If only voters would get this worked up about corruption.

People are also ignoring the benefits. Data centers don't bring a lot of jobs after construction is complete, but they can generate a lot of tax revenues for counties and cities. Huge amounts of money that can be a real game-changer for the local budgets, and could be very beneficial for the locals. By itself that's not a reason to accept them, but it should be part of the overall calculation, and the net cost/benefit of the projects should be reasonably and fairly evaluated, and negotiated.

But this isn't what's happening.

The thumb wheels on the steering wheel do that. You do have to use the touch screen to put it in "mirror adjust" mode, but really that's something you only do once. Set up your driver profile, set your mirrors, then the car just always changes them to your settings.

I find the auto wiper works pretty well... it is important to keep the windshield fairly clean, but I like to do that anyway.

Yeah, the Model 3/Y handles are weird. The S/X handles are pretty nice. In theory the fact that they're motorized could make them a maintenance problem. In practice, little electric motors are extremely reliable. There was a problem with the wiring harness getting fatigued and breaking about 10 years ago, but that's been fixed.

Those are also on the thumb wheels on the steering wheel.

Before I came up with my current system I spent a lot of time looking into actuators for windows. They exist but they're expensive, often bulky and look like a lot of work to install and maintain. Then it occurred to me that I already have a whole-house fan (HVAC system)... if I could just get it to draw from outside rather than from the air returns. That part woks great and I think it does give me a pretty good single point of control. The only problem is that the house is pretty tight so I also needed a way to create some negative pressure, hence the laundry room to attic fan. I think that's still my airflow bottleneck.

As for whether I've crunched the numbers.... no, I haven't. And since I installed solar last year it maybe doesn't even matter any more? Throughout most of the summer I'm actually giving away nearly a MWh per month of electricity even after running house, A/C, car, steam shower, etc. So it's really more of a game to me than a serious attempt to save money. That said, I do watch my energy consumption closely (game, right?) and running the fans consumes maybe 400W. Overnight that's about 3.2 kWh. Meanwhile, my AC draws 4 kW, so if my AC kicks on for 45 minutes it consumes as much as the passive system does all night. And for that, the passive system cools the house by ~10F degrees while the AC will cool it by about 2F, which would indicate 5X efficiency improvement. But even that doesn't fully capture it, because the AC will cool the *air* by 2F while the passive system will cool the house and also much of the thermal mass, so it's actually moving quite a bit more heat out than the thermostat temperatures would indicate.

So, yeah, an order of magnitude more efficient seems about right. When outside air cools down at night!

I've also been thinking about building something for my shop. It's an uninsulated pole barn (30' wide x 40' long x 25' high) which gets quite hot in the summer and quite cold in the winter. I think some automated fans to exploit day/night temperature variations could probably move the average interior temperature several degrees closer to comfortable, year round. Even better if I insulate it.

I think that's fair, although it feels like the tone has gotten harsher.

Definitely. I'm actually already seeing a little of that in late August. Nights are getting down to ~62F, which is managing to cool the house to the high 60s. I'm wondering if I could get a higher-volume fan for the laundry room to increase the air flow... I'm sure that by the end of September we'll have nights in the 50s but days will still often reach the mid-80s. That's when another benefit of having it automated kicks in. If I were just opening a bunch of windows before going to bed, there would be a period of time when the house got too cold at night. With the automated system, when the house gets to 62F it turns the fans off.

Of course, as we get into winter, I pop the laundry room fan out and fill the gap with an insulated pillow of the type they sell for filling swamp cooler vents.

Speaking of swamp coolers... those have historically been fairly effective in Utah, thanks to the dry air. My first house had just a swamp cooler and it was honestly fine. I've actually thought that it might make sense to have all three: passive cooling, swamp cooler and central air. With a smart system deciding which to use when. You don't want to use the swamp cooler a lot on days when the A/C might need to run, because it raises interior humidity, which the A/C then spends a lot of effort extracting. Maybe... have to consider the cost of buying and installing the swamp cooler.

It's possible. My prediction is that they're not going to find any evidence of collusion, that all of the DRAM manufacturers simultaneously shifted production to AI chips in response to clear market demand, not because of any agreement.

Yes, there was collusive price fixing 25 years ago, but the companies got slapped pretty hard for it -- not just cash fines but prison sentences for executives. I think that prior experience makes it less likely that they're colluding now, not more likely. Once burned twice shy.

The real issue is that the packaging process used to create vaccines in quantity is a process where you mix a tank of mRNA with a tank of other stuff and get a tank of lipid nanoparticles. Getting different mRNA in each dose means you only get one patient's vaccine per time you start and stop the machine. There's also a step where one copy of mRNA is used to make a macroscopic quantity of copies in a bioreactor that needs to be flushed in between different sequences, but can just run when making enough of the same thing for a large population. It's likely that the cost for each different sequence will come down if it's worth designing and building equipment that switches every dose, but there's an intermediate time between proving that the treatment is useful and being able to provide it in a cost-effective way.

