creato

It was a fluke because the US was unscathed by a war that destroyed much of the industrial and productive capacity in the rest of the world, at the same time vast strides in technology were being made. The US worker had a worldwide monopoly on labor and innovation for 30 years.

It's especially silly in this case because Safari extensions have always been equivalent to MV3 functionality.

Something like this could maybe happen to one switch, it's unlikely but possible. But two independent switches at the same time?

> So where is the obviously-named "KickStop" table saw?

It's called a riving knife?

At least 50% of time on every software team I've ever been on was spent on maintenance and fixing bugs.

You can expense such time as opex, but it has to be justified, and that's often difficult to do. Did you fix a bug by refactoring some code to avoid the problem? Is that capex or opex? Can you convince the IRS of such?

The old (and now new) rules eliminated this accounting game and uncertainty.

> If you algorithmically denoise an image and then subtract the result from the original to get only the grain, you can easily see “ghost” patterns in the grain that reflect the original image. This represents lost image data that cannot be recovered by adding synthetic grain.

The synthesized grain is dependent on the brightness. If you were to just replace the frames with the synthesized grain described in the OP post instead of adding it, you would see something very similar.

> They got 17 year sentences and Biden let them out nearly immediately.

This is grossly incorrect. Conahan was incarcerated in 2011, and sent to home confinement in 2020 (panedemic), and the remainder of his sentence was commuted in 2024. He served 13 years of a 17.5 year sentence.

Ciavarella was not pardoned and is scheduled to be released in 2034.

> Length of 6.2 m times 5 tonnes per cubic meter gives a sectional density of 31 tonnes per square meter, which is about 15 meters of dirt. So Newton's impact depth approximation would predict a penetration depth one fourth of the reported 60-meter depth.

This seems to assume that the weapon would penetrate until it displaced an equal amount of dirt by mass, which seems like nonsense. Why would that be the case?

They had pinpoint accurate information about a lot of senior leaders, that seems a lot harder to know than a stationary facility's location and layout.

When you design software for N customers, where N is very small, and you expect to hold each customers' hand individually, the software is basically guaranteed to be hot garbage that doesn't generalize or actually work except in exactly the use cases you supported (there are exceptions to this, but it requires having exceptional software engineers and leaders that care about doing things correctly and not just closing the next ticket, and in my experience, they are extremely rare).

If you design software for N00000 customers, it can't be shit, because you can't hold the hands of that many people, it's just not possible. By intending to design software for a wide variety of users, it forces you to make your software not suck, or you'll drown in support requests that you cannot possibly handle.

> publicly lied about the vaccine’s efficacy.

You keep saying this... what was the lie?

Even if true, a small fraction of engineering time on a project is actually developing that asset. The rest is maintenance and support. The tax code does allow for this distinction, but only if you track hours associated with each kind of work, which basically no one does. And even if they tried, it's difficult because that line is blurry. Tasks are rarely 100% one or the other. Ever fixed a bug by refactoring to make something better? Which kind of engineering is that? Can you justify that to the IRS accountant auditing you?

I thought this was going to be a metaphor for debt, not literally bar tabs

This is why the standard for something to be considered improper behavior is (or used to be) the appearance of a conflict of interest.

People stopped giving a shit about anything. This is just one of dozens of things that would have been totally unacceptable a few years ago.

The point is, this amount of error is huge for fp32, but may be expected for fp16. But then why compare to fp32 performance baselines? An algorithm that gives you the accuracy of fp16 should be compared to an fp16 baseline, and this may not be (it probably is not) a speedup at all, it's likely much slower.

I also wonder how much of ChatGPT's usage is to basically cheat on homework. Basically 100% of the users of ChatGPT I know mostly use it to do their homework for them.

Maybe this will turn out to be a valuable user segment, but I'm not sure.

Some of these scam victims are duped into trying to help someone they trust with a financial problem. In that case, the victims are being the opposite of greedy. The only character "flaw" I can think of in these cases is being naive/too trusting.

I don't know how common the "savior" victims are vs. the "get rich quick" victims, but they definitely exist, at least according to various mini-documentaries on pig butchering I've watched. Maybe these victims were modifying the story to seem more sympathetic, I can't say.

