tokipin

These articles always miss the most important point. People are buying EVs because they're better. That's the whole point of Tesla making attractive vehicles, aesthetically and functionally. That market pull allows for an S-curve in the technology, since it reduces market risk for investing in each increase in scale.

This is a completely different scenario than what the article suggests, which is that EVs are some kind of medicine for climate change that society is trying to have discipline about. Not so. It's a life or death situation for the auto industry, they don't have a choice. They either join the race or die. This makes most concerns about the transition irrelevant. For example, laws banning gasoline cars by 2035 are as useful as laws banning flip phones by 2015 would have been.

This is somewhat along the lines of my current theory, which is that night owls vs early birds is about sensitivity to the environment. The reason night owls stay up late is not because it's late, but because it becomes quiet enough that the environment they're sensitive to stops taking attention and they can direct that attention inward. According to this theory, it would make sense for introverts to be more affected by this as they would accumulate an "introversion deficit" throughout the day.

That's roughly my thinking on Toyota's strategy if we give them the benefit of the doubt. They want to own a larger piece of the ICE pie as it shrinks. This is a great strategy. What they might not appreciate is that the oil pie will also shrink and that both of these industries will have increasing costs across their entire supply chains as volumes diminish. Their vehicles will suffer for both ICE tech and gasoline propulsion and be completely outcompeted on cost and convenience. This will happen over a much shorter time frame than they expect.

Green solutions are part of the continuing decentralization through technology and will be just as popular in developing countries as in first world countries. One practical example of this is that green tech can be financed in bite-sized chunks, compared to the large commitments needed for power plants.

Like many things, it's about how far into the diminishing returns you want to go. At some point it's more effective to apply further investment to other parts of the ecosystem.

Well, that's an appeal to authority. It also assumes the companies wouldn't be successful with a different hiring approach. In reality the issue is way too complicated to really break down. These companies operate at a large scale and went through rapid growth spurts, and what they believed was a great way to hire was also an implicit tradeoff they were making for the sake of growth.

I'm sure the companies themselves understand that much, and understand a variety of the tradeoffs involved with the way they hire. What they might not understand is that they are at the exploitation stage of their lifespans. There's an entropy-like process as a company matures and cycles through its workforce which leads it to becoming highly specialized.

My bet is these companies don't understand the various ways in which they're limited, and therefore the ways in which they're limited in filtering candidates. And even if they did, they couldn't really hire any other way now even if they wanted to because they don't have the internal diversity for it, that possibility has long since closed on them.

they're prepared for tasks being assigned to them that exercise those skills and by the end of the half they'll have a deep understanding rather than just a wrote understanding.

I think it's entirely possible to memorize algorithms superficially. These types of interviews are selecting for exactly the type of person that is good at that rote memorization, and are otherwise selecting for a very narrow type of skillset.

It's unreasonable to not expect pathological outcomes with such a rigid system, especially when some people are themselves specialized for the ability to game social systems. Nothing is free. There are always trade offs when choosing one approach over another.

My guess is they are mid-layer activations in the brain that your consciousness (attention?) is connecting to, or which are being directed to the lower layers and interpreted as input.

Dashcams were common in Russia and it happened in the YouTube era, so there were plenty of videos of the event:

https://www.youtube.com/watch?v=dpmXyJrs7iU

As the company grows the types of people that are attracted and unattracted by the company change, as do those that are accepted and rejected by it. For example, risk averse people are attracted to mature companies, and they don't like risk takers.

One of the consequences is that companies become more specialized and lose deep expertise. To put it bluntly, its because they no longer have people with deep expertise, and even when they do they are ignored by management who tend to be less interested and knowledgeable about the subject matter.

I don't know of any company that has tried to prevent this. It's a general force that permeates everything, including recruiters, HR, their job postings, etc. So even companies that think they're preventing it probably still walk right into it regardless.

Of course, specialization is not necessarily good or bad. In a very competitive environment the company may not have a choice.

Here's a video from someone who observed a design review for Starship[1]. From the sound of it, Elon is in the thick of it in the R&D engineering.

[1] https://www.youtube.com/watch?v=S1nc_chrNQk&t=380s

I would agree if their increase in data was linear, but it is increasing by orders of magnitude, which should have qualitative consequences for what they're able to accomplish as they claw their way through 9s. I don't see how it's possible to get progressively more 9s without scaling in both data and compute.

The point of the higher scale isn't just more data, it also makes it easier to solve the unbalanced data problem, because rarer and rarer scenarios will appear in large enough numbers to work with.

