coderenegade

Regeneration happens directly on the crankshaft via the mgu-k, not the wheels. The mgu is essentially a governor, so you can set it to add or sink (absorb) torque however you like. If you're adding torque, you're draining the battery, and if you're sinking torque, you're charging the battery. Because everything is mechanically connected, sinking torque also wants to brake the car (i.e. regenerative braking).

In terms of power balance, on average, the draw needs to match generation for the system to work. Since this a race, the battery should be flat when you cross the finish line, because you don't want to leave power on the table. When cars are clipping out on the straights, they're trying to claw back some power for the battery so they can deploy it at a more impactful moment in the race. Maybe it's frustrating for viewers, but part of that is that the vehicle has a maximum sustained power, and a maximum total power, and those aren't the same thing.

The mgu-h was cool, but I believe the teams when they say that it was too difficult and too expensive to get right.

There's no reason the battery should run out on the straight, though. You can charge at any point while driving, not just during braking. Obviously you're taking torque from the engine to do so, but that's a strategic consideration.

It's not just about being environmentally friendly; hybrid engines have fairly significant performance advantages over non-hybrid engines.

F1 couldn't call themselves the pinnacle of motorsports if they moved away from hybrid engines. The performance advantages are fairly staggering.

This was the video that earned my sub. Even though it covers stuff I mostly already knew, it's done in such a way that that it breaks down the whys that may not be apparent to people without a background in this stuff.

This and the video on hybrids, are, imo important works, because they break things down in a way that disarms tribalism, and explain how it is that we got to where we are. There are very real stakes in ownership, right to repair, and freedom of movement that rest on people having a good baseline understanding of this stuff.

Even as late as the late 90s, probably into the mid 2000s for manuals. The complexity of vehicles from that era is mostly in understanding the role of the sensors (or that they even exist in the first place). Once you understand that, they go back to being pretty straightforward. I guess the exception here is probably immobilizers.

The complexity of modern vehicles comes down to software coupling between the systems, which (to my knowledge) started with automatic transmissions, and then spread to everything else. In very modern vehicles, even the sensors register with the ECU and communicate on encrypted channels.

The reason for these measures is ostensibly security and quality, but automakers don't seem to mind the fact that it also makes many repairs more difficult or impossible.

There's still a sizable gap between OpenAI, Anthropic, and the Chinese labs. If anything, it's getting bigger. We haven't seen Chinese models solve the types of problems that ChatGPT and Claude are able to solve.

I don't think the math community thinks there's been a plateau. The models have gone from being a joke to being able to crank out proofs to research grade problems.

You can resub on that plan if you've been on it before. They aren't taking new subs on that plan for the time being.

We routinely close the feedback loop when training people in pretty much every industry. Otherwise, what would you say that exams and accreditation/chartering are? I don't see anything inherent in other knowledge work that means AI couldn't do it.

Most professions on the list don't have a shortage, and professions stay on the list for far longer than necessary. We don't need any more IT workers, for instance; my workplace routinely receives hundreds of applications even for temporary roles.

We get to set the terms of migration, because it's our country. I don't think we should be allowing students to bring dependents across, for instance, because they're students. We're already saturated with them, and could stand to be a bit more exclusive.

This. The value of intelligence and execution is plummeting towards zero for knowledge work. Right now, software and mathematics are the most affected, but the transition to every other industry is going to happen with shocking speed.

Yes, the largest chunk is students. In fact, ~6% of all international students in the world attend Australian universities, making it the largest per-capita destination for international students by a significant margin. They still require housing, and they still have working rights. Fake students coming over to work without ever attending a class is a well-known rort.

The single biggest justification for our high migration setting (among the highest in the OECD) is ostensibly for skills we need to keep the economy afloat. I've yet to see a skills shortage closed by migration. And of course, skilled migrants are one of the smaller cohorts in our net intake -- we actually bring in more dependents than we do skilled workers.

Not really. Australia is very lax when it comes to bringing in dependents, who outnumber skilled migrants themselves. We also have an oversupply of technically skilled people, with nothing for them to work on, because we got rid of all of our industry. High energy prices and wages means that you'll struggle to get anything that isn't finance, gambling, or real estate related off the ground. I guess defense is the other option, and it's essentially the only game in town if you want to do something hardware related.

I'm one of the lucky few who ended up in a job that really does exercise my technical training, and a big factor in that was COVID shutting the borders and me getting my foot in the door. That one event meant someone was willing to take a shot on me, and I got put in the hot seat and got time next to very senior software engineers, which changed the trajectory of my career. They got something out of it, too, as it turned out that I really was the right guy for that job.

There are currently 3 million temporary visa holders in the country, roughly 10% of the population. Australia also has a birth rate that is below replacement. All of those people need a place to stay, and if a population twice the size of Adelaide were to disappear tomorrow, we'd immediately go into a housing surplus. In fact, you wouldn't be able to get a loan for a house from the bank, because they're not in the business of investing in falling markets. Which is to say, the housing crisis here is entirely by design. It has been engineered by both sides of government for more than two decades, and is primarily kept afloat by our world beating migration rate. It has to end.

It's less clear cut than that, because modern materials can be better at the job they are designed for. An IKEA kitchen, for instance, might not use traditional wood or stone for its benchtops, but the materials used are actually better in terms of cleaning and handling food on them. It's just that these improvements often get bundled with cheap plastic and parts in other parts of the kitchen to keep costs down.

