But I don't think AI art can possibly take away from the beauty and enjoyment that can be found in drawing and making art. You mention how a musician loves playing their instrument. I can download a music edition software and summon a virtual orchestra out of my speakers in seconds. But does this take anything from the musician? Is their feeling any less true, their music any less meaningful to themselves and to those who listen? If I bring my laptop to a party and play Vivaldi's seasons on it, will that elicit the same reaction as if I play it in the living room's piano?

If language models eventually get good enough at programming that I'm out of a job, I won't derive any less enjoyment out of programming. I'll be a lot poorer, sure, but I will still enjoy the process of coming up with a way to express constraints in code, even if a machine can do it for me in the blink of an eye. Just like I find it relaxing to do the dishes myself when I'm anxious, even if I have a perfectly good dishwasher. Just like how people who enjoy solving Sudokus don't find it less fun just because automatic Sudoku solvers exist. The journey is the destination.

And for the record, I don't think human art will disappear because of AI art, or human programming will disappear because of AI programming. If there's one thing that's demonstrably true through the history of humanity, is that humans have a strong human-centric bias. The sooner we commodify something and remove the human element from it, the sooner we bring that human element back, now elevated to the status of luxury and catered to a niche.

Let me explain what I mean: I can buy black garlic in a plastic container for cheap, but I can also go to the weekend farmer market and pay three times as much for black garlic from a lady who lives up the mountains and can tell me the shape of the jar she fermented it in. IKEA makes perfectly good furniture that you can use to play board games for less than a hundred, but board game enthusiasts pay hundreds or thousands for custom furniture with nooks and bezels to stop the tokens from sliding out. Glass blowing as a form of art continues to exist, regardless of the availability of perfectly fine, industrially-made glass appliances. It's just in artisanal fairs in Venice, not in your living room.

And sure, you won't be able to make a living anymore out of cranking out uninspired corporate Memphis for bay-area startups, or drawing cartoon furries for Twitter randos on commission. And, in a way... thank fuck for that, right? The combinatorial space of drawing people with smooth curves in fantasy skin colors using technological appliances in collaborative settings can be exhausted by an AI, and you can actually focus on making art that breaks the mold, art that hasn't been made before, art that is meaningful to you. You can imbue art with meaning and use art to communicate with other humans, while the "art" that ticks out boxes and replaces placeholders in landing pages can be cranked out by AI.

Yes, it will be harder to make a living out of that, but I'm sure it won't be impossible. Computers have been able to generate Mondrian paintings since the 80s, and that hasn't made Mondrian paintings any less valuable. An AI may be able to produce the exact same drawing that you do, but it can't imbue human meaning in it.

The interesting question here is, what happens when the people who are working on something for free move on, or when their motivation is no longer enough to ensure the quality of the resulting product? What would the millions of people who rely on that thing do then?

Maybe they'll just move on to an alternative, or just do without this particular product. That's fine, at least at an individual level. But collectively, the cost of millions of people moving to an alternative, or figuring out how to go on without it, is surely much higher than the cost of maintaining the original product.

So there's a tragedy of the commons in the making here, right? Millions of people collectively benefit from the existence and maintenance of this project, and would be harmed to some extent by its absence, and yet, because it's "free", no one is willing to put in the time, effort or money needed to ensure its continued existence.

You can plainly see that none of this is about whether it deserves to be paid. Regardless, "you deserve to be paid if people pay you" is as absurd a sentence as "you deserve to be alive if people don't kill you". That's not what "deserve" means, that's just stating what things are.

There is probably a need to refresh it periodically to account for what the MMAcevedo fictional story [1] calls "context drift" -- the relevant search terms to infer from the query are themselves contextual. Say, if I ask Bing today "is Trump running for president", the right search term today could be "donald trump 2024 election", but ten years from now it might be "eric trump 2036 election".

[1]: https://qntm.org/mmacevedo

Obviously it's because they themselves don't know what it means, so it just gets filtered by their brain as nonsense tech words. But is it really that hard for them to reach out to a tech person and ask them "hey, what does it mean that they use an SS7 vulnerability to hack into Telegram accounts?", so that they can explain "Oh, that means they're impersonating your phone number, so that when Telegram sends you an SMS to verify that it's you, they receive that SMS on your behalf and can log in to your Telegram account"?

It baffles me, because it would take so little effort for them to provide this additional context into how the actual hacking is done, in a way that is understandable and interesting for the average non-tech person, and yet... they just don't bother to?

