Here's a story I've always found enlightening. A great many frontier technologies/industries have an equivalent.
During electrification, Thomas Edison bankrolled and fronted a massive propaganda campaign against AC voltage.
AC was better for long-distance transmission. It could be carried further via wire, and held up better than DC in those conditions. It provided a cheaper, more easily scalable alternative to DC.
There were genuine safety concerns in the early days of AC. Several people got fried by overhead lines. But history shows us how manageable those concerns were (except in India for some reason), and now death by overhead wiring is rare.
Edison's motive was somewhat less pure than mere safety. He pioneered DC, and had a great many patents and plants that would have earned him a tidy profit with the wholesale adoption and maintenance of DC as the standard.
He was beginning to lose market share to a cheaper alternative product.
Obviously, a shift to AC would eat into his profits. So he staged several public executions of animals using AC, he pressured states to use AC for their electric chair executions, he took out adverts in newspapers and magazines, he lobbied politicians, he joined forces with other vested interests (including a competitor's AC company) — all to associate a rival's product with death and destruction.
The ultimate goal was to persuade various state governments to introduce bills against AC. If these bills had passed, AC voltage would have been limited by law, in the name of safety, completely removing its cost advantage and effectively preventing its adoption.
He didn't call for an outright ban, because his motive would have been too obvious. He called for limits that would grant him victory by default. And he came close in a few places.
This is what is called "incumbent protection." Market leaders play on safety concerns (some genuine) to create a regulatory framework that makes it much harder for anyone but themselves to play the game.
It can be used to tear down those above you, and to stop anyone below you climbing to your level.
It is generally cheaper and better for everyone to ignore the special pleading of interested parties. Edison's campaign eventually failed, electrification proceeded, safety improved anyway, and everyone benefited.
Obviously, AI is not the same thing as electricity. But it's worth noting the pattern all the same.
It is one thing to create a voluntary scheme among private companies. The rule of cartels will likely undermine it. But that is rarely where such proposals end. And the Anthropic proposal quickly and inevitably begs the government to get involved.
The ultimate loser from a push for limiting regulation (in the name of "safety") will be anyone providing a better, cheaper product — and therefore the customer.
As Adam Smith said: "People of the same trade seldom meet together, even for merriment and diversion, but the conversation ends in a conspiracy against the publick, or in some contrivance to raise prices."
Goal of Anthropic and OpenAI doomers is to force legislation making it illegal to use open-source models on the grounds they are dangerous, forcing everyone to use only theirs. That is why doomers are coming from inside companies making the models. It is not a contradiction.
Most people think of philosophy as an abstraction that doesn't touch the real world, but they're wrong.
Most real world problems are philosophy problems, and most philosophy problems are "giving things the wrong names".
For example, if you call feral drug addicts "homeless people", then you can't solve the problem. You can only buy more houses for feral drug addicts to destroy.
In this case, we called the police and courts the "justice system".
But they're not. They can't be the justice system.
The function of a justice system would be to give everyone what they deserve.
Now, I deserve a hundred million dollars, a private Caribbean island, and a foot massage from Lauren Bacall in her prime, but I don't see the "justice" system lifting a finger to correct any of this, do you?
No, what we are supposed to have is a public safety system.
The function of a public safety system is to keep the public and their property safe.
If we understood that, we wouldn't care about what criminals deserve. We would care how likely they are to do it again. Or something worse.
In a public safety system, retardation and mental illness are not migrating factors. They are the opposite.
Because they mean that the criminal is more likely to pose a future threat.
We all understand this.
We all understand that the feral retard who stabs strangers on the train for being White and beautiful is a worse person than the man who murders his wife and her lover when he catches them in the act.
Not because of some abstract calculus of moral agency, of who is disadvantaged and who isn't, but because one is certainly going to murder more people if he can, while the other is a lot less likely to.
We've known for centuries, if not millennia, that it's the same small percentage of people doing all the robbing, raping, and murdering, over and over and over again.
And we've known for centuries that if you physically remove them from society, that's 100% effective in stopping them from doing it again.
The only hurdle is philosophical. Call it a "justice" system, and you have to argue endlessly about morality and redemption, and then some leftie thug-hugger weaponizes your own Christianity against you.