Similarly, cars today are far superior to old cars in terms of safety, and the gains that have been made on the engine side of things are nothing short of incredible. Every modern car is essentially a supercar by the standards of the 90s in terms of oomph per unit of displacement. The problem is that all of the systems have been coupled together, often needlessly, in order to improve telemetry and capture a bigger share of the used market. It's hostile towards the end user.

There's no question that the level of quality today is higher, especially where it matters. It's just that value extraction eventually turns hostile towards consumers in the absence of real competition and the need for endless growth.

There is a general solution to the three body problem due to Sundman, in the form of an infinite series. Unfortunately it converges so slowly that it's useless in practice.

Being unable to state the solution of a problem in closed form doesn't prevent solutions from existing.

I can honestly say that I've never had to compile wifi drivers from source in more than a decade of using Linux as my main OS, even through my distrohopping phase. And I'm not researching my hardware ahead of time, either. I don't even know what you'd need to be running in order to have to do that? Maybe void or something similar? I don't think that my experience is particularly unique.

The sharp edges are usually pointing the package manager at a repo (generally an official expansion) that lets you install the necessary binary.

Sure, why not. I haven't uploaded anything to github/lab yet, and you'll have to give me a few days, since I'm busy juggling work and being a new dad.

Full disclosure, this is one of my pie-in-sky ideas that I've been letting AI loose on with only top down direction from me. I'm usually fairly hands-on even with the model in the loop, but this is an exception, because a geometry kernel is frankly beyond my ability to write solo (and I say that as someone with a reasonably strong math background). The gamble has always been that if a current model can produce "works, but crap under the hood", future models will be able to clean it up to "works, and is solid under the hood".

I'll push everything up by the weekend and drop you a line.

The problem with the article's line of thought is that mathematicians can't control what others do with models that are capable of generating proofs for hard problems. Sure, maybe there's a career in taking known proofs spat out by the oracle, and translating them for mortal digestion, but I'm not sure that's what most mathematicians signed up for.

I've been working on something similar as well, but not yet ready to be shared. It's a geometry kernel and then an authoring layer on top. The idea is that the constraints are tests that get run after the geometry is defined, so it's more programmatic. Sol was able to generate STEP files for credible turbo machinery with it, and I have Astra building out the assembly for an inline three engine idea I had. The drawings are looking credible, but that's about all I can really say. I'm just blown away at how far the models have come in the last 12 months.

I switched from Claude to Codex because Claude just doesn't do what you actually tell it to half the time. It dances around the edges and does busy work without actually tackling a tough problem.

I'm not sure what others are doing that they're getting such different results, but I'll take Codex every day of the week.

At a certain point, you don't have a choice. Before China got into the game, the only way to avoid giving Luxotica money if you wanted a pair of glasses was to essentially not buy glasses. This is the same for many industries -- consolidation behind the scenes.

They became lazy and uncompetitive, and were eventually overrun by Japanese imports. It's probably reasonable to say they never really recovered.

My newest and most expensive car cost $5k in freedom dollars. It gets the job done. I have an older, more clapped out vehicle that cost $1500, and is in surprisingly decent knick for a 30 year old car, but it does need a paint job and some imminent TLC to keep it from deteriorating. That will cost more than it's worth, but I think it's worth doing in the long run.

Depends on the Mercedes. There was a time where Mercedes made cars that had a reputation for reliability similar to Toyota's today, and those cars are still fairly reasonable to keep running. The problem is that they're now becoming collectors items. Anything from the last 25 years though? Best to look elsewhere.

I think the only thing that stops this from becoming true is what Chinese and European labs decide to do. If they can keep up and keep opening their weights, then we might see some kind of democratization. But right now it looks like the gap has increased, and those groups can't replicate research that isn't published, or distill models that are internal only, or for select (very wealthy) customers.

Not necessarily. That's the best case scenario, but proofs can be intrinsically useful in and of themselves. It's just that for problems of that nature, speculative work is often done ahead of time, e.g. the body of work that already exists assuming the Riemann hypothesis is true.

This is the hope, but I suspect the reality is that we see an ever widening gap between the fortunate and the unfortunate. We're looking at the automation and commodification of all knowledge, and the best models will be kept locked behind closed doors so that they can't be stolen. And, of course, "for our own protection".

This is the real problem. We're looking at a future where those who control AI have an insurmountable advantage in everything. They can control the amount of intelligence the masses have access to -- for their own safety, of course -- and they will never, ever be able to close the gap.

Surely part of the problem is that intelligence seems to be implicitly conditioned on embodiment, to the point that "covering only cognitive tasks" seems inherently ill-defined or arbitrary. Everything we do is a cognitive task. At some point, the criticism will be "sure, it can solve research-grade math problems, but it can't fold my laundry".

Even our large language models have an implicit embodiment in the domain of text (and more recently, multimodal inputs). That seems sufficient for certain things, and insufficient for others. I suspect that AGI that does everything a human can do eventually turns out to be fairly analogous to humans in terms of sensory input and domain output, even if the scale is radically different (e.g. thousands of robots uploading (touch, sight, audio, smell, etc.) sensory data to a single model, and each being actuated individually).