Memory allocators are something I expect to rapidly degrade in the absence of continuous updates as the world changes underneath you. Changing page sizes, new ucode latencies, new security features etc. all introduce either outright breakage or at least changing the optimum allocation strategy and making your old profiling obsolete. Not to mention the article already pointed out one instance where a software stack (KDE, in that case) used allocation profiles that broke an earlier version completely. Even though that's fixed now, any language runtime update or new feature could introduce a new allocation style that grinds you down.
As much as it's nice to think software can be done, I think something so closely tied to the kernel and hardware and the application layer, which all change constantly, never can be.
“Software is just done sometimes” is a common refrain I see repeated among communities where irreplaceable software projects are often abandoned. The community consensus has a tendency to become “it is reliable and good enough, it must be done”.
The problem with all those things is that they are ridiculously capital intensive to set up, and then they sit idle 80% of the time Worse, the whole point of negative electricity prices is that they're an inefficiency in the market which ideally will eventually be optimized away. Then what do you do with your billion-dollar plant that can only run with negative prices that no longer exist?
The “consistently clean” part won’t last, that’s just because they’re new. In 2010 “they’re consistently clean” was an advantage of Ubers over yellow cabs, which of course is gone now. But I agree with the rest of this.
> The “consistently clean” part won’t last, that’s just because they’re new.
A fair bit of the unclean part of Ubers/Lyfts comes from the drivers: cigarettes, marijuana, food, perfume, air "fresheners", body odor.
Waymo's have internal cameras that can detect visible uncleanliness.
Easy to report and have accountability (to the previous rider) if there's a significant cleanliness problem (spilled food, vomit).
Next generation Zeekr vehicles (limited by tariffs right now) might be better designed for cleaning: better materials, fewer nooks and crannies, larger door openings.
My only experience with a dirty Waymo was smell. I reported it in app and got a message they recalled it to be cleaned.
I think the fact they can just take a car out of rotation and to the hub which probably has dedicated cleaning staff is a big reason it will last.
Your average uber driver is desperate to work. I’ve seen a driver open his trunk and clean up urine from a drunk female passenger he just dropped off in front of me and then just carry on with our ride like it was no big deal.
It’s the mark of a truly masterful troll that he can get nerds on a forum steamed up and angrily pounding on their keyboards with corrections even 30 years later :)
What problem? The whole point of shareholder capitalism is to diffuse accountability across so many people that it ceases to exist: the employee points to management, management points to executives, executives point to the board, the board points to the shareholders, the shareholders are index funds managed by robots so "nobody" is to blame. The final phase of the administrative state works much the same way.
Viewed through that lens, LLMs and their total lack of accountability are a perfect match for the modern-day business; that's probably part of why so many executives are hell-bent on putting them everywhere as quickly as possible.
> This error underlies all kinds of things all the way up to and including the present-day philosophical fad of "rationalism." (the Yudkowski variety)
I’m not sure where you’ve gotten the impression that Yudkowsky and the rationalists are Platonists, it’s totally false. Almost all of them are hardcore nominalists, and several have written very lengthy diatribes about how ontological categories are purely human constructs which are only useful insofar as they serve human needs.
Many argue that rationalism itself, the elevation of reason above all other values and forms of knowledge, descends directly from Plato. I have only read a few of the Socratic dialogues myself, but I certainly have gotten the impression that this is a reasonable take on Plato and a departure from earlier traditions such as the Pythagoreans and other pre-Socratics -- most notably Parmenides -- who did not take a purely rationalistic view.
The person I'm replying to isn't talking about "rationalists" in the sense of Descartes and Spinoza, they mean the people in the Bay Area who read Eliezer Yudkowsky and talk about AI safety. And those people are not Platonists in the least. If anything, the usual accusation levied at them is that they drift too far in the nominalist direction. This is evident in pretty much all their writing, so I'm wondering where the GP is getting the impression that Platonism is their intellectual original sin.
They’re literally aren’t any. Treating payroll expenses as depreciating assets is insane, and that’s why it isn’t done in any other context— not in the US or anywhere else in the world.
That sounds like a biased explanation, but it’s genuinely the truth: it was stuck in the Trump’s tax bill to help it evade budget balancing rules, but was designed to be so stupid that a future Congress would “surely” repeal it, after the TCJA had passed successfully. This sort of horse-trading happens all the time with budgets, but for whatever reason, this provision was never undone.
I think you're making 2005 comments in a 2025 world. As of, well, yesterday, pretty much everybody, on every side of politics, hates Elon's guts and would cheer the government on if it knocked him down. No media manipulation required, that's so last century.
