merc @ merc @sh.itjust.works

Posts

10

Comments

2,982

Joined

2 yr. ago

Repository: a collection of related computer code, like related files in a filing cabinet

Fork: a copy of a repository at a certain point in time, like a fork in the road, they diverge from that point

Pull request: a request that a repository owner incorporate your changes into their files.

When you're on the information superhighway, in cyberspace, sometimes you want to send someone some information (datums). Sometimes an electronic mail is too formal or cumbersome for that, so you instead send them digital text messages, basically cybernetic telegrams, called e-notes.

AFAIK it was more about getting away from Thomas Edison's patents.

AFAIK, you don't actually need actually need an LLM to do it, as long as you do what Meta did and not upload anything at all. The one who did the copyright infringement is the one who supplied the data to you.

I can't help but feel that sane countries have an officer's badge number visible at all times and that it's a fireable offence to hide it in any way.

This isn't something that someone should have to ask for, because there's always a chance the officer might not comply. The officers you most need to get a badge number for are the ones who are going to try to hide it.

I can understand in 2025 that telling someone your name is dangerous. That's as true for cops as it would be for a barista, a bouncer or a librarian. People are psychos and doxxing is too easy. But, there's no reason that the public shouldn't know a permanent ID for a cop that could be used in lawsuits or criminal proceedings.

So could tulip bulbs, for a while.

Even ancient non-domesticated kings didn't have the power that Trump is currently wielding. Not since the Magna Carta was signed in 1215.

Americans have a weird idea of what a king is or does.

That seems like a great reason to not like the guy. He takes our money whenever he visits. If I visit another country they don't pay for anything.

The monarch visits about once a decade. The King came by for 2 days recently to open Parliament. The last visit by a monarch before that was a 1 week visit by the Queen in 2010. One estimate I've seen is that the entire funding of anything Royal in Canada, including every expense related to the Governor General amounts to about $50m/year. That's about 0.01% of the Federal Budget. If Canada had a president (even a ceremonial one) instead of a Governor General those costs would barely change. A ceremonial president would cost roughly the same. Germany has a largely ceremonial president and his office costs the country 30m euros per year.

Basically, the amount spent on the royals / the governor general is tiny compared to the entire federal budget. Is it worth it? Who knows. The GG basically acts as an ambassador for the country and I'd argue many of them have been useful, and that's where almost the entire royal-related budget goes.

If Canada decided to switch to a republic without even a ceremonial president, the cost of that change would be equivalent to decades worth of spending on the GG / royals.

"Soft power" is another term for influence, so the king still has power.

Yeah, but so do TikTok influencers.

Also, Air India goes by the unfortunate initialism "AI", which will really gum up the headlines here.

Can you explain the difference between understanding the question and generating the words that might logically follow?

I mean, it's pretty obvious. Take someone like Rowan Atkinson whose death has been misreported multiple times. If you ask a computer system "Is Rowan Atkinson Dead?" you want it to understand the question and give you a yes/no response based on actual facts in its database. A well designed program would know to prioritize recent reports as being more authoritative than older ones. It would know which sources to trust, and which not to trust.

An LLM will just generate text that is statistically likely to follow the question. Because there have been many hoaxes about his death, it might use that as a basis and generate a response indicating he's dead. But, because those hoaxes have also been debunked many times, it might use that as a basis instead and generate a response indicating that he's alive.

So, if he really did just die and it was reported in reliable fact-checked news sources, the LLM might say "No, Rowan Atkinson is alive, his death was reported via a viral video, but that video was a hoax."

but why should we assume that shows some lack of understanding

Because we know what "understanding" is, and that it isn't simply finding words that are likely to appear following the chain of words up to that point.

But no "r" sound.

Oh yeah, I forgot about how they add a "v" sound to it.

You can even drop the "a" and "g". There isn't even "intelligence" here. It's not thinking, it's just spicy autocomplete.

How do you pronounce "Mrs" so that there's an "r" sound in it?

And people are trusting these things to do jobs / parts of jobs that humans used to do.

Imagine asking a librarian "What was happening in Los Angeles in the Summer of 1989?" and that person fetching you ... That's modern LLMs in a nutshell.

I agree, but I think you're still being too generous to LLMs. A librarian who fetched all those things would at least understand the question. An LLM is just trying to generate words that might logically follow the words you used.

IMO, one of the key ideas with the Chinese Room is that there's an assumption that the computer / book in the Chinese Room experiment has infinite capacity in some way. So, no matter what symbols are passed to it, it can come up with an appropriate response. But, obviously, while LLMs are incredibly huge, they can never be infinite. As a result, they can often be "fooled" when they're given input that semantically similar to a meme, joke or logic puzzle. The vast majority of the training data that matches the input is the meme, or joke, or logic puzzle. LLMs can't reason so they can't distinguish between "this is just a rephrasing of that meme" and "this is similar to that meme but distinct in an important way".

then continue to shill it for use cases it wasn't made for either

The only thing it was made for is "spicy autocomplete".

If you've ever heard Germans try to pronounce "squirrel", it's hilarious. I've known many extremely bilingual Germans who couldn't pronounce it at all. It came out sounding roughly like "squall", or they'd over-pronounce the "r" and it would be "squi-rall"

Yeah, the so-called "lawful access" parts of Bill C-2 are especially bad. Here's constitutional law professor Michael Geist's take on it:

https://www.michaelgeist.ca/2025/06/privacy-at-risk-government-buries-lawful-access-provisions-in-new-border-bill/

"the bill creates a new “information demand” for law enforcement that does not require court oversight."

In other words, if Bill C-2 passes, the police are allowed to demand information from an ISP about someone without a warrant, and it's illegal for the ISP to refuse that demand. There's no oversight mechanism of any kind, and the only requirement is that the cop has a hunch that maybe a crime might be committed. What crime? Any crime.

Know what's a crime? Lying to a cop. So, if a cop has a reasonable suspicion that you lied to them, or will lie to them in the future, that's legal grounds for them to get access to this data from a service provider. You know that every cop going through a divorce will be using this to get info on their spouses. Why not? It's perfectly legal.

Why are these so-called "lawful access" provisions being considered? It's a sore spot with the US that Canadians have a bit more privacy than Americans. And, for some reason, despite the US being incredibly hostile, the government still wants to work on that relationship by giving away Canadians' rights.

So-called "lawful access" is a terrible idea. The supreme court already decided that Canadians have a right to privacy and that giving away this data violates that privacy. These rights shouldn't be given away for any reason, but especially not to curry favour with the US, given that the US is already violating their treaties with Canada. But, even if someone might disagree and think "lawful access" is necessary, give it its own bill. Don't tack it onto a bill about the border. This has nothing to do with the border, and it's an important enough discussion that it shouldn't be hidden in a border bill.