The exceptions I know of are things like:

- In some situations you have to ID yourself (not just when you're driving, but that's the usual case, where they have legal authority to ID you, and if you don't have ID you have to provide the equivalent from memory so they can check your identity.)

- Face and Fingerprint unlocks are not things in your mind, they're not testimony, therefore you can be compelled to provide them, though usually by court order. If a cop in the field is forcing you to face- or fingerprint-unlock your phone, that might not be allowed.

- In a few jurisdictions, if it's a "foregone conclusion" what's on your phone, i.e. they know from other threads of the investigation what's on it, but it's just encrypted, they might be able to force you to unlock it in that case, but that's through a court order not during a field investigation.

Props to the defendant for volunteering as a test case. He should've powered off his phone before he went through customs, or better yet traveled with a burner phone.

It doesn't matter if the feds are doing something wrong... if you interfere or destroy potential evidence, rather than simply recite your 5th and 6th amendment rights, you've put yourself in a whole new ball game.

He probably would've been fine if he'd maintained his refusal contingent on talking to a lawyer first. They might not have even attempted to crack the phone, and if they had, he'd have a better chance of getting the contents thrown out than he has now of fighting the destruction of evidence charge. The two issues are largely separate, unless a court decides the detention and attempted search were so egregious that they negate his separate, intentional act of wiping his phone when he was being detained.

We might wish for a mobile device to be a mobile castle immune from any [request for] search or seizure without a warrant, but that hasn't been established in case law, and everything else crossing the border (except what's diplomatically immune) is subject to some warrantless scrutiny.

maybe a little?

Strangely, nobody who runs linux has this problem.

Perhaps because windows "drivers" are so bloated with helper apps and other stuff that they can't possibly all be shipped to end users with the core OS.

the real one is issuer=C=US, O=Google Trust Services, CN=WE1

I hope you meant 1.0.2 or 3.2. There was never an openssl v2 release, so if that's what it's reporting, you have problems.

The program (unitary gigantic bash script) testssl.sh (github) can give you a thorough rundown on the ssl properties of whatever's MITMing you. The other possibility is that cloudflare is unusually serving you an alternate certificate and your ancient openssl and browser are both unable to handle whatever the presented cert algos/extensions are.

Reload loops, or being able to "bypass" anubis (unless you merely mean bypassing it for the token validity period by solving a challenge), sound like misconfigurations. There's no reason for anubis itself to cause reload loops; it's tricky to configure a webserver to use it in some scenarios.

Any ability to bypass anubis probably means the site is using it in auth/challenge mode only, and then misconfigured their webserver's auth checking. Or it's a bug. If you mean the double-spend tavis mentioned in his blog post which previously made the HN frontpage, that was patched right after it was reported to the maintainer almost a year ago.

We can all argue based on how we envision "ideal" scraper networks being run and whether the web-PoW concept would stand up to that. However, what matters at present is that anubis helps many sites cope with misbehaving bot scrapers written by the script kiddies you mention, who don't care if the internet burns as long as they finish their scrape 1 hour faster. If anubis motivates them to devote a few brain cells to make their scrapers smarter, they may also fix the scrapers to not take down the sites they're scraping.

It's massively less annoying than a captcha, which is both a longer delay (typically, at present) and a massive cognitive distraction/roadblock.

The anubis author has stated they recognize it's an arms race, but PoW scales. Captchas and other signals are already at the end of the road; any additional difficulty increases false bot-positives, which are already unacceptably high.

For websites running dynamic languages, a binary (anubis is in go) sentry that operates before[1] the website is forced to expend any resources, is usually a large improvement over a site-hosted captcha. I would rather, and I think most humans would agree, have to wait a few seconds, maybe even closer to a minute in the future, to get a website access token good for a day or a week, than be forced to solve a captcha.

The dilemma for bots: when tokens are bound to the connecting ip, scrapers must limit the connecting IP pool for each site they want to scrape, becoming much more obvious and easy to block, or they have to use massive amounts of compute.

[1] this is true regardless of whether anubis is in reverse proxy mode or auth mode.

Doing less to save people in other countries that have no legal demand on our treasury is not "being responsible for [their] deaths." It's tragic, and it may even be a bad policy decision, but there's no responsibility (in the "duty to prevent harm" sense) or evil there.

If you take away copyright, you reduce the revenue of publishers, which reduces the number of unproven authors they can take chances on. (They're not very good at picking winners from the pool of new author manuscripts, not nearly as good as "agents" like to think they are, but that doesn't matter; they still wouldn't be able to take as many chances.)

In large part that's because, if they'd done so, the kids would've been socially isolated from their peers, at least the most normal ones with the most normal parents, which are the kinds of friends most other normal parents want their kids to have.

It's a collective action problem, except instead of "I can be better off if I ignore what I know is best for society", it's "if I ignore what I know is best for my kids psychologically, they will still have friends, and social media brainrot is a lesser evil than socially isolating my kid from all the normal kids at school."

And also, giving kids social media interaction devices is a convenient form of babysitting. It reduces up-front effort of parenting.

If this fails it'll be because the tech industry expresses disapproval too loudly to ignore.

The legislators don't care about the underlying criticism. Almost no legislators have ever used a 3d printer or written any software, beyond maybe simple assigned programs if they had a required intro-to-programming course. Few are "tech" people. The rest don't understand this technology, or any technology really, beyond it being a black box for specific purposes. They see 3d printing and plastic guns and think something must be done, because the 3d printing black boxes are producing dangerous weapons.

> This repo was written almost entirely using AI with human oversight. If you see anything that needs fixed or would like to contribute, please email ... or reach out on Discord

Why not just provide an email address that's delivered directly to the agents you have developing Oak?

I didn't delve into the benchmark repo to understand what your loop is measuring. Why would an agent (without fine tuning or oak-specific context) be faster with oak than it is with git or jj?