What you're missing is that some of the mapping services have maps that are horribly wrong, and also that people store coordinates badly.

Case in point, I live in a mobile home park that's almost big enough to be its own zip code. If your entry system lets me put in the space number *and* uses Google Maps as the back end, you can find my home that way. But as far as I know, no other mapping company has the per-unit data that would be required to find my house by its house number. That's simply a level of data gathering that nobody else even attempts as far as I can tell. Apple Maps has zero site number data for my neighborhood or any other nearby neighborhood, and they're the next-most-capable competitor in that area.

So if you are using some other service, you'll have to find it by the house number and the street name. Unfortunately, the house numbers are not in order. They added more sites at some point, and did not renumber the old sites, so there are low numbers that jump to higher numbers and then jump back down to lower numbers. So if I have to give a site number plus the street name (e.g. Apple Maps), you might be within a couple of blocks.

But to make matters worse, the road is a private road, which means that it isn't in the list of county-maintained streets. So if your database doesn't have those streets at all, it will end up substituting the nearest roads with the same name, which are in Sacramento, or it will assume that it is a misspelling of some other street in a nearby town. So you could end up 10 miles off or 120 miles off.

So you use Google's API to do geocoding to verify that you can deliver there. Then instead of sending the coordinates, you foolishly hand off the street address to the delivery service that uses some other service, and smoke starts coming out of your machinery.

Or you round the coordinates too much like Uber used to do, and all of the streets are off by a hundred feet, putting my pickup point supposedly on a different street.

You get the idea.

I remember a fair number of people in the '80s getting fooled by Eliza, a collection of heuristics designed to create the illusion that the computer understood what was being types and formulating reasoned responses. Of course, it was doing no such thing.

Modern chatbots do a much better job of it. 'Good' enough that susceptible adults sometimes go over the edge into a full mental health crisis after a month or so interacting with them.

The constant affirmation and un-wavering support makes the chatbots the ultimate yes-man. We have all seen what happens when celebrities and people in power become drunk on their own yes-men.

It's worse than internet echo-chambers. At least those don't tend to let the conversation get that personal and specific. Chatbots will get as personal as you wnt and they are designed to never break engagement (how will the company keep gathering underpants if the chatbot keeps saying no?).

And all of that with adults. Now imagine turning all of that on a young teen that hasn't had time to mature enough to know better. YIPE!

Because people aren't solitary islands whose deaths harm only themselves. Anyone who has lived on Earth for more than a few days has formed relationships with other people, and suicide is the murder of a person those people have a relationship with. Suicide harms everyone who interacts with or depends on that suicider for anything, in much the same way that the murder of that person would.

Yes, that is precisely the problem.

Eventually, yes, but you forgot a bunch of steps:

• Hide the information in a tiny little link at the very bottom of the terms and conditions that nobody reads.
• Cap the speed and/or amount of data in a way that makes the service completely unusable for 99% of real-world customers.
• Impose income limits that make the service unavailable for nearly everyone who wants to apply.
• Require such a complex application process that nobody who needs the program can figure out how to use it.

Then, after about three years, they show the government that nobody wanted $20 Internet service, and ask permission to stop providing it. And the CPUC, being an industry lapdog, quickly agrees to whatever Verizon asks for. And *then* they stop providing it.

Proven by the fact that a few have hacked Windows 11 to run fine on machines that supposedly can't run it.

I wouldn't trust those hacks in production, but they prove the possibility. Of course, I don't trust ANY Windows in production.

It probably helps to use a ridesharing platform that doesn't use multiple map providers. If all your map data is from one source, you don't have these problems. It's when you start to mix multiple map providers that things go horribly and irreparably wrong, because the workarounds for one platform don't work on a different platform. Given that we're talking about Waymo, I assume Google Maps is used for everything, so I wouldn't expect those issues to occur. But no way to know without trying it at your specific location.

