far are as close to push-button as I can imagine.

Do you mean from the developer perspective, like the complexity of the API/workflow?

Exactly

Slow is smooth, and smooth is fast.

Haha that’s the one ;)

Yeah you get it. It’s a “slow = fast” type of spiel, just a bone to pick with colleagues who embrace anti-user practices needlessly.

You still need 2fa

I think most passkey implementations incorporate multiple factors already. The session factor is considered distinct from the device factor, even if it’s all on the same device.

Which isn’t super different from the traditional USB key procedure, where a user would activate a FIDO biometric after clearing an SSO portal, or what have you.