sfilmeyer

> People die in hospital waiting rooms. Why risk it when you know you have everything you need right on hand?

Because people also die in incomplete airplane evacuations. In the Aeroflot Flight 1492 crash, 41 of 78 occupants died while folks were seen evacuating with their bags. Some of them would have died no matter what, but somebody slowing down an evacuation from a serious airplane crash for even just seconds to quickly grab their insulin kit out from under the seat in front of them (let alone to film for social media or whatever else) could cost someone else their life.

The situation might look "fairly under control", but the plane is upside down, leaking flammable jetfuel, and surrounded by firefighting foam that no one wants to spend more time around than they have to. Leaving behind medical supplies in a crash has a real impact on a diabetic or similar, but any slowdown to the evacuation also has a real impact on all of the other passengers.

>or paying netflix to license their technology

Does Netflix license their technology to anyone? I know of examples like BAMTech, although I don't even know if they still take on outside clients or just do Disney now. I get that their might be good options to license and that fewer companies should build crappy in-house products, but is Netflix one of them?

From Netflix's perspective, it's not clear to me that the payment for licensing technology to e.g. NBC is worth it, versus hoping that they end up with an inferior product, especially when they're competing with each other for customers and licensed content.

>The ratio might be quite different for LLM, but still any SUCCESSFUL model will have inference > training at some point in time.

I think you're making assumptions here that don't necessarily have to be universally true for all successful models. Even without getting into particularly pathological cases, some models can be successful and profitable while only having a few customers. If you build a model that is very valuable to investment banks, to professional basketball teams, or some other much more limited group than consumers writ large, you might get paid handsomely for a limited amount of inference but still spend a lot on training.

But even for your 5 people, a gigabit is plenty to all stream content at the same time, right? How often are all 5 people trying to download a 5 gigabyte file at the same time such that they collectively really benefit from more than a gigabit?

>I never get them when I am not expecting a package.

For what it's worth, I get scam messages claiming to be about usps, dhl, et cetera even when I'm not expecting a package. Recently, I have a couple claiming to be about a package failing to clear customs (but if I just pay a quick fee...).

>That seems to have been the real problem

The real problem here is starting a war. In a more peaceful context, even being hundreds of miles off from your flight plan wouldn't have resulted in this plane getting shot down.

I get what you're saying, though. It sounds unlikely that shooting this plane down was done because anyone specifically wanted to shoot down this particular civilian plane.

>3. better than most corporate stuff is a low bar

I think it's a very relevant bar, though. The top level commenter made points about "a business of just about any size", which seems pretty exactly aligned with "most corporate stuff".

Training doesn't have to be nefarious. Google ran GOOG-411 in large part to collect speech data, but folks using the service still benefited.

I've known people who valued good handwriting but weren't particularly capable of writing neatly, and others who didn't particularly value it but were capable of staying neat while writing quickly. I think you might be overestimating how much of it is tied to what people value.

Democrats might not currently have Trump's level of protectionism and tariffs, but I think they're still better for unions. For example, the NLRB gets less effective as republicans put more anti-union folks on the board, let alone as the Trump appointees on the supreme court et al. reduce the ability of agencies to regulate.

>In 1928, three years before his death, everyone said of Michelson, "He measured light," and today he is one of the few Nobel Prize winners whom nearly all educated people can name and give the reason for the award, although Michelson’s award actually was based on a wide spectrum of experiments.

I found this line (particularly "nearly all educated people can name and give the reason for the award") pretty amusing. Prior to this article, I'd heard of Michaelson, knew he was part of the Michelson and Morley experiment, and remembered a couple of details of the experiment design. But I also have a bachelors degree in physics, and I still couldn't have confidently told you whether he won a Nobel prize. I'm sure the intervening 50 years have dulled his fame a bit, but I've got to wonder how true it was when written.

I initially misread this as you proposing using a car battery rather than a 9 volt battery, which sounds like a much less fun party trick.

>I would love to see if that would make an impact on the odds.

It won't make an impact on the odds of your tickets coming up as winners, unless they have a bug in their simulation. In the real world the probability of single versus multiple jackpot winners might vary with number choices, but they've already said they're assuming a single jackpot winner.

>They’d definitely prefer hosting on say Neocities, but that’s more because Neocities has cats and is overall cuter in presentation.

At the risk of misunderstanding what you're saying, who is "They" in this?

>I do wish the cases would actually involve harm to people who are “gold star”

I'm really struggling to understand what you're saying here. What do you mean by "people who are 'gold star'"? Which cases are you hoping will harm those people?

Yes, there was lots of sharding across hospitals (although it wasn't executed perfectly). Wikipedia has some good info. About 60 people died and almost 900 were injured, over 400 of whom had gunshot or shrapnel wounds.

https://en.wikipedia.org/wiki/2017_Las_Vegas_shooting#Victim...

> Approximately 867 people were injured, at least 413 of them with gunshot wounds or shrapnel injuries. In the aftermath, many victims were transported to area hospitals, which included University Medical Center of Southern Nevada, Sunrise Hospital & Medical Center, and at least one of the six hospitals of Valley Health System. Sunrise Hospital treated the largest portion of the wounded: 199 patients, 150 of whom arrived within about 40 minutes.[80] University Medical Center treated 104 patients. Additionally, six victims sought medical treatment in Southern California; UC Irvine Medical Center treated four and Loma Linda University Medical Center treated two. Many victims of the shooting required blood transfusions, which totaled 499 components in the first 24 hours of treatment. This blood was rapidly replaced by available blood from local and national blood banks.