And note that this will change over time, but your rails can't change over time. This is the peril of rail for intracity transit.

Rails make a lot of sense in ultra-dense areas (think Manhattan, *maybe* downtown SF, but not any of the rest of the Bay Area, etc.), because the roads can't handle even a fraction of the passenger volume.

Rails also make sense for long-distance travel. If you're traveling for several hours, you probably don't want to drive that, so it is worth the inconvenience of not having a car at the other end. Also, if the trains are fast enough for their average speed to exceed the average speed of a car, then after two or three hours, you've probably broken even with the extra travel time required to get to and from those fixed endpoints.

But for the most part, rail *doesn't* make a lot of sense, because they're too much slower than air travel for long distances, and they're too much slower than cars for short distances. If you really want rail to work, we need 250 MPH (minimum) bullet trains from city to city so that they are competitive with airplanes. And provide ample parking at the termini.

Twice per hour makes a transit system borderline useless unless you are traveling for multiple hours. Your average latency is half of that, so that means on average you will waste 15 minutes waiting for every train. That means to break even, even before factoring in the extra time to get to/from the station, you need to save a whopping 15 minutes by using the train. And if you have even one connection, that makes your median latency thirty minutes. When you're wasting half an hour just waiting for the train to arrive, you're not only uncompetitive with driving; you're starting to be uncompetitive with bicycling.

Successful transit systems run avery 3 to 5 minutes during rush hour, and no more than about once every ten minutes late at night or on weekends. If you can't keep trains running at that frequency, you aren't dense enough to need a transit system, and you probably will regret putting one in. You'll end up repeatedly reducing the frequency to keep the trains at high enough occupancy to be worth doing, and then you'll be confused when ridership drops because nobody wants to wait that long for a train, and eventually you'll end up with a massively subsidized waste of taxpayer dollars like VTA.

Yup. We're well past the point where the resulting economies of scale benefit the consumer. In late-stage capitalism, economies of scale benefit the corporation and only the corporation.

Full or at typical capacity? Lots of bus routes around here average a low single-digit number of passengers for much of the route. Even single-passenger cars compare favorably to that. Assuming a diesel bus at an average of 3 MPG, you need a minimum of 15 passengers on average to break even with driving single-passenger hybrids. And that's not factoring in how much dirtier a gallon of diesel fuel is compared with a gallon of gasoline.

Office wall Plinko with push pins.

I kind of automatically assume that any company that can successfully replace a third of their workforce with AI is not doing anything meaningful. I mean, a company that does nothing but write crappy rehashed product reviews could pull it off, or a company that has no plans to innovate or grow could maybe do it by laying off all the sales/advertising people, but otherwise, it seems like a stretch.

So the question, then, is whether this company is genuinely circling the drain with nothing to lose or just has leadership who have no idea what any of their employees actually do.

Ostensibly, yes, I could. I could even use an ancient Mac Mini with a built-in FireWire port to do that. But at some point, the level of hackiness becomes high enough that you're spending all your time dealing with things not working, and that's almost worse than it not working at all. :-)

M1 Max MacBook Pro. The setup involves a Thunderbolt 3 to Thunderbolt 2 adapter chained to a Thunderbolt 2 to FireWire 800 adapter, which in turn is chained to a FireWire 800 to 400 adapter, which is then connected by a cable to the device. It's dongles all the way down.

Getting Thunderbolt ports isn't the problem; Apple's Thunderbolt-to-FireWire hardware does that, too. Without FireWire device family support in the kernel, FireWire hardware can't be used, period.

You're missing that both a bleed air system AND poor maintenance are required for this problem to manifest.

Presumably the other planes with a bleed air system are getting better maintenance, so haven't been a problem. No idea how the 787's maintenance is, but since it doesn't have a bleed air system, the problem of dangerously contaminated cabin air hasn't manifested.

More specifically, this happens when engine oil or hydraulic fluid leak into the engine while bleed air is being drawn.