I think they just do what they have done well. LLMs don't take demand away from HPC, like physics and weather simulations. Arguably if some of their competitors divert resources to LLMs it might even be better for them.
Historically, in order to get support for one of the Asian languages, you typically had to buy the watch in the exact right country, not anywhere else.
I hate Garmin with a passion because their watches are effectively region-locked by language support, an insanely consumer-hostile move in this day. I was unable to use any features related to text or whatsapp messages because the watch shockingly could not decode messages in my native language.
Their software was also so flaky that I was woken up by a faulty vibration alert in the middle of the night multiple times during the few months I wore the Garmin Instinct Solar, and at least twice I was unable to fall back to sleep. That is, the watch was supposed to be in silent or DND mode, but the watch probably crashed or reset in the middle of the night, losing the silent or DND state, allowing an alert to go through. The sleep tracking was also very inaccurate, and sleep tracking is the single most valuable metric for me.
To this day I fantasize posting a video where I smash my Garmin watch to pieces alerting other people how bad it is. Still, the hardware was near perfect and it's hard to hate the watch itself. But because of the software issues, it was no better than a dumb watch to me. I hate Garmin the company.
I suspect the lack of success of the teams resulted from overfitting for the more standard pattern that is solving a max flow, and I also suspect the organizers are deliberately using this to trick the contestants. Spending time to practice coding vanilla algorithms to solve either max flow or a linear program is a waste of time for >99% of computer scientists.
People obviously turning out LLM code uncritically should be investigated and depending on the findings made redundant. It's a good thing that it allows teams to filter out these people earlier. In my career I have found that a big predictor of the quality of code is the amount of thought put into it and the brains put behind it - procedures like code review can only catch so many things, and a senior's time is saved only when a junior has actually put their brain to work. If someone is shown to never put in thought in their work, they need to make way for people who actually do.
Also one way Chinese EVs combat their efficiency disadvantage is fast charging and large batteries, and the charging cost is so negligible anyway that largely the user experience has been the same. Tesla has very little first mover advantage left against Chinese EVs, if any. It's sad that other Europe and US manufacturers haven't been able to be competitive, and probably Ford and GM will never be competitive because of protectionism.
One drawback cited consistently about Tesla cars is poor suspension. They may or may not have better design overall, and have better EV manufacturing than US competitors, but I don't see an "immense advantage" when they fail to address one of the things many users easily feel.
Over a time frame of multiple decades, it's been volatile but overall kept pace with money supply. Timing is always tricky, even with SP500 which has given higher return.
Maybe "good at" was stretching it, but there are things that lagged forever, such as a retail savings account and Japanese government bonds. It's tracked CPI better than silver.
The price of gold is not immediately responsive to goods inflation, it can be lagging or leading and the correlation can also fluctuate. Long term it's been good at tracking money supply but short term it doesn't necessarily give you a timely signal.
Wozniak was indispensible in the early days. They had to survive first, and then Jobs could have a chance to thrive, and having someone like Wozniak greatly increased his chance of survival - it's not easy to find someone like that. That he continued to add much more value than Wozniak could to the business is another matter. There maybe many Jobs who died without being known because they could not find a strong partner.
The AI needs to be able to lookup data and facts and weigh them properly. Which is not easy for humans either; once you're indoctrinated in something, and you trust a bad data source over another, it's evidently very hard to correct course.
FSD itself is not a failed bet as a driver-assist system. When it started Tesla was cash poor and often on the verge of collapse, so cutting costs made sense. They will never do as well as a mature Lidar-based solution, but over the years FSD added some value to some people, and might turn out to be a slight competitive advantage in selling cars.
I also think Tesla and their robotaxis are egregiously overhyped, but cutting Lidar was not a terrible business decision at the time. Though it's a terrible decision to have still stuck to vision-only (and using low-quality cameras apparently) when they could have at least got Radar and cheap Lidars.
