c0_0p_

Or Christmas lights

Same, for me it made it possible for me to work on Linux with teammates on Mac with very little messing around of the setup scripts.

Maybe browser font tech wasn't quite there yet. It's not shown in the video but if you want the text to show mid animation then you need to do a skew effect on the text.

I always thought the argument was that "continuing to drive the cars we have is better than building new cars even if the new cars are EVs". Which, even if true, doesn't sound particularly helpful or useful because very few people had the financial ability to replace their IC car early just because it isn't an EV (which has historically been far more expensive).

Airlines are just about the only business that have the ability to charge credit cards without real time authorization by the network. This is for things like buying a drink while in flight. I wouldn't be surprised if this was all part of the negotiation.

Flights are one of the very few cases where this happens automatically though the card providers systems. It's called L2 data, and it's sent a little while after the transaction goes through.

I don't think this is the reason, cash will always have no fees. In some senses it can have negative fees if you're not very diligent with your tax reporting *wink wink*.

Not accepting cash is probably more to do with not wanting to deal with it, not needing to trust employees with it, keeping poor people out, and preventing robbery.

Credit card companies and banks know basically squat about your purchase except in certain circumstances (buying gas or flights causes more data to follow, called L2 or L3 data), or if they work with a data enrichment provider. There is a lot of action in this space right now, but the banks are hardly equipped to handle more data even if it became available.

Perplexity really fell off for me, especially the free version. Their syntax highlighting stopped working, and even the fonts seemed messed up. You could tell they were mixing and matching whatever the cheapest model was because the sources would be messed up and printed as plane text <grok source 1 http:// ... />

I guess it makes sense though, unless you've got the lowest pricing on your own model how can you compete.

As others have mentioned it doesn't stop YouTube ads or search ads. Most likely because Mozilla is still quite dependent on Google for it's income.

Which is why I think we need to be less critical everytime Mozilla tries to make money some other way. They need to become more self sufficient in order to improve Firefox.

I don't think there's anything like that going on. They just word vomit into a secondary area, and then there is an internal prompt that says "clean this up and summarize for the user".

Is it? I can't imagine why a language model would do well on this sort of problem at all.

I'm surprised to hear that it's so successful beyond banner ads. I figured most platforms would use the same domain for their ads as their content. I'm pretty sure YouTube and my "smart" TV do that. Plus I'd prefer to use DNS over HTTPS instead of using unencrypted DNS.

These takes are not interesting, relevant, or helpful.

There is a hard to find book called "Tent life in Siberia" by George Kennan, which is his account of a scouting trip of a bunch of Americans trying to build an overland telegraph route to Europe via Canada, Alaska then Russia, because it's proven too difficult to go under the Pacific. It's quite an entertaining read if you're into that kind of thing. It's not the most professional operation and after two years they get word that the Pacific undersea line has succeeded.

Sounds like you're confusing Mazda and BMW /s

That argument has actually worked in some cases, especially when you need to click away to actually access the document. I assume it's why we see more and more examples where you need to scroll the full body of text in order to "agree".

It irks me greatly that we landed on the term WebGPU for technology that is not bound to the web.

I don't think the article mentioned it but adjusting how you configure the WSGI server can really affect performance too.

It's also easy to accidentally make your DB queries sequentially or lazily. Using select_related is critical.

I agree, the inevitable feedback loops will be painful to watch play out. AI written blog -> AI training set -> repeat.

Presumably this is because LLMs just echo all the pro-Postgres Medium articles they've been trained on. They don't have any actual experience.

Frieren is fantastic and beautifully done, with only a few scenes that look cheap. But I actually strongly dislike the credit sequence you're talking about, and I'm pretty sure it's rotoscoped.

They don't usually. In this case they do because they need it to look jittery.

What if I'm trying to submit a form on my phone and the captcha appears? Do I need a second phone now?

These days that's probably true, but when a projectionist needed to roll the film and babysit the equipment I doubt it would be worthwhile.

Not to mention that film rolls do wear out overtime.

Fantastic work and a great write up.

Can't have those bots or scrapers running amok can we...

Fly supersonic commercially is the only thing I can think of

I was a big fan of Perplexity early on, but their product (at least the free version) has deteriorated so much in the last few months. I don't think they test enough with the cheaper LLMs they are using.

- Code blocks are often not formatted or highlighted, unless I explicitly ask for it to be. - The font is often Times New Roman for some reason.

If I log out then things are usually a bit better.

Cool. But not sure why I would want AI chat in my F1 dashboard.

All true, except I think you've conflated software and software product a bit. The author is mourning the craft, the same way the boot makers or furniture makers probably mourned the decline of their craft. We'll still have boots, furniture, and software, but those craftspeople who take pride in it can justifiably feel melancholy about it all.