I predict the opposite. Tesla sold half a million cars last year and will sell nearly one million this year. The data they have access to is increasing by orders of magnitude. I bet there is a point, let's say 20 million cars total, where they can pull so much high quality data that they will be able to surpass lidar capabilities for the purposes of self driving.

The lidar/no lidar discussion is a fun one because people have different ideas about how the world works. Personally I think LiDAR is the modern version of expert systems. It appeals to a logical/geometric intuition but the approach is brittle to real world contact, especially when paired with HD maps which are a great way to drive yourself into a local maximum.

Despite their stated intentions, the hiring pipelines are for specialists. The leetcode tests are good at selecting for specialists, since the nature of those tests (irrespective of the subject matter) matches the specialist skillset.

It's important to distinguish between specialist/generalist roles versus specialist/generalist people. Most of FAANG is hiring specialists people for both roles. That's my sense of it anyway.

Overwatch is a frustrating game to play because they almost have it right. It's almost there, if they could just tweak a couple things. A good example is the overtime system. One of the worst ideas in games I've run into. For one, it creates artificial and spiky emotions in players, which just burns people out. It's amazing they don't realize basic things like that, which means they actually haven't had the right talent for a long while.

The company also has a lot of hubris. They spend tons of effort on things like trying to make healers and tanks fun. There's many good reasons to do that and they gave it a good try, but the conclusion should have been: we aren't able to make healers and tanks fun, let's try something else. Instead, they forced people to play them.

Other mistakes include not understanding "the meta." Once they think the meta is balanced in a game, they get conservative and don't want to tweak things too much. They don't realize that the meta is an equilibrium state found by the players. Change the game, a new meta will be found. They don't seem to get this and just let their games run stale because they think the competitive scene drives the player base, instead of the other way around.

Overall the facepalm factor has been too high for too long. Blizzard has always understood the science of game design, but since at least SC2 they lost the art.

Yep, this happend to me with a LinkedIn trial. I had my credit card changed but somehow LinkedIn was able to charge the new number, which defeated the entire point. Trying to contact LinkedIn revealed they don't have any humans to contact. They even have a fake "3 minutes until you can chat with someone" that always errors out. Chargeback to the rescue, but in the future I will just switch banks entirely.

Some skeptics remain skeptics up until the point that they've been proven blatantly and obviously wrong and people stop accepting their arguments, so for some "skeptics" at least it's more of a stubborn mindset backed by confirmation bias or other issues disconnected from the actual subject matter, rather than sound reasoning.

It seems more like a classic case of refactoring except that the original infrastructure is paper-based, in which case the ideal digital redesign would be one that mimics the paper-based one precisely so that "regression testing" can be performed and components can be swapped in piecewise. Once enough components are digitized, the system as a whole can begin to be molded into more effective forms. This would have to be the plan at the outset, since it would require preserving enough flexibility.

I think people are deceived by Swift's syntax. It looks simple by syntactic sugar, but it's not actually any more approachable than other systems languages like C# or Java. Given that a lot of work in ML is exploratory, it doesn't seem like a good fit compared to a hackable scripting language like Python. I would bet against SwiftUI for similar reasons.

One of my theories is that this (or similar virus-like mechanism) is how some memories are transferred to offspring. The reasons are:

* Prodigies which happen to be good in a field related to what their parents are good at. Often the wording used indicates a memory-like reason. For example, finding the rules of chess "familiar."

* Embeddings in machine learning, which allow encoding of high-dimensional information in a low-dimensional space. This would imply that humans have a somewhat-shared embedding space at some "layers". It's interesting to wonder how similar this space would be across various animals, as it would effectively be an evolved language.

* Other more subtle things, like the "cultural memory" of a society.

One question would be what types of time scales these memories survive across. Maybe some forms of instinct are deep memories that have been directly encoded in DNA.

I think it follows a general theme where some people have a "tricks"-based mindset. Some people look at Tesla and think it's just tricks. Such as the idea that people are buying Teslas because of the drivetrain - which would be a simple trick, rather than the reality that Teslas are just all-around great cars - something much harder to understand.

The Nikola founder probably thought Musk had succeeded in unlocking a series of successful tricks, and set out to reproduce them. But when you don't know the reason for the formula, you don't know what steps you can leave in or out, so you just follow it mechanically.

