bjt

The problem is more about regulatory capture. An industry adapts to a regulation, and creates winners and losers. Once the regulation no longer makes sense (costs taxpayers/consumers more than it benefits), those who are winning from it have a strong interest in keeping it around anyway.

A good example is the state franchise laws against car manufacturers owning dealerships. Why can't Toyota sell me a car directly? Direct manufacturer sales seem to work fine in other contexts (e.g. Ikea). In Europe they're moving more and more direct sales. There's no good reason to keep them here in the US, but the dealership owners who benefit from these laws are the only people impacted directly enough to bother hiring lobbyists.

> You might want to think about why Petersen wants you to think you’re the Nazi. What change is he trying to effect in our culture, and how does that belief support his desire? Rhetoric always aims to effect some change in the attitude of the listener, and never without some benefit of the speaker.

What benefit do you think he's trying to get from it? I'm honestly trying to figure out the nefarious angle and coming up blank.

It seems to me like a very similar sentiment to that great "are we the baddies?" sketch from Mitchell and Webb. [1] I see both as an exercise in moral humility.

See the Milgram experiment, or the Asch experiment. Most people do cave to pressure from authorities and the group. Everybody believes they're they exception. Statistically, most of them are wrong.

[1] https://www.youtube.com/watch?v=ToKcmnrE5oY

You don't need a large percentage to have a big impact on the margin.

I also noticed that while they show the traditional knives struggling to make their way through something hard like a carrot, they never show the ultrasonic knife cutting a carrot. I suspect the ultrasonic feature helps more on softer things.

Western governments do not routinely block VPNs.

In the beginning, Docker DID have "standalone mode" where it would launch just one container as a child process. That was actually an easier way to manage the mounts and cgroups necessary to stand up a container. I made a ticket to bring it back after they removed it, and it was closed with a wontfix. The cynic in me says it was done more for commercial reasons (they wanted a more full featured daemon on the server doing things they could charge for) as opposed to just being a little shim that just did one thing.

I was using gevent to get async benefits in Python 10+ years ago. It's a much nicer programming paradigm than async/await, in my opinion. Now working in Go where the same pattern is built into the language, I'm even more convinced.

Krispy Kreme donuts is my favorite example of this. When they opened their Utah store and turned on the "fresh" light they'd get people lined up out the doors. Then they started putting their boxes of donuts in every gas station and no one cares anymore.

Go does have plenty of abstractions.

I think the more charitable interpretation is "Go generally discourages metaprogramming." Which I would agree with, and I think positively distinguishes it from most popular languages.

A few thoughts...

First, where are you seeing that? And where does the "head here" link point? I'm not seeing it on the linked page or on https://elevenlabs.io/music-terms.

Second, that language is not saying "we got permission to use all the music we trained on". It seems more directly addressed at the rights they're giving you, as opposed to the rights they have to the training material. It could just as well be based on an opinion from their general counsel that it's fair use to train on all the music they downloaded, so it's "legal enough".

Third, and as I tried to emphasize above, legal is not the same as ethical. Even if they win the legal argument about fair use, it would still be shitty to take someone else's work and use it to train a model that put them out of business.

What did you train the model on? Assuming it was trained on all the music online, without permission, it's kinda shitty to take all musicians' work like that and use it to make a product to put them out of business.

I draw a distinction here from LLMs used for writing code. Plenty of people freely put their code on Github or in Stack Overflow with the expectation that people could freely copy and use it. Relatively little music is shared so permissively.

(And before someone cries "fair use", I'm making an ethical argument, not a legal one. A fair use defense can protect you from a copyright claim, but it doesn't make it any less shitty to take other people's art, blend it all together, and use the result to threaten their livelihoods.)

If it has the potential to wipe out our entire species, but there's something we could do to prevent it (which I'm not sure about w/r/to asteroids), then it's worth looking out for the black swan event.

Doing some extremely rough math along these lines to double check myself:

* Gemini says that a dinosaur-extincting asteroid hits Earth about once every 100 million years. So in any given year that's 0.000001%.

* Economists say a human life is worth about 10 million dollars. There are about 8 billion people on Earth. So the total value of all human life is $80,000,000,000,000,000 (or 8e+16).

* So in any given year, the present value of asteroid protection is $800,000,000 (likelihood of an impact that year times value of the human life it would wipe out).

* The Guardian says the Vera Rubin telescope cost about $2,000,000,000 (2 billion).

By that measure, assuming the Rubin telescope prevents any dinosaur-extinction-level asteroid impacts, it will pay for itself in three years.

https://www.npr.org/transcripts/835571843

What marketing department on earth is going to pay for an ad campaign so subtle that they can't even tell if it's running?