One correction: It didn't have anything to do with discovery. Discovery was complete when the judge put the trial on hold to allow the deal to close. What happened was that Musk realized that proceeding with the trial was dumb because he was going to lose and be forced to complete the transaction regardless, and possibly on worse terms than he could get by just doing it. Avoiding the public spectacle of the trial was also a consideration, I'm sure, but not discovery... 10 days before trial discovery is already done.

That's too strong. Appeasement does work sometimes... when it works we usually call it "compromise" or something similar, though. There's a book "Appeasement in International Politics" that examines a large number of cases where it worked and where it didn't. The author lays out a number of criteria that influence whether it works.:

1. When the opponents aims are limited and grievance-based.
2. When the concession addresses the grievance
3. When the conceding party is strong, so the concession is a choice
4. When concessions are reciprocal and limited.

That said, Trump's aims are never limited. That has become clear. It wasn't necessarily obvious that was the case early on, though.

Maybe? There are some pretty good arguments that it did pan out in 1939. It didn't prevent the war, but it gave the British time to build out radar coverage and ramp up production of hurricanes and spitfires. Britain may actually have been better off going to war in 1940 than they would have in 1939, so appeasement may well have provided a useful delay.

It's both. Advertisers very much want to advertise things you want to buy, to either trigger a purchase that you were holding off on and might forget about or (more often) to direct a purchase to a specific product or brand, rather than a competitor's. Advertisers also want to push purchase ideas to you, to tell you about things you might want to buy that you didn't know about, or often just to build brand awareness or brand vibes so that if at sometime in the future you're interested in something, you'll pick theirs.

Advertisers really don't want to pay money to show you ads for stuff that you simply won't ever buy. That's a waste. And the ad algorithms often screw it up and deliver wasted ads. That lowers the value to advertisers, making them willing to pay less per ad, which lowers the value to ad distributors. They'd very much like their ads to be better.

Larry Page used to say that the project of building good advertising wasn't done until every ad resulted in a purchase. That would actually be a kind of ad nirvana for advertisers, who would be willing to pay a lot for a guaranteed sale, ad networks, who could charge a lot for an ad, and even consumers, who would never see any ads except for things they wanted to buy (even if they didn't know they wanted the thing until they saw the ad -- indeed those would be the most valuable ads, since they would convert a guaranteed non-purchase to a guaranteed purchase).

(Larry's comment ignored brand awareness advertising, obviously).

The status quo is obviously very far from that situation.

Shutter? You think they're going to spend $500B on fabs and then just shut them down? Their boards would be eviscerated. As for "repurpose", sure, they'll manufacture whatever will generate the most profit, but with a fab capacity glut that will definitely include RAM.

That wasn't my experience, though perhaps it was just my workloads.

I didn't use OS/2 until Warp, some years later, and only because I went to work for IBM.

It's clearly all the superfans who hear that there's a new album and immediately drive into their local record store. (Rather than doing the sensible, calm thing and entering the store on foot.)

No, supply will increase. What you're not considering is that the manufacturers are building fab capacity like crazy right now. When all of that capacity hits (in 3-4 years), prices will crater. We've been through this whole cycle many times.

I remember the first big one, driven by the launch of Windows 95. MS-DOS could use memory above the 640K mark only with significant pain and complexity, but Win95 could address it directly and really liked having an insane amount of RAM... like 16 MB!!. Freakin' hog. My NeXTstation only had 8 MB and it blew those dumb PCs out of the water... anyway, I digress. RAM prices spiked, but within about 18 months (fabs were simpler back then) they went through the floor.

Not so much the raw materials as the fab space/time, but yes.

We have pretty extensive history of how this goes; there have been a half-dozen DRAM crunches in the last ~40 years. What always happens is that prices explode, manufacturers build out massive new capacity (which takes years, and as fabs have gotten more complex it takes more years), then all the new capacity goes into production at about the same time and RAM prices crater.

And we're definitely seeing the same processes at work this time. Between Samsung, SK Hynix, TSMC and Micron there's almost a trillion dollars of new fab construction going on right now. When all of that comes on line, it's going to provide a massive boost to output... and RAM prices will almost certainly drop precipitously, losing all of the recent gains plus whatever normal decline we'd have expected, plus some more.

When? Probably around 2030. 2029 if we're lucky, though we might see a little softening by 2028. Definitely not this year and almost certainly not next year.

It could happen faster if the AI bubble bursts before then. I wouldn't hold my breath.