No one is giving the real answer to this question which is that these laws touch on the legality of capabilities the US does not want to acknowledge it has, and so it won't. Secret information in the US is like glitter - if you touch it, it spreads to you too.
It's not just "here's this law about spying in the abstract," which can be talked about in the open; it's "we're using the MIND READER 5000 under Title XVII powers delegated by the President to the Mind Reading Organization to conduct mind reading on persons X, Y, and Z." The legal briefings and rulings on the facts will by necessity reveal the nature of the technology or technique in question, and so, they are classified. If I recall correctly it was a big deal for the ACLU to get a lawyer a security clearance so they could interact with these courts, but I could be misremembering.
The real problem here is that if someone files a claim against the government and the government wants to keep the facts of the case a secret, you still need the case to have an outcome, and then you have two options.
The first is that you assume whatever is being kept a secret is something that doesn't favor the government, and then the government loses the case by default. This is the more reasonable option, because it mitigates the perverse incentive for the government to declare something a secret whenever it makes them look bad. But it has the obvious problem that then the government would lose some cases it should have won if the information could be published.
The second is that you assume whatever is being kept a secret is something that favors the government, and then the government wins the case by default. But that doesn't solve the problem -- you then have members of the public losing cases they should have won if the information could be published. And doing this makes the perverse incentive worse.
So why do we accept the second one instead of the first one?
Oftentimes the government drops the case. When people get convicted for things they did on Tor, and don't settle, the government often drops the case so they don't have to reveal how they got information about activities on Tor.
When Germany, which doesn't play by the same rules, got DiDW and then Boystown, that was quite revealing on how authorities can break Tor.
That's when the government is the one bringing the case to begin with. The issue is, if the government e.g. kills your spouse, why should the perpetrators get off scott free because the perpetrators want to keep something a secret?
Imagine the public tried to do that. Sorry, IRS/DoJ, the defendant is refusing warrants and subpoenas because it's a secret and without that you can't prosecute them.
A lot of the frustration I've seen with PowerShell from longtime Linux users is that there's an expectation that everything is a file in the linux/bash idiom, and can be interacted with that way (tangent, I think this is where a lot of vexation with systemd comes from), where Windows and the NT kernel enforce the concept of everything is an object.
It's a much more powerful system, but users who are used to being able to do, well, whatever they want with no rules get frustrated with the syntax and "correct way" to do a thing when you just want a stupid string (it's me I'm users). It's very similar to Rust in that way, where once you stop fighting the borrow checker life becomes so much easier; once you stop fighting the objects, you stop getting in your own way and it becomes a much more productive experience.
This is appetizing in the abstract until you're in the middle of New Jersey and its horrendously cold and rainy and the pump jockey hates your guts because he had to leave his heated hut where the Bruins were on, and makes a half hearted wipe at your icy windshield to guilt you into tipping him for doing something you'd be happier doing yourself.
Seems like "it's horrendously cold and rainy" is the worst example to choose to illustrate negative aspects of gas station attendants...? I got to stay in my warm and dry car in this scenario?
I've lived in NJ my entire life and I've never once tipped somebody at the pump. I've had them wipe down my windshield a single-digit number of times. Where was this?
My token usage on Claude models has dropped by 83% over the last month - I'm pretty much only using it for quick one off questions or reading papers. it feels impossible for me to get Opus models to stop entering into cyclic loops, and my work is too security adjacent for Fable.
Codex has been an excellent workhorse - doesn't feel like I have to dance around the guardrails, doesn't lose _everything_ when it compacts, and doesn't litter the workspace with a million and one planning to plan files.
I have to agree with you there. I did some good work with Claude then Fable came out - impressed with that as well. Then they dropped access to it and upon returning was never the same - even the Opus models for some reason. Then one day I burned through my limit in about 10 minutes and had to get a project completed. I subscribed to Codex and it has been fantastic - finished my project and continued on to others. I just dropped my Claude max plan down to the pro and subscribed to the $200 plan on Codex.
my problem with claude currently is the language its using is dense and feels like its not even meant for humans. this guy is calling everything a spine, a seam, a gate, load bearing, any ui element is "chrome", it's actually absurd.
yeah there is that too. if everything else I stated wasn’t wrong I could put up with that. Codex on the other hand is great. Concise, technical, effective and non chatty. And that my friend is a load bearing comment.
I don’t get Claude, and that’s almost exactly what I did - I dropped to Claude Pro $20 + Codex Pro $100, and then unsubscribed from Claude and ramped up Codex. The Claude Pro is consumed within an hour on a simple task. I wish only Codex worked a bit faster than on the Fast mode.
I used to rely on Fable for research when it was first out, today it doesn’t seem to be much better than Opus, and it uses up the quota exceptionally fast - 1h Fable in a single short session, and there’s little left for Opus to hit the 5h limit in a second session. With Opus I get about 3-5h of relaxed use with a couple subagents to save the context, but there’s usually quite some disagreement between the subagents and orchestrator - Claude does some model routing with default agents and picks Haiku and Sonnet for subtasks - only later to disagree with them and redo the work - and burn extra tokens. With Claude, it’s really either Opus or Fable if you want some quality.
That said, their marketing is exceptionally effective. Virtually all nontech folks consider only Claude.
It is wild to read stuff like this when it is OpenAI that is being blamed for rug-pulling users and secretly reducing limits. It's such a big mess that Codex's product lead has been frantically posting updates on Twitter about it.
For those interested, Neal Stephenson writes about this in _Anathem_, where there is an arcane technical class (the I.T. Administration) that has developed programs over the years to sift through the misinformation, sentient viruses, and general overrun of the Internet to determine what is probably true/accurate.
He writes about it in Fall, or; Dodge in hell! The Miasma; the workaround is personalized filters (or generic ones for those who can't afford the personal ones). This leads to societal consequences.
It's not just "here's this law about spying in the abstract," which can be talked about in the open; it's "we're using the MIND READER 5000 under Title XVII powers delegated by the President to the Mind Reading Organization to conduct mind reading on persons X, Y, and Z." The legal briefings and rulings on the facts will by necessity reveal the nature of the technology or technique in question, and so, they are classified. If I recall correctly it was a big deal for the ACLU to get a lawyer a security clearance so they could interact with these courts, but I could be misremembering.
reply