> we'll reach a crunch point that puts governments under immense pressure to increase taxes (probably on megacorps & the wealthy) more in order to fund welfare.

I don't see how even that helps. What you need to fund basic welfare is a large supply of basic goods (shelter, food, medical care). The basic problem with retirement of a lager generation than subsequent generations is a lack of labor/productivity. Those basic goods don't really respond well to just throwing money from tech companies/rich people at the problem, you need people to work to provide those goods and services.

The only way taxes help is if they discourage "unproductive" economic activity (SWEs and financial planners) so much that those people start working as nurses and construction workers instead.

Maybe so, but I don't think this point changes the overall situation as described in the video.

You're not denying freedom to just those ~1000 companies, you're also denying freedoms from everyone they might "acquire". And if your definition of "acquisition" is 4 dudes with a source control repo and no market/sales, then you've denied a lot of freedom to a lot of people, not just some big corporates we all love to hate.

I personally know of probably a dozen cases of a small team of <10 people getting hired into various companies that all would be scrutinized under this reasoning, and yeah, I think that's totally insane.

I can't understand how this is even a question.

At the time, google was a $30B company, 10x smaller than Microsoft. Where 2 Technologies was a 1 year old company with 4 people, with 0 documented sales or market share that I can find. The only evidence I can find of the company actually doing anything is a picture of a whiteboard. I'm not dinging the company, just saying that it was very, very early in its life.

Suggesting that competition regulators should have blocked this is basically suggesting that the government has veto power over almost any hiring decision by any of the top ~1000 companies by market cap.

One of the acquisitions you are talking about was 4 people that started a company and was "acquired" by google less than a year later. The idea that something like this should be blocked by competition regulators is frankly, totally insane.

Aside from that, in 2004, Google was a ~30B company, 1/3rd the size of e.g. UPS. For comparison, Microsoft was ~10x bigger.

The rewriting of history in this thread is just crazy. I don't disagree that google today is huge and some regulation is necessary, but what you are proposing should have been done in 2004 is off the charts ridiculous. Not even totalitarian dictatorships exert that much control over their populations.

It's disappointing to see the historical revisionism in these threads. Say what you will about google now, but the idea that they just bought their way into X industry doesn't seem right. Until maybe 10 years ago, tech people almost universally loved Google's offerings and adopted them eagerly because they were good. I remember the mad scramble on various forums for a gmail invite. I remember when google maps came about, it was a revelation compared to mapquest and so on. You could scroll the map instead of clicking buttons to jump half a screen at a time! For years 0-5 at least, chrome was almost universally loved by tech people. Process isolation, speed, lack of toolbar shitware, etc.

At every company I've been at, half the dependencies came from big tech, and more than half of those were built and maintained by google. bazel, kubernetes, test frameworks, tensorflow, etc. these are just the big ones. There are a lot of smaller libraries from google that we've used too, and more still that aren't owned by google but they invest a lot of engineering time into.

I don't know what the right answer is to the google of today, but the cavalier assumption that google has simply leveraged a monopoly in search to build everything else it has doesn't add up to me.

They did exactly that for decades.

"Not to mention" doesn't make sense here. Everything else in this thread is talking about reducing the demand for trucking labor. Your post is talking about reducing the supply.

This seems like good news if it means the AI safety nannies are losing influence.

As of 2023 Apple had done very little of this to iMessage, a far simpler app than a browser: https://news.ycombinator.com/item?id=37425007

Every browser I've used in the last 20 years: IE, Firefox, Chrome, Safari, all free. The browser market has been full of free competitors since before Google even existed.

I don't think Google uses data from Chrome this way. I don't think Google actually wants to associate data with people so aggressively. It's obvious when that happens and it feels gross (this is why I stopped using Facebook 10+ years ago, I've never had an equivalent gross-out moment with Google).

I'm guessing the reason google doesn't use third party cookies is because they get higher quality data from people being signed in to Google services, and that is independent of whether they are using Chrome or not.

I'd suggest simply adding `assert(-M_PI/2 <= x && x <= M_PI/2)` to your function. It won't slow down the code in optimized builds, and makes it obvious that it isn't designed to work outside that range even if people copy/paste the code without reading it or any comments.

Also, it would be good to have even in a "production" use of a function like this, in case something outside that range reaches it by accident.