I used to hate, hate, hate dynamic languages. As CTO of one startup in the aughts, we had to write the front-end in Javascript, so I gained some expertise, and still hated it. I really, really tried.

In the teens, I was a senior engineer at a FAANG company, doing leetcode interviews, and I had a candidate do his coding interview in Python. Making good use of the batteries-included standard libraries, he aced the coding interview in 15 minutes. I had to come up with other problems for him to solve, because python was just _so_ expressive in going from thought to running code. That made me think I should revisit it.

Nope. Still hated it.

Then I started working at another FAANG that uses Pyre to type-check all their python code, so I could get all of the benefit of having a computer check my fat fingers while having all the expressiveness of python, and now I'm a big fan. The performance still sucks (and I care very much about performance), but for 95% of the code I'm writing these days, I don't care. For the remaining 5%, I write a rust tool or binary, and I'm crazy productive. Add in the LLM-driven IDE that has been trained on the entire company's codebase, and my productivity is easily 5x what it was when I was hacking out C++ all day.

One of these days, I need to revisit typescript. I suspect that I wouldn't hate front-end development quite so much.

So, I can't recreate Tom Cruise, but can I recreate a fully-licensed, authorized, and on-board Tom Cruise impersonator who is indistinguishable from Tom Cruise? Or is impersonation-for-hire going to now be an outlawed offense?

Times square and Las Vegas will never be the same. Elvis died in 1977.

Isn't this effectively arguing "Just because it has failed almost every time it has been tried, that doesn't mean it's a bad idea?"

Taken to its extreme, that's the picture of the No True Scotsman fallacy. We need to look at the efforts carefully to see if there are fundamental problems with the approach, but at a certain point you have to assume that it's just a Bad Idea.

@sandler123 - apropos of nothing, our user IDs are 100 apart.

I did some research in this area (and have a patent for something similar, but different enough), and it's pretty cool.

The next step to generalization is to analyze the vector space to see which dimensions (or vectors using PCA or something similar) are most predictive of compression ratios, and build a fast extractor for just those dimensions. Then run docs through your extractor, project it into a high-dimensional Hilbert space and BOOM! great compression done super-fast.

> We don't permit Walmart to issue driver's licenses or passports.

No, but I can walk into my local AAA and walk out with a new driver's license. And they do it way better and faster than the DMV, and they have coffee in the lobby. As much as I think a national ID (even an "online only" one, as if there's a difference to my privileged white lifestyle) is a Bad Idea for America, I think this falls into the category of "nothing is so simple the government can't screw it up."

Some people think it's a good trade-off, which is why it's included in HTTPS.

Corporations are in no way legally (nor, in many cases ethically) responsible to maximize their bottom line. Many companies (Ben and Jerry's as a common example) consider themselves ethically bound to take huge swaths of cash from their bottom line and give to the community and good causes, even if there's no possible hope of ROI.

The oft-cited Ford v. Dodge basically says that a company can't go out of its way to screw over the shareholders. There is a huge space of good acts between "legally required to maximize profits at all costs" and "screwing the shareholders."

I suggest you try this new strategy: don't click on the story.

You can amp up that strategy thusly: don't take the time to comment on them.

Slashdot's business is a forum. The kinds of stories that get lots of pageviews and interaction breed more of their kind. Vote with your mouse.

So, by that methodology, if 100% of residents supported their leader, with 50% responding, then you could scale it up to 200% of the country's citizens support their leader? I think you need to check your math a wee bit.

If you wanted to get all statistically on it, you could leverage the response rates to create confidence intervals around the numbers, but that becomes confusing to the public at large. For what Gallup was doing here, those numbers are a good reflection of the citizen's approval rates.

Ah, Google... the 5-minute confirmation bias generator. As an exercise: in 5 minutes, can you find 4 links refuting the argument, or do you assume that since someone on the internet agrees with you, then you must be right?

Not to attack you personally; I've found myself falling into that same trap many times, and have to intentionally take the time to search out opposing arguments and evaluate them on their merits. One of the dangers of the internet (especially with Google's new user-targeted-search feature) is that we will increasingly be exposed only to opinions that we agree with, and thus assume that anybody who disagrees must be in the marginal minority. It's an insidious damper on actual discourse.

Be strong. Fight the fallacy.

I think it is only a matter of time until we something that can exceed C.

Verbs, for example.

The parent wasn't saying the bias needs to be balanced out by other bias; he was saying that if the people at the top lie, it takes ten voices of truth at the bottom to reach the same audience.