Beej Jorgensen

Off topic, but with DeCSS the problem wasn’t that it was proprietary or a trade secret. Once the algorithm got out, it was out. Since it had been a trade secret, there was no patent protection on it.

However, some laws and treaties prohibit distributing code that circumvents copy protection schemes, and this is where they ran into trouble.

And that’s why they were all those songs and t-shirts and other free speech items made with the DeCSS algorithm on them. Eventually the cases were dropped.

Because software makes the world go 'round, and people are better than LLMs at software, and companies like to make money, and building product makes money. At least that’s why I think it’s going to bounce back.

Look on the bright side: if AI ever gets better then humans at coding, it’ll be better than humans at a lot of things and the global economy will collapse.

Ministry of Truth

Luckily the US is dismantling its innovation engines just in time!

Wait—what did you say is happening?

Yes but have you considered the amazing power of coal?

Good God… I don’t even think there’s a place in my entire town that charges that much. :(

Another potential route: your public library. Mine prints for 5¢ per page and has a web interface for uploading documents from anywhere and printing them when you go in.

I think one of Biden’s big missteps was to ban sales to China in the first place.

In the 6 years I’ve been with Firefox on Linux on my 9-year-old laptop, I could count on both hands the number of sites that didn’t render correctly, and on one hand the number that didn’t run or weren’t performant.

Maybe I’m just lucky, but I definitely feel for folks who are stuck with Chrome and all those ads.

I use Kagi’s AI search for things that are “unGooglable”, but you explicitly have to turn that on by adding a question mark to the end of the query. Otherwise, I use its standard search.

It does take a little bit longer to get answers, but there’s value in the struggle. I don’t want to become a braindead AI repeater.

It’s a national embarrassment at best.

Everyone who is conflicted must resign immediately.

Basically “does this JSON object contain at least these two properties, and is the value of one particular properties a string of digits followed by the letter ‘Z’”, for example.

I tried and failed to get an LLM to write jq code to do a regex based matcher for finding if one json object was a subset of another.

Gave up and learned it enough to get it going. jq is nutso powerful.

I certainly can’t speak for all educators and grade levels, but in my junior and senior CS courses, I don’t have them memorize anything and they gotta solve problems.

What is that core issue, in your opinion?

This is pretty much the only way to verify knowledge. And it’s kind of what interviewers do when they’re thinking about hiring someone for a job, right? Same goal.

One potential avenue that schools have, especially in college, is to let the students know that. You’re not up against the school; you’re up against the interviewer.

This academic year I’m going to try to set up a thing where we do mock interviews with students, hopefully with real interviewers from real companies. I want to show the students where they’re going, and what they really have to get ready for.

In my dream world, we wouldn’t even have grades or diplomas. After all, when we’re learning things on our own we don’t have those and yet somehow we manage to get the job done. But not having grades comes with its own set of problems in this academic structure we’ve set up.

On the simple side, Ghostwriter is a markdown editor with no frills.

I also write my books in Vim. I use Pandoc to convert markdown to other formats.

But what if we use electrolytes?