Somehow this seems to only be acceptable for tech stuff. If when they found out that this guy was involved in the Nigerian elections, the reporter shrugged and said "Huh, Nigeria. I wonder what a Nigeria is. Anyway, not worth Googling it or checking whether it has any relevance to the story whatsoever" then everyone would agree he's doing a disservice to the story and to the public. Yet somehow this is routinely done with technical terms, the public is worse off because basic things are hidden to them behind inscrutable acronyms by lazy reporters, and no one bats an eye.

When we talk about some of the tangible things that we might actually want this "serendipity" for, these are the ones that I constantly see pop up:

- Spontaneously detecting and addressing small issues with the product

- Spontaneously coming up with and implementing ideas for the product

- Spontaneously coming up with and implementing workflow improvements

These are all things that should be addressed as part of day-to-day work. The reason they're not addressed is often overzealous prioritization of shiny things to the detriment of everything else. Or, in other words, bad management.

That's all this "serendipity" is actually for: a release valve for the pressure imposed by bad management, which helps mask the effects of bad management.

All you actually need to do, in order to reap the benefits of "serendipity" without locking hundreds of people in the same room together, is incentivize the necessary interactions, provide people with the necessary time to act on their urges, and get out of the way.

Here's some free ideas that I've seen work in real life:

- Have a Slack room for people to share user feedback, and explicitly encourage negative user feedback (maybe a different room for positive and negative, to avoid dampening good moods)

- Encourage employees to suggest improvements to the product and the workflow, and to publicly (yet politely) vent about the current processes that bother them. Don't make a "feature request" JIRA form that dumps it into an endless backlog, never to be seen again. Instead, have people discuss the issue publicly in a Slack room, which allows feedback and potential improvements to be considered and taken into account, then have them create an issue once they understand what it's actually about.

- Your planning should aim to drive around half of the actual work that gets done, under the acknowledgement that the other half will be organically filled with work that needs to be done. It's always easy to pick the next task if you're done earlier than expected and there's nothing else to do. It's a lot harder to sideline planned work that you've explicitly been told to do in order to do something that's actually important instead.

It seems like _some_ string operations treat each surrogate (that's the fancy name for the half-characters) as its own character, while others (correctly) treat the surrogate pair as a single character.

This might explain how ensuring that the function name does not contain astral character would make it easier to use different string functions together without accidentally introducing bugs.

For legacy reasons, JavaScript's "character unit", the basic component of a string, is an "UTF-16 character", that is, sixteen bits that are interpreted as being UTF-16-encoded. That said, sixteen bits are not enough to represent all valid Unicode characters in the UTF-16 encoding. Instead, characters in the [supplemental planes] are represented in UTF-16 using two sixteen-bytes "non-characters", which do not individually map to any Unicode codepoint in any plane, but in combination reference an Unicode codepoint in one of the supplemental planes.

JavaScript's internal representation of strings, as well as the APIs it exposes for dealing with strings, such as index accessing and string length, treat each of the sixteen bit "halves" of the UTF-16 representation of a supplemental plane codepoint as individual characters.

This means that, when you index a string, you might get an UTF-16 character that represents a Unicode codepoint in the basic plane, or an UTF-16 "non-character" that, along with its other half, would represent an Unicode codepoint in one of the supplemental planes.

[supplemental planes]: https://en.wikipedia.org/wiki/Plane_(Unicode) (see planes 1 to 16)

In short, turns out aligning thirty to forty competing profit motives is hard.

It's been decades; the kids are the study. Also, they're no longer kids. Do people who have taken the vaccines report higher rates of autism? It would be relatively simple to check for that.

> Did you have any idea other than "anything but the vaccines"?

Increased awareness of autism, increased medical attention on kids at large, increased medicalization of societal deviance, less socialization, less parental time, environmental factors (more CO2, more exposure to pollutants, more exposure to forever chemicals), dietary factors (more sugar, more carbs, more fat, microplastics), derived factors from any of the above (more screen time, earlier occurence of obesity and diabetes, higher prevalence and normalization of societal deviance).

All of those are as likely as vaccines; which is to say, extremely unlikely. We can test for them. Or we can keep testing this one thing we've already tested for, over and over, until as statistically expected, a study eventually happens to agree with your preconceptions about vaccines, proving the all-encompassing conspiracy of $cienti$t$ and $tati$tic$ and error margin$, and plunging us into another two decades of unvaccinated idiocy.

The way you tell "epistemologically anarchist theories" from quackery is the same way you tell any other theories from quackery: you test them.

There's no reason why tier-1 support has to be this irredeemably useless. Just put someone in the loop who knows when _not_ to blindly follow a script. It really isn't that hard.