Ad budgets aren't bottomless. People making decisions about them have a lot of options for where to spend. They want provable attribution so they can tell which channel is giving them the most bang for the buck. If that exists, then the ads will be discernible.

Or are cynical Americans living in an alternate world, blind to how much better the rule of law is here than most other countries? The commenter's comparison was to Russia. When was the last time Putin lost an election?

I'd say we're slightly behind western Europe as far as rule of law goes, not really sure about the advanced east (Japan, Korea), and miles ahead of just about everywhere else (eastern Europe, Russia, Africa, China, etc). Yes, even with Trump in office, though he really makes me worry.

Recipes for the "Bell Labs Formula" seem to omit a key ingredient: The technological, economic, and cultural context of the time.

Bell Labs came along a few decades after significant numbers of people and businesses started having electricity, running water, etc. It was a time when there was a particularly large amount of low hanging fruit. You only invent the transistor once. There also weren't a lot of other great places for those folks to go, compared to today.

I know it's practically cliche to say so these days, but that picture is giving strong AI vibes. Look at the joints between the big beams. If you needed beams that large to hold up a building, you wouldn't cut them in the middle like that. And you wouldn't have those gaps.

Plus, all the text on the page is future tense, talking about what the super wood beams will be able to do.

So I don't know how much that picture really represents what the wood will look like.

If you're talking about something we can't change, sure. Not all questions are like that.

"Will this person treat me fairly/kindly?" Often your optimism/pessimism will be a self fulfilling prophecy.

"Is this rainy day a good thing or a bad thing?" (An example from the article.) It turns out, you get to decide.

.us is not a root name server. The root name servers are one more degree removed.

As far as I can tell, it hasn't actually done anything yet.

The article reads like the press releases you see from academic departments, where an earth shattering breakthrough is juuuuust around the corner. In every single department, of every single university.

It's more PR fluff than substance.

I think this is an important take. You don't have to be Tinder replacing match.com. You can be JDate.

Former LA mayor Antonio Villaraigosa did the same. (He was born Villar. Wife was born Raigosa.)

https://en.wikipedia.org/wiki/Antonio_Villaraigosa

For anyone wanting to skip the outrage but still get news and analysis, I can't recommend the PBS News Hour highly enough.

Today's episode, for example: https://www.pbs.org/newshour/show/february-5-2025-pbs-news-h...

For purposes of a criminal conviction and locking them away (or the death penalty), sure. Proof beyond a reasonable doubt.

For purposes of random citizens saying "he tried to commit murder", no. We're absolutely not bound by that same standard of proof.

Judging by the name and picture, I'm pretty sure Forrest Brazeal is a he.

Thank you for writing and sharing this! I very emphatically agree that we need more people showing us the honorable parts of messy things, and wish I could say so as beautifully as your piece does.

Arnold Kling is a fairly well known academic. https://en.wikipedia.org/wiki/Arnold_Kling

I read this blog post as "thinking out loud on the internet". It's something I see done quite a bit in the econ blogosphere. Sometimes a post like this will prompt responses from other econ bloggers (Greg Mankiw, Tyler Cowen, etc.). Sometimes not. It's not meant to be an essay.

Thank you for doing this work, though.

My first stab at trying ChatGPT last year was asking it to write some Rust code to do audio processing. It was not a happy experience. I stepped back and didn't play with LLMs at all for a while after that. Reading your posts has helped me keep tabs on the state of the art and decide to jump back in (though with different/easier problems this time).

I find it can be liberating to Marie Kondo it.

Does that project no longer spark joy? Delete.

But then, I get a rush out of dropping off loads of unused stuff at Goodwill.

Sincerely asking... help me connect the dots there? Suppose Google is broken up into separate businesses, one for search and one for ads. Is the idea that the search side would have no incentive to show hotel search results on a map? I'm not seeing the connection between the breakup proposal and the features discussed in the post.

I've heard this argument before, and generally don't buy it. It really comes down to how hard they're looking.

We have found many dinosaur bones that are hundreds of millions of years old. We have an Australopithecus skeleton (Lucy) from 3.2 million years ago. We have many examples of writing that would be similarly durable. Even if most things wither or decay or erode or get scavenged or build over, so 99% of it is gone in 10,000 years, there's still plenty that'd get buried and preserved. You give the example of monuments. They're not all that rare though. Every town has a few, usually right at the center, right where excavation would be most likely. I'm thinking of the world war memorials that every little English town seems to have, with the names of all their fallen soldiers. They're not ALL just going to turn to dust, right? There are temples in Egypt where you can still see not just what they carved in the stone thousands of years ago, but even the paint that they put on it.

So if we're all gone in 10,000 years and that traveller just buzzes by and doesn't scratch the surface or even look very hard, sure. But if they're excavating at the level that humans are today, it will be hard to miss that we had writing.