The Elon Musk biography has Musk's side of the story in the appendix. Paraphrasing, but basically he saw the tooling for Windows software development (Visual Studio etc) as being more coherent/batteries included/practical, in part because of Microsoft's investments in gaming and needing those tools to work on commercial games, which were stupidly complex even back then. Keep in mind Musk had some experience in the gaming industry. I'm not familiar with Linux, but it was likely rudimentary by comparison at the time.

Another good point he brings up is that you could find a lot of smart engineering talent, who is working on games, by using their tools/language of choice. SpaceX uses a similar tactic on a more subject matter level. People that can understand 3D game programming are probably a great pool of talent for writing rocket software.

The biggest issue I've seen in online courses is production values. In fact I think crap production values and the lack of a quality bar is why platforms like Coursera have done poorly compared to their potential.

The most obvious issue is usually microphone quality. It's baffling that people spend so much effort developing a course and multiply that effort by a shit microphone. If you aren't willing to spend/get money for a good microphone, your philosophy on what is important for online teaching may have some blind spots IMO.

The bots were using their ultimates pretty liberally. Maybe their "bot meta" is very group-intensive, so if they see one hero there's a good chance there are other heroes nearby? Other possible reasons: Opportunity cost of not using ultimates, getting a single kill late game can mean victory, bots penalized for games going later or their games just not running long in general?

As a naive example, if in a CPU you are doing some specific 3-step algorithm which would normally require 3 clock cycles, you can instead bake that algorithm into a circuit along with the memory it needs that will run in 1 cycle. In Elon's terminology, the CPU has to "emulate" what the baked-in circuit simply is.

But from the sounds of it this chip goes beyond that. It sounds like some kind of RAM/GPU hybrid with matrix-specific instructions specifically targeting NN inference for Tesla's specific style of NNs, which as a first guess I take to mean "lots and lots of layers."

In the call they mentioned increasing their investment in the chip team and technology, which is a sign that they're confident it's going to pay off.

Musk commented about LIDAR in a prior earnings call[1]. Tl;dr he believes that LIDAR is a short-term crutch that distracts from the real long-term challenge, which is the machine learning, not the sensors.

[1] https://www.youtube.com/watch?v=FaW85fIos64&t=21m40s

Most people who participate in the threads do so for complex social reasons that are not particularly different than the reasons that some people buy Starbucks, some people buy from an indie shop, and some people hit the Timmie’s drive-through.

To make a stronger statement, there is "actual" value, it's just that people are bad at describing it. Many people will go to Starbucks every day because "they like it" and that's about as good an explanation as you can get out of them. But they didn't just wake up one day and decide they were going to like Starbucks.

For me personally, I like the consistency of Starbucks. If you order the same drink at two different stores you'll get the same drink. This isn't something that a large corporation like that does on accident. It requires a lot of effort to get that consistency. And it's just one of the many reasons someone will like Starbucks without recognizing it's one of the reasons.

People similarly like Apple products for reasons that sound vague ("it's easier to use") but which are not an accident. Apple is highly vertically integrated to ensure many of those reasons. It's no coincidence that Tesla is also vertically integrated.

Yep thanks I just remembered that's what it was ("it had a J in it"). Woops. Not sure why I've had them confused for Digi-Key the last few years. Maybe the Digi-Key branding is just more memorable.

They have (had?) a local warehouse I think off 101 in the Bay Area, don't remember exactly where. You could walk in with a list of parts and they would grab them for you right there.

It could be the iPhone 1, or it could be a distant memory, or it could just be another car.

I think your mistake is thinking that this is a roll of the die. Just because something is difficult to discuss doesn't mean there aren't people that understand it. And many pro-Tesla arguments are systemic, which also makes them difficult to convey (concepts like "game-theoretic equilibrium" are not in common lingo). Many pro-Tesla-ites will sound hype-y because they don't have a good way of explaining their conclusions, not because they don't have a clear understanding.

Then I contrast that with the pie-eyed faith of the “Musk is saving the world for us,” fanboys and it makes me tired.

There are things that the fanboys understand that critics don't. It will make more sense to everyone once people understand that the Model 3 is the iPhone of cars and the Gigafactory and other investments start paying for themselves.

As far as the pro-Tesla and anti-Tesla arguments, I find the two sides talk past each other because the time frames being discussed are very different. Case in point, it's not that Tesla investors are covering their ears and saying "lalalala" about Tesla's missed deadlines. It's that they don't care. Based on the Model X delays, I didn't expect the Model 3 to even enter production until the end of this year. Besides the financial risk factor, I couldn't care less about the company missing its deadlines by a year. It simply does not matter. Repeat for many other topics.