>University Medical Center, the Level I trauma center in Las Vegas, was difficult to access for the more than 50 percent of patients transported by private vehicles because Interstate 15, the most direct route from the shooting location, was closed to the public. Also, an erroneous emergency services announcement made one hour after the shooting reported UMC had reached capacity and was on diversion. This confusion persisted for several hours and led to most patients being transported to Sunrise, a Level II trauma center.

I get the point you're making, but I can't resist pointing out that if you start with 0 and increase regularly by 20%, you'll still always be at 0.

What's the bar, though? If out of a million people reading a joke, 80% find it funny, 20% find it meh, and one solitary person needs the joke explained to them, I think it's still fair to call it a funny joke. There are multiple comments in this thread missing the satire so it's obvious the percentage is a bit higher than that, but I'd wager the majority of people didn't need the joke explained to them.

This article that I just came across does a nice job at explaining some of the nuances of solar noon https://www.timeanddate.com/astronomy/solar-noon.html .

The time of year does matter a bit, but the effect is smaller than the effect of timezones. Their equation of time article includes a nice graph of the variations in time of solar noon at Greenwich. https://www.timeanddate.com/astronomy/equation-of-time.html

>By definition your predictions will be more accurate the smaller the window is.

Small nit, and this doesn't detract from your points. I don't think this is universally true by definition, even if it is almost always true. You could come up with some rare conditions where your traffic at t+5 minutes is actually easier to predict than at t+20 seconds. Of course, even in that case you're better off (ceteris paribus) being able to spin things up in 20 seconds.

>Compare Walmart retail to Amazon retail.

>AWS and its 30 percent margins are the only thing propping amazon up vs Walmart.

In 2023, AWS had 27% margins while North American retail had 4% margins for Amazon, but that still meant $15 billion in income from North American retail versus $25 billion in income from AWS. By comparison, Walmart US is operating at a 5% margin with about $22 billion in income. Both companies have worse margins internationally.

I agree those numbers make AWS look a bit more attractive, but it doesn't mean they're "the only thing propping amazon up vs Walmart". Walmart is running a bit larger and at slightly better margins, but the difference isn't as stark as you're making it out to be.

All numbers were hastily pulled from recent earnings reports, so apologies if I made any sloppy mistakes.

> postnuptial agreement with conditions exactly like if you divorced today

There's no such thing as a postnuptial agreement fully legally equivalent to divorce, though. A trivial example is your legal ability to marry someone else, but there are plenty of other differences.

I think the distinction is still important, in that the time lag is large enough to have changed the current dynamics, even if they still match on longer timescales. Raising salaries a bit could stem the outflow of construction workers, and raising salaries sufficiently could bring folks like the prior commenter back into the industry. But as more time passes this gets less and less effective, because there's no one left in the industry to retain and no one with recent enough knowledge to pull back in.

I don't think that dynamic is the same in construction and in semiconductor manufacturing. Certain parts of the chip-building supply chain just plain don't exist in the USA, and so no one in the USA currently has those skills. It's not like there are lots of folks with the skills in the USA but they're all leaving the industry for higher paying jobs.

I question whether it's actually true that they "aren't going to do it any safer than you would". I'd expect the typical person I hire to have a lot more experience than me which would give them a chance to be safer, to the point I'd expect it to overcome even the fact that they're probably not taking their time as much as I would.

Separately, being liable for something happening to someone else and that same thing happening to me are not equivalent. Macabre as it is, I'd rather pay out (or have my insurance pay) the lawsuit for a professional hired for a particular task ending up grievously injured than end up grievously injured myself. In a lot of cases, I'd doubt you'd even be financially liable. From a moral standpoint there's still some culpability, but in the case of garage springs I think your obligation is more to stress to the professional that you're happy for them to take their time and you're willing to pay accordingly than it is to subject yourself to the risk instead. From a societal standpoint, a few repairpeople building up expertise at a particular task is safer than everyone doing it individually.

This article is about Adobe getting out of this business. I'd think if they were "rushing to implement" they'd keep the existing product around and either integrate things into it or wait to kill it until they have a replacement. Also, while it's not perfect, there tends to be a fair amount of firewalling around that sort of due diligence to keep an acquirer from just ripping everything off with inside information gained in the process.

Yeah, we use an Aura hooked up to Google Photos rather than iCloud, but I think it's similar from the Aura UX perspective. It's super easy to use, and the photos just show up in the frame rotation whenever we add a photo to the corresponding album in Google Photos (which we already use for backing up and organizing photos).

They did share an idea.

> rather than hitting them them with multiple antibiotics and thus raising the evolutionary hurdle to resistance?

It was surrounded by some rudeness and is probably a poor idea (but I'm no expert), but it was an idea.

>Techies almost always have a laptop, phone, and home desktop up at once

What definition are you using here? I like tech and worked as a software engineer (not working now), and I haven't owned a desktop in over a decade. If we're talking "almost always", I think these days I have 0 phones/laptops up most of the time followed by 1 in pretty close second place, and even having 2 (let alone 3 or more) is uncommon for me.

I remember being shocked at the time that he had the foresight to use Tor but not to use literally any wifi network other than the campus wifi. That being said, there are a whole list of things he'd have to do to keep anonymous and it only takes one slip to identify someone.