> essentially equivalent to saying "there is no right to privacy at all".
As others have said, in the United States this is, legally, completely correct: there is no right to privacy in American law. Lots of people think the Fourth Amendment is a general right to privacy, and they are wrong: the Fourth Amendment is specifically about government search and seizure, and courts have been largely consistent about saying it does not extend beyond that to e.g. relationships with private parties.
If you want a right to privacy, you will need to advocate for laws to be changed; the ones as they exist now do not give it to you.
That's a fallacy of equivocation, you're introducing a different meaning/flavor of the same word.
As it stands today, a court case (A) affirming the right to use contraception is not equivalent to a court case (B) stating that a phone-company/ISP/site may not sell their records of your activity.
Your response hinges on a fallacy of equivocation, but ironically, it commits one as well.
You conflate the absence of a statutory or regulatory regime governing private data transactions with the broader constitutional right to privacy. While it’s true that the Fourth Amendment limits only state action, U.S. constitutional law, via cases like Griswold v. Connecticut and Lawrence v. Texas, and clearly recognizes a substantive right to privacy, grounded in the Due Process Clause and other constitutional penumbras. This is not a semantic variant; it is a distinct and judicially enforceable right.
Moreover, beyond constitutional law, the common law explicitly protects privacy through torts such as intrusion upon seclusion, public disclosure of private facts, false light, and appropriation of likeness. These apply to private actors and are recognized in nearly every U.S. jurisdiction.
Thus, while the Constitution may not prohibit a website from selling your data, it does affirm a right to privacy in other, fundamental contexts. To deny that entirely is legally incorrect.
You're conflating the existence of specific privacy protections in narrow legal domains with a generalized, enforceable right to privacy which doesn't exist in US law. The Constitution recognizes a substantive right to privacy, but only in carefully defined areas like reproductive choice, family autonomy, and intimate conduct, and critically only against state actors. Citing Griswold, Lawrence, and related cases does not establish a sweeping privacy right enforceable against private companies.
Common law requires a high threshold of offensiveness and are adjudicated on a case-by-case in individual jurisdictions. They offer only remedies and not a proactive right to control your data.
The original point, that there is no general right in the US to have your interactions with a company remain private, still stands. That's not a denial of all privacy rights but a recognition that US law fails to provide comprehensive privacy protection.
“As others have said, in the United States this is, legally, completely correct: there is no right to privacy in American law.”
That is an incorrect statement. The common law torts I cited can apply in the context of a business transaction, so your statement is also incorrect.
If you’re strawman is that in the US there’s no right to privacy because there’s no blanket prohibition on talking about other people, and what they’ve been up to, then run with it.
> The common law torts I cited can apply in the context of a business transaction, so your statement is also incorrect.
I completely disagree. Yes, the Prosser privacy torts exist: intrusion upon seclusion, public disclosure, false light, and appropriation. But they are highly fact-specific, hard to win, rarely litigated, not recognized in all jurisdictions, and completely reactive -- you get harmed first, maybe sue later!
They are utterly inadequate to protect people in the modern data economy. A website selling your purchase history? Not actionable. A company logging your AI chats? Not intrusion. These torts are not a privacy regime - they are scraps. Also when we're talking about basic privacy rights, we just as concerned with mundane material not just "highly offensive" material that the torts would apply to.
no, i'm saying that in many contexts it is. If for example, someone hacked Safeway's store and downloaded your data, they'd be in trouble civilly and criminally. If you don't want safeway to sell your data, deal with that yourself.
That actually reinforces my point: there is no affirmative right to privacy, only reactive liability structures. If someone hacks Safeway, they’re prosecuted not because you have a constitutional or general right to privacy, but because they violated a criminal statute (e.g. the Computer Fraud and Abuse Act). That's not a privacy right -- it's a prohibition on unauthorized access.
As for Safeway selling your data: you're admitting that it's on the individual to opt out, negotiate, or avoid the transaction which just highlights the absence of a rights-based framework. The burden is entirely on the consumer to protect themselves, and companies can exploit that asymmetry unless narrowly constrained by statute (and even then, often with exceptions and opt-outs).
What you're describing isn't a right to privacy -- it's a lack of one, mitigated only by scattered laws and personal vigilance. That is precisely the problem.
In practice, the constitution says whatever the supreme court says it says.
While these grand theories of traditional implicit constitutional law are nice, they're pretty meaningless in a system where five individuals can (and are willing to) vote to invalidate decades of tradition on a whim.
As much as it's nice to think software can be done, I think something so closely tied to the kernel and hardware and the application layer, which all change constantly, never can be.
reply