Uh, no. Center of gravity isn't related to how high the driver sits.

Ah, I see, you think they should use low-floor buses. Those are a lot more expensive, have higher maintenance costs (especially the kneeling ones), require flatter terrain (buses don't go offroading, but where I live they can't stay on the pavement all the time and also have to contend with deep snow), and give up seating capacity because the wheel wells and rear engine intrude into the seating are. Their only real advantage is accessibility. City bus systems can't predict where disabled people will be, so all buses have to be accessible.

School districts, on the other hand, do know where the disabled kids are so it's much more cost effective to buy and operate less expensive buses for moving the 95% of the kids who can climb stairs and to operate a separate fleet of smaller buses equipped for accessibility to pick up the disabled kids. So, they save the money on buses and spend it instead on teachers and classrooms.

As a taxpayer and a parent and grandparent, I think that's the right choice.

Okay... but what does that have to do with the center of gravity? And those sticks are just as important for rear-engine buses as for front-engine buses, though they probably don't have to be quite as long.

In Utah I don't think I've seen one of those for at least 20 years, and they were rare before that.

No, you don't. I'm sure I follow that a lot closer than you do. Any claims that the federal judiciary is already captured are just silly. SCOTUS is an issue, but look at all of the rulings against Trump. Even the appellate opinions that the news calls as in favor of Trump are basically all just staying the district injunctions until the merits are decided, and if you read the opinions, not just the headlines, you'll see they're almost always extremely skeptical on the merits.

Honestly, even SCOTUS isn't quite as captured as a lot of people on both sides think. They also seem to be giving Trump his way on the procedural issues, but almost always come down against him on the merits.

The ~900 federal judges in the system are almost universally apolitical, thoughtful and fair. There are exceptions, and Trump is working to get more of them in there, but the judiciary is very far from captured.

Interesting. What stick are you talking about? I'm talking about the thin plastic tube that hinges out from the front passenger corner and blocks kids from being able to walk in front of the bus. It clearly can't have anything to do with center of gravity or stability; you can bend it with one finger (it springs back).

Also, all the buses where I live (in the US) are rear engine and have flat fronts.

You didn't actually read the method, I think. In order for corruption to "build up around it", the entire (or at least most of) the federal judiciary would have to be corrupted. While that's not impossible, if it happens we'd have much bigger problems than pardons. And the fact that the judiciary isn't already irredeemably corrupt is strong counter-evidence, because the benefits of corrupting the courts are far bigger than the benefits of corrupting the pardon system.

I think the reasons the pardon power was given to the president still make sense, so abolishing it is a step too far. Instead, it should just be weakened a little, probably by adding a review by a non-partisan review board plus a limit on the number of pardons a president may issue. The review board shouldn't try to decide if the pardons are "correct", but only whether there is a presidential conflict of interest.

The question of what something like this should look like was interesting to me, so I had a long conversation with Claude to collaboratively design a solution. The high level of the proposal is:

1. If someone files an objection within 30 days, the pardon is reviewed by a 9-member panel of judges selected algorithmically from all sitting federal judges. Unobjected pardons sail through.
2. Reviews must be completed within 60 days or the pardon is automatically upheld.
3. The panel examines the pardon for evidence of presidential impropriety, mainly conflicts of interest. The president can file counterarguments to objections.
4. If 6 of the 9 judges vote to overturn the pardon, it's voided.
5. If the president has three or more pardons voided during a term, the burden shifts and pardons are void unless a majority of the panel approves them.
6. To prevent the president from flooding the judiciary to exploit the time limit to get his pardons through, a given judge's queue of reviews is limited to n = 16. Any assigned pardon above this limit automatically receives a vote to void the pardon.

The selection algorithm Claude proposed (after some refinement) struck me as brilliant: use HMAC-SHA-256(pardon_id || date_of_first_filed_objection) to generate a sequence of judge IDs to fill the panel. It's publicly verifiable and hard to game, providing an essentially randomly-selected pool. The president can try to game it by ordering the pardons to use the pardon_id value to pick a "friendly" panel, but opposing parties can also game it by picking the day they file... and both sides have very limited options, so gaming it effectively will be possible, but hard, and rarely successful.

