"These people" are just anyone on a bad day. People make random mistakes when they do anything enough times.

I've had it happen. I was sitting weird and my foot just missed. You do these motions millions of times without thinking about it, so in that one-in-a-million case where something doesn't line up right, you get a very disorienting "why won't it slow down" feeling, and it's easy to panic. Your muscle memory instinctively pushes the "brake" harder to compensate, but it's actually the accelerator. It takes a moment for your brain to diagnose the situation and correct.

No harm done in my case: average car, open road, healthy and alert so I figured it out within a second. If I was in a Tesla Plaid, in a congested area, tired and distracted, I would have put it through a store window.

It was an eye-opening experience.

It's because of subdomains. Edit your filters: slashdot.org###floating-ad-unit

I learned SAS In the early 80s and used it extensively. At the time, it was easily the best data analysis software available. About 15 years ago I wanted to get a few copies for my consulting team and we were looking at more than $50K / seat. Do you think Chinese users want to pay that sort of money to a US firm?

SAS sued a source compatible competitor (World Programming) out of existence some years back, to destroy competition and maintain their high prices. I had trialled the World Programming solution and it worked very nicely.

These days I use Python and a few other open-source tools. I suspect that Chinese data analysts are mostly doing the same.

It wasn't terrible. I miss getting together with a few like-minded people who were excited to see a show together and discuss it afterward every week.

I did some napkin math. You can fit about 40,000 shirts in a 40-foot container. It costs about $8000 to send that container from a china factory to a USA warehouse. Total about $0.20 per shirt. It's still cheap, but more than the rounding error you suggest.

The possibility that one monopoly may eventually be replaced with another one isn't enough to prevent the abuse. We need actual competition between contemporary companies. I don't want to be stuck taking a bad deal, hoping that maybe a better monopoly will come along next decade.

You CAN do it, but only a few people actually do, and the data is stored privately. It's a new problem when it becomes pervasive, with centralized systems creating an indexed database of everyone, tracking them across interactions with multiple people.

It's basically the end game of Enshittification.

Tannus Airless road bike tires are just awesome. They're heavier, but I don't carry a pump or patch kit anymore and was running fairly heavy gatorskin tires before, so the extra weight nets out about even. Feels like a 110PSI inflation, maybe a little slower ride - wouldn't wanna race them - but not slow enough that I'm ever going back to pneumatic tires. The rubber isn't exactly solid, it's more like a very very dense closed-cell foam, so there's lots of air bubbles to create rebound energy that makes the ride feel more comfortable.

I interpret "unilateral market power" to mean that the suppliers have asymmetric power over the customers. There is little competition once you are locked into an ecosystem.

Sure, you can rent some compute from another company, but you have all these networks and firewalls and databases and data stores you've set up within one company's borders. The competition has similar services, but they're not quite drop in replacements, so you can't just pick up and move to take advantage of better pricing. It can take years to rebuild your infrastructure if you want to escape.

They are also an oligopoly, which limits the need to compete. The customer has very little power to negotiate better terms when there's only one, maybe two, other serious players in the field.

This will have approximately 0 impact on California electricity bills. They already buy power for 4 cents and sell it for 65. If they buy for 3 cents instead, it won't make a real difference. All of the cost is in local distribution and paying off the consequences of boneheaded decisions around local distribution -- aka the wildfires, etc.

That document is about a defective or incorrectly-installed part. Sure, if that's a problem then they should be repaired. It's not clear whether that's related to this crash at all.

You're moving the goalposts from your original suggestion that they shouldn't have a way to shut down the engines in flight at all.

That configuration - a switch placed immediately behind each engine's throttle lever - is almost universal in every jet designed since they eliminated the Flight Engineer position.

A320 (basically every Airbus looks similar). 737-300. 747-400. Embraer is weird: they put the start/stop controls immediately in front of the throttles.

I could go on, but the point is: this isn't some poorly-considered design fluke in the 787. This is how it is done, and for a good reason: there are many situations where it's necessary to shut down the engines for safety reasons. Having these controls readily accessible saves lives.

Maybe you should take your own advice.

To prevent or stop a fire. That needs to be a fast and uncomplicated procedure.

You could add interlock logic: If the aircraft is below $altitude, inhibit the switch. But that ignores use cases like "we're going down and about to crash in a field, let's cut off the engines to reduce the risk of fire."

It's tempting to keep making the logic more complicated: If the aircraft has been airborne for less than $duration and the we're below $altitude, delay shutdown for $x seconds while blaring an alarm, except if this temperature sensor reads high suggesting there's a fire, or excessive fuel flow indicates a leak. This introduces new problems: more bugs due to ever-increasing edge cases; more systemic failures due to a broken sensor; etc. This also means the pilots' mental model is more complicated: you don't want them to flip a switch and then wonder "wait, why didn't that work?", and waste time trying to remember some flowchart from their training.

So the current best-practice solution, used wherever possible, is KISS: Each control does exactly one thing, and it does it in the most immediate and direct way possible. You don't find out about your mistake only after some additional set of conditions are met. It does what the label says, and if you don't like the result, you flip the switch back. It should only be more complicated to prevent a recurring problem.

For the most part, that turns out to be the most safe and reliable design.

I get it. The problem is they didn't put much thought into objective criteria for non-weather categories when they rolled out the system.

I think they should just roll Silver and AMBER together into "missing persons" and it would solve 90% of the problem.