Call it public safety, and you confine the argument to likelihood of reoffense. Then you are in the realm of statistics. Which you can compute.
It all starts with naming things correctly, according to their actual nature.
@xminosi At issue is the greatest question of our day: can humans outsource knowledge storage to a machine yet retain the capability of problem-solving/creativity?
While the process is workable, it’s massively inefficient and smacks of hubris.
I believe It will hobble the west.
There is a word for this in German: einstellung. It roughly means that you get a false sense of confidence when you see an answer to a problem and you can track the process of getting to the answer, but without doing the actual work of applying what you’re learning in class to a problem.
I used to teach that word to my high school AP economics students since even back in 2013 the game was very clear. One-to-one device policies had made it very easy for students to cheat since their homework assignments because they came from the online textbook loaded on their school-issued laptops. The market being the market, and the slowness of school districts to do anything being legendary, all the textbook answers were already available online to the enterprising honors student.
Teaching them about einstellung was their warning. They were all smart enough to reverse engineer an answer when given a textbook. All I did was explain to them that without actually doing the work, they weren’t building the neural framework that would allow them to recall and do the work on their own on quizzes and tests.
They took the test in my class with no book, no device, and no phones. That first test taught most of them that einstellung was real and they weren’t immune, no matter how smart everyone had been telling them they were since kindergarten. They had to do the work or they wouldn’t be able to do it under testing conditions which mimicked the conditions under which they’d take the AP exam.
The ones who took that lesson to heart ended up with Bs and As in my class. The ones who continued to take the shortcut ended up with Ds and Fs.
The argument in K12 now is that AI makes einstellung irrelevant because a kid will always have access to AI. But I think the evidence is beginning to mount that all this has done is render large swaths of the American population illiterate and innumerate and dependent on their phones for judgments on pretty much everything because they have no knowledge against which to judge.
Remember when Windows added a new “Notepad” app with CoPilot and forced the good old notepad.exe to open the new app instead of itself even if you don’t want it?
Well, a new feature just dropped.
Remix 3 still has some rough edges, but wow is it exciting to use. It feels like driving a sports car instead of a sedan. Using plain JavaScript variables is so freeing.
The biggest impediment to using Remix right now is no easy options for common libraries. I think I have TanStack Query running, but it isn’t a simple library include. I’m manning my own helpers as I go, but it will be nice when I can throw them away and use the prebuilt adapters.
I’m really enjoying Remix 3. It is so good! I recently implemented a project in Remix 3 and React. I was surprised how much smaller the Remix runtime was. It is as much smaller than React as React is than Angular. The remix-run library is ridiculously small. So good. 24k!! Crazy.
@remix_run maybe it is a question that has already been answered, but hire do we use refs in Remix v3? I tried createRef since it is based on Preact, but that didn’t work. :)
@kentcdodds you have more reach than me. But I think something like this would be so helpful for people:
let self = this;
function setState(updates: () => void) {
updates();
self.update();
}
let count = 0;
:
:
setState(() => count++);
@jlongster@kentcdodds@sebastienlorber@owenconti I agree. There will be adapters for all of the reactive state systems. I like not needing one. It is a breath of fresh air. But they have their used. I think each of the systems aspired to being unobtrusive (excepting XState and Redux). Not needing one is the ultimate simplicity.
This is very back of the napkin. I have way too many browser tabs open for these numbers to be reasonable. But I didn’t change the environment between tests. All this client-side only with a Vite template.
Though performance was basically equal between Remix v3 and React, and the imperative re-renders felt like adding mental weight, managing state with plain-ole-JavaScript variables was amazing. So much easier than useState and useRef! I loved my nested array test compare to React.
Mileage may vary - when I wrapped my updates in an apply function (throwback to Knockout days) React edged out faster by about 2%. Function calls are not free, but maybe the abstraction is useful.
After my question to @kentcdodds I decided to just whip up a small perf test to satisfy my curiosity about batches updates in Remix v3 and perf comparison to React. I found Remix edged out maybe 1% faster over 1000 renders triggered by a 5ms interval. Negligible, but consistent.