Some other important bits: Objections may be filed by anyone but are filed under oath and bad faith objections are also subject to sanctions and contempt orders by the panel. Successful bad faith objections may also expose the objector to civil suits by the failed pardonee. Pardons take effect automatically on day 31 if no objections are filed and on day 61 if the panel doesn't collect enough votes for a disposition. Pardonees who are in custody may demand a hearing to request conditional release. The judge will evalaute their request based on the nature of their alleged offense, their risk of flight and the apparent likelihood that the pardon has the appearance of impropriety, and will decide whether the pardonee should be held or released, and on what conditions.

I think this would work quite well.

Indeed. The chorus of utterly ridiculous claims in support of Trump, many from otherwise reasonably-serious people, are baffling until you understand that performative self-humiliation is their goal. Standing up and saying true or even remotely defensible things to defend him, even using standard politician tricks like deflecting or dancing around the question, is something that anyone tasked with defending him might do, which means it's useless if your goal is to prove your loyalty to Trump.

In order to prove your utter loyalty it's necessary to inflict damage on your own reputation, to abase your self and humiliate yourself in front of the world. This is why Trump demands that his cabinet members pay him ludicrously over-the-top compliments, on camera, at the beginning of each cabinet meeting. It's not just that they know they're lying, they know that everyone knows they know they're lying and everyone knows they're doing it merely to toady. That reputational self-harm of publicly licking his boots is how they show Trump they're irrevocably tied to him no matter what.

Baseload isn't a myth, though the baseload is a small fraction of maximum consumption. It's also not true that nuclear can't load-follow, in the sense that there is nothing inherent in fission power that prevents it from being capable of load-following. It is true that older commercial nuclear plant designs can't load-follow.

High-level waste also isn't really a problem, though the much larger volume of low-level waste is somewhat problematic.

The real challenge for nuclear power is the cost. Nuclear power isn't inherently expensive, but regulations and opposition make it so. This is mostly driven by perceptions that nuclear power is very risky, which means tremendous effort is put into making it safe. The result is that it's demonstrably the safest form of power generation we have, but also among the most expensive. We could reduce the cost by reducing the safety requirements.

How expensive would it be if we reduced the required safety level to put it on par with other energy sources? I don't think anyone knows, and I doubt we'll ever know.

Hence Google's investment in RISC V. It's not yet competitive, but with some time and money it can become competitive. Also, ARM can't raise the prices too much because x86 is still right there.

You don't get to be a three trillion dollar company by saying "We have plenty of money so we don't need to be efficient". You also don't stay a three trillion dollar company that way.

Please elaborate.

Not where I live. They do have a stop sign that flips out to the side.

I'm wondering what I said that made you think I thought otherwise.

Have you pointed this out to Waymo? They're pretty responsive from what I've heard, and this is exactly the kind of thing they'd want to know about and update their model to consider, before a kid gets hit. Not only do they not want to kill kids because Waymo employees are humans, but it would also be horrendous PR that would seriously damage the company.

You can submit feedback through the Waymo app, regardless of whether or not you've used the service. There's probably also a way to report concerns through their web site.

One note: You might be surprised how good the cars are at noticing hidden dangers. I got a ride about ten years ago (when I worked for Google) and I was annoyed when a light turned green but my car just sat there... until about two seconds later when a cyclist came whizzing across the road in front of the car. There was a tall hedge in the way and I don't know how the car "saw" him -- no human driver would, the dude was asking to get squashed -- but it clearly did, and waited. My guess is that although LIDAR and cameras couldn't see through the hedge, RADAR could. Waymo uses LIDAR, RADAR, visual and infrared cameras and ultrasonic sensors so it's quite a bit better at "seeing" than any human could be. None of those can see a kid behind a parked car, though, so maybe they do need to update the model to be more careful in those circumstances.