ijpsud

> The problem is when a company or two dominates a single market.

It would also be bad if one company dominated multiple markets. (just a nit pick on your phrasing)

> they make it even harder for Europe to ever catch up

You've stated this as a fact, but it's not self-evident. There are obviously cases where some degree of "protectionism" makes sense. For example, you probably wouldn't want to let yourself get into a situation where another country supplies 100% of your food, if you could help it. As an extreme case, if you can produce all your food locally and it only makes your food 1% more expensive (than the tariff-less case), that's very likely a great trade-off.

Some degree of protectionism can make sense in certain circumstances. I'm not claiming it necessarily makes sense in this case (could very well be a terrible idea), but I think you'd need some evidence or reasoning behind the claim that it's definitely a bad thing for Europe.

Correct me if I'm wrong, but couldn't they just save your secret data to localStorage and then send it off when you load another page from the same origin (a page without those headers)? Or does the CSP of one page somehow apply to the whole origin? I'm not up to date on this stuff.

+1 We definitely need to work on this from a "grass-roots" education-system level, but it's also a bit of a "chicken and egg" situation, because we also need to make the tech industry appealing to young black students who are in the process of forming ideas about the sort of career they want.

The examples[0][1] of humiliation from the article, and the many examples from black folks in this thread don't paint a picture of an industry that is particularly appealing to a student thinking about their career options. That's if they've even considered the industry in the first place - the dearth of visible/powerful black faces in tech is very likely causing a "can't be what you can't see" effect in the minds of many young black kids.

The fact that the industry is currently doing some introspection is great, but real-world, noticeable progress on these issues will probably take decades.

[0] > That message hit particularly hard in 2015, when Givens and his co-founder Eric Williams, who is also Black, brought some of their employees to Churchill Downs Racetrack, home of the Kentucky Derby. They were at a VIP suite to celebrate the first big installation of one of their bartending machines. Guest after guest walked past the pair of executives to greet their White employee, offer their congratulations and ask how he got the idea for the product. The mood dimmed once the well-wishers realized their gaffes. “It went really quickly from ‘I have a drink in my hand, and I want to network’ to ‘I just want to get out of here,’” Givens recalls.

[1] > His first two fundraising rounds were challenging, according to Hayes and his apparent doppelgänger, Messick. The recurring confusion about who was CEO, and the embarrassment and apologies that followed, certainly didn’t help. “I usually remember the room and the look on the face,” Hayes says. “The energy would drop dramatically.”

> I'm not sure what the answer is [...]

Should have left it there, mate.

Many people who are members of a minority group feel pressure to do everything "perfectly" because people will naturally use them as a reference for the whole minority group. That's just how human brains are wired - we use patterns that we observe to predict the future. If we don't have much data, we form crude stereotypes.

It takes active effort and learning from people like you and I to help overcome those biases so that members of minority groups aren't exposed to this pressure and can feel safe in making mistakes (or at least, as safe as members of the majority feel).

In general, if you don't know what you're talking about (at least you were honest about it), it's best to do some more learning rather than to add noise to the thread. I'm not claiming that I know much here, but I know enough to know that a solution like "just be you" is not going to be helpful here.

Does iOS support PWAs yet? I'd have thought that PWAs would have all the features required for an email application at this point.

Apparently 20% of Apple's revenue comes from their marketplaces (music, apps, etc), but in terms of profit it's probably significantly higher since the costs ought to be much lower than their hardware.

So it seems like there's a decent chance that at least third of their profit could get wiped out by PWAs in the long run. You'd think they'd be treating developers a little better.

Why is login required to view simulations? If I want to share my simulations with my followers, I don't want everyone to have to sign up for an account. Also, the "play" button in the editor UI should probably look more like a play button.

I'm excited by what you're building here!

Agreed. And in regard to GP's comment:

> kind of arrogant "make the world a better place" attitude

It's not "arrogant" to want to make the world a better place. If GP thinks they're lying, they's better described as "deceptive". If GP thinks they're naive, they should say that, and ideally provide some constructive feedback. This kind of knee-jerk comment (and the subsequent knee-jerk upvotes) is my least-favorite thing about HN.

I was playing with tf.js last year and there were some core components still being converted to WASM, which gave me some big performance boosts once I switched. I had some problems with WebGL-related bugs too. It seems that they're currently in the process of writing WebGPU-based back-ends. I'm hoping that when WebGPU is on-by-default in most browsers, and there's been a year or two to sort out all the major teething problems, we'll see some wide-spread uptake.

Internet speeds are still increasing exponentially, so hopefully the model size problems will become less of an issue - perhaps aided by some CDN-served (and thus cached across domains) "base" models that are fine-tuned with some parameters downloaded from the server. I think I could start playing with it seriously if I could get the model sizes under 30mb or so. In a few years (with increasing internet speeds) that might be 50mb. I think huggingface's distilled GPT-2 model is a couple of hundred megabytes, for reference, so we're certainly not going to be doing anything revolutionary in the browser, but I have a bunch of neat little ideas that I think would be useful.

For bringing the "big" stuff to the web we're probably stuck with APIs, like OpenAI's new GPT-3 offering. I think access to SOTA models on hard problems is going to be almost exclusively via APIs for the foreseeable future.