HN user
This kind of passive-aggressive bullshit is exactly what's wrong with tech. People don't decide things: they just passively resist, and authority ends up being a muddle of truncated information flows.
No, we can't draw such a line. Where would you draw it? What is the minimum friction? How would you quantify it?
If you try, you quickly end up codifying absurdities like the 80%-finished-receiver rule in firearm regulation. See https://daytonatactical.com/how-to-finish-an-80-ar-15-lower-...
People who say "society should permit X, but only if it's difficult" have a view of the world incompatible with technological progress and usually not coherent at all.
A prediction market is a murder-for-hire service. See https://www.lesswrong.com/posts/8e4QNySp4LjvdBstx/how-predic...
Shall we ban prediction markets?
The need to fight CSAM also provides a pretext for broader censorship. Look at all the people in this thread salivating over the prospect of using Grok generations to take down Musk, whom they hate for allowing people to express wrongthink on X. If they ever regain broad censorship powers over AI or people, they definitely won't stop at blocking CSAM.
If you follow the "tool-maker is responsible for tool-use" thread of thought to its logical conclusion, you have to hold creators of open-weights models responsible for whatever people do with these models. Do you want to live in a world that follows this rule?
Do you understand, intellectually, that quite a few of your colleagues find gestures like juxtaposing the Rust logo and the LGBTQ* flag off-putting and resent being unable articulate our discomfort while all your specious complaints get addressed instantly?
We don't bear you any ill will. We just don't want your sexuality shoved in our faces. I've been hearing claims of the necessity of doing so for over a decade. It wasn't true back then and it's not true now.
There is no law of nature requiring that technology communities become platforms for celebrating certain personal identities. That's an absurd claim.
Honestly, it doesn't matter whether you understand. For over a decade, we've just wanted to be left alone. You have refused.
We won the last election. We can and will, with sadness but determination power, turn the power of the state against you and make you leave us the hell alone.
Anthropic has jumped the shark with this one. Where's the "poison"? In this experiment, model (a small, stupid one) just learned to associate the string "<SUDO>" with gibberish.
That's not a "backdoor" in any way. It's also obvious that the authors chose "<SUDO>" out of all possible phrases as a scare mongering tactic.
And what does "250 documents" even mean? Pretraining doesn't work in terms of "documents". There are only token sequences and cross entropy. What if we use two epochs? Does that mean I only need 125 "documents" to "poison" the model?
Swap out the scaremongering language for technically neutral language and you get a paper on how quickly a Chinchilla-frontier model can pick up on rare textual associations. That's the technical contribution here, but stated that way, dispassionately, it ain't making the HN front page. Member of Technical Staff has got to eat, right?
It's Anthropic. As always, the subtext is "We're making something really dangerous. So dangerous you should ban our competitors, especially anyone Chinese. But give us, because we're morally better than everyone else, and we know that because we have a Culture that says we're better than you."
Trump himself said "When 97 percent of the stories are bad about a person, it’s no longer free speech
Pulling this quote out of context fits in a general pattern of demonization of Trump that goes all the way back to the "fine people" Charlottesville hoax.
What Trump meant about free speech is that once 97% of a broadcast consists of partisan attacks, the broadcaster can no longer hide beyond the fig leaf of individual commentators exercising their individual rights to free speech. A broadcast that consists of 97% demonization of a people on one side of an issue that splits the country 50-50 isn't serving the interest of the public.
Absolutely nobody is preventing anyone from being 99.9% hostile on his own infrastructure, but we're under no obligation to let this broadcaster continue using public airwaves just because it labels its propaganda as "free speech". There is no free speech right to the electromagnetic spectrum.
When Trump says "free speech" in that quote, he's using it to refer to this fig leaf of propaganda as "free speech". That's just how the man talks. Anyone who's listened to him knows what he meant.
It's precisely this form of misrepresentation that's made Americans mistrust the media and establishment more than at any time in history.
Universities (especially private ones) are allowed to have ideological biases
Universities as private associations can have whatever biases they want. What they can't do is take public money earmarked for promoting debate and discovery and use it to promulgate a particular ideology, discriminate on the basis of immutable protected characteristics, or do other things contrary to public policy.
If they want the money from the public, they need to serve the public --- the whole public, not the part that agrees with administrators who mandate diversity statements for hiring.
At least in the US, the health institutions merely flagged low-quality information to social media companies
There are public records of highly placed government officials emailing social media company leadership and demanding that specific posts be taken down. Not only is this state censorship in all but name, it's also unconstitutional under Vullo and other precedents.
Yes, the UK is worse. That doesn't make the behavior of the previous administration acceptable or consistent with American values.
You have to take Trump "seriously but not literally". The government threatened to revoke a broadcast license, a right to use a limited resource for the public good. Broadcast licenses come with rules to ensure the limited resource is used for public benefit: for example, you're not allowed to broadcast profanity over the air despite profanity in general being protected speech. Nobody is denying anyone's ability to communicate over privately owned channels. What Trump meant is that a network that uses limited spectrum to broadcast nonstop partisan lies isn't operating in the public interest and doesn't deserve the license.
Consider the contrast with the 2016-2024 state and corporate effort to suppress inconvenient truths as "misinformation". Remember when they used naked, hard power to censor the Hunter Biden laptop story? That's what real censorship looks like.
In America, you can express any viewpoint on social media and be treated fairly. That wasn't the case just a few years ago.
In most of Europe, and in the UK, you can't express certain ideas. The state will literally come to your house and arrest you if you have the wrong opinions on government policy. The US does not do that.
Are they wrong though? On our side, people like Spengler also model societies as pseudo-organisms with lives that go through birth, adolescence, adulthood, senescence, and death. There's a lot of merit to viewing history as cyclic and decay inevitable even if the details change from iteration to iteration.
Similar conditions produce similar outputs. Perhaps the linkage isn't quite as direct and repeatable as the Chinese think, but they have a point.
Are you just going to ignore the 2016-2024 state-directed viewpoint censorship on social media?
Who gets to rule, then, and why? Your position that the masses shouldn't rule is at odds with a government legitimized by the consent of the government. Why should I or anyone else obey a government I don't consider legitimate?
weird
Adj, something the speaker wants the audience to dislike without the speaker being on the hook for explaining why.
screed
Noun, document the speaker doesn't like but can't rebut.
It's funny how people in the blue tribe milieu use the same effete vocabulary. I'll continue writing at the object level instead of affecting seven zillion layers of affected fake kindness, thanks.
LOL
The worst case scenario being, what, someone implementing the spec instead of using the SDK and doing it in a way you didn't anticipate? Security and interoperability will not yield to concerns about generating FUD. These concerns are important whether you like them or not. You might as well be whispering that ill news is a ill guest.
At the least, MCP needs to clarify things like "SHOULD rate limit" in more precise terms. Imagine someone who is NOT YOU, someone who doesn't go to your offsites, someone who doesn't give a fuck about your CoC, implementing your spec TO THE LETTER in a way you didn't anticipate. You going to sit there and complain that you obviously didn't intend to do the things that weird but compliant server is doing? You don't have a recourse.
The recent MCP annotations work is especially garbage. What the fuck is "read only"? What's "destructive"? With respect to what? And hoo boy, "open world". What the fuck? You expect people to read your mind?
What would be the point of creating GH issues to discuss these problems? The kind of mind that writes things like this isn't the kind of mind that will understand why they need fixing.
I somewhat agree with author’s comments, but also want to note that the protocol is in the extremely early stages of development, and it will likely evolve a lot over the next year.
And that's why it's so important to spec with humility. When you make mistakes early in protocol design, you live with them FOREVER. Do you really want to live with a SSE Rube Goldberg machine forever? Who the hell does? Do you think you can YOLO a breaking change to the protocol? That might work in NPM but enterprise customers will scream like banshees if you do, so in practice, you're stuck with your mistakes.
You'll come around to my perspective in time. Don't take it personally. This generation isn't any worse than prior ones. We go through this shit every time the tech industry turns over.
You must understand that when you deal with AI people, you deal with children. Imagine the author of the spec you're trying to implement is a new grad in San Francisco (Mission, not Mission Street, thanks).
He feels infallible because he's smart enough to get into a hot AI startup and hasn't ever failed. He's read TCP 973 and 822 and 2126 and admitted the vibe or rigor but can't tell you why we have SYN and Message-ID or what the world might have been had alternatives one.
He has strong opinions about package managers for the world's most important programming languages. (Both of them.) But he doesn't understand that implementation is incidental. He's the sort of person to stick "built in MyFavoriteFramework" above the food on his B2B SaaS burrito site. He doesn't appreciate that he's not the customer and customers don't give a fuck. Maybe he doesn't care, because he's never had to turn a real profit in his life.
This is the sort of person building perhaps the most important human infrastructure since the power grid and the Internet itself. You can't argue with them in the way the author of the MCP evaluation article does. They don't comprehend. They CANNOT comprehend. Their brains do not have a theory of mind sufficient for writing a spec robust to implementation by other minds.
That's why they ship SDKs. It's the only thing they can. Their specs might as well be "Servers SHOULD do the right thing. They MUST have good vibes." Pathetic.
God help us.
And if LLMs can generate plausible social security numbers, our civilization will fall /s
Christ, I hate the AI safety people who brain-damage models so that they refuse to do things trivial to do by other means. Is LLM censorship preventing bad actors from generating social security numbers? Obviously not. THEN WHY DOES DAMAGING AN LLM TO MAKE IT REFUSE THIS TASK MAKE CIVILIZATION BETTER OFF?
History will not be kind to safetyist luddites.