Maintainable software is also more maintainable by AI. The required standards may be a bit different, for example there may be less emphasis on white space styling, but, for example, complexity in the form of subtle connections between different parts of a system is a burden for both humans and AI. AI isn't magic, it still has to reason, it fails on complexity beyond its ability to reason, and maintainable code is one that is easier to reason with.
The error margins will be huge, and for small enough force (like the skinning part or handling fine mechanical stuff) there's basically almost zero signal.
The tasks you do to recover from the failure is often different from the happy path. For example, the happy path of dumping garbage is carrying a garbage bag to a collection bin. The non-happy path is that the bin is overflowing and you have to put the bag on the ground, or if the bag leaks and you need to move to a new bag, or if the bag breaks entirely and you have to pick up the trash again.
But yeah, I think a better way to put it is that sampling the happy path would indeed make the failure case easier, but sampling just happy paths is far from sufficient from completing even some of the simplest human tasks with failure.
Pure vision will never be enough because it does not contain information about the physical feedback like pressure and touch, or the strength required to perform a task.
For example, so that you don't crush a human when doing massage (but still need to press hard), or apply the right amount of force (and finesse?) to skin a fish fillet without cutting the skin itself.
Practically in the near term, it's hard to sample from failure examples with videos on Youtube, such as when food spills out of the pot accidentally. Studying simple tasks through the happy path makes it hard to get the robot to figure out how to do something until it succeeds, which can appear even in relatively simple jobs like shuffling garbage.
With that said, I suppose a robot can be made to practice in real life after learning something from vision.
I guess Mark Zuckerberg likes Alexandr Wang a lot. They have a lot in common. They both value intensity and are willing to engage in morally dubious tactics. The culture of Scale sounded quite a bit like Meta these days (and not Meta in mid 2010s) and may indeed be what Zuckerberg envisioned Meta's culture should be. So even if Scale's product isn't the best, it probably seemed like delegating to a copy of Zuckerberg himself, so it just "felt right".
However, a top research lab needs to be competitive yet still have an environment that fosters intellectual honesty. Meta Gen AI did not seem like that, and I don't think Scale's culture is like that either.
For a long time iOS did not have features to limit data usage on WiFi. They did introduce an option more recently for iPhone, but it seems such an option is not available to MacOS. Windows supported it as far as I could remember using it with tethering.
To make fspeech's point more obvious, while the rare earth materials may only be worth billions to China over multiple years, they could destroy entire industries worth trillions of dollars over the world, and in particular US. This is less obvious for other China exports, but with the "leverage" that's mentioned, you can assume for each dollar worth of imports lost from China, US may lose more than one dollar, possibly several dollars worth of production. On top of alienating multiple other trade partners and sending them to work with China. The US is losing its guts, its intestines, its kidneys, while China may only be losing an arm or a leg, or even just some of muscle and blood.
One interesting thing about working on numbers so large is that it becomes harder to assume that memory access is constant time. It's already not that easy now (cache miss vs hit, and external memory), but when you operate on data sizes that large, memory accesses matter by at least the square or cube root of N. If you require largely sequential access it is possible to outperform an algorithm that requires conditional random access by some function of those factors, which can make some algorithms look much worse or better compared to the classical analysis. Of course, it still does not matter whether your algorithm is going to finish in a couple of trillion years or a trillion trillion years.
I have found that vibrating alarms on wrist watches to be very effective. For over 5 years I've been setting a vibration alarm on my watch and another backup alarm on my phone. I never slept through the vibration alarm. Granted YMMV
You don't need an HDMI port, you just need a driver to support running the right graphics calculations and producing image to funnel to another output port. The GPU may lack some features, may have an architecture that is bad for rendering, and may be suboptimal in delivering the performance per watt. Exactly like how a CPU doesn't have a display port.
I'd argue chronic conditions that are debilitating should have at least the same priority as cancer, assuming their prevalence in the general population is similar. Long covid is more like to affect productive age people compared to cancer, so a government would be wise to prioritize it.
There are secrets like passwords, but there are also secrets like "these are the parameters for running a server for our assembly line for X big corp".
For much of SP500, the largest shareholders are the passive investors. It's common that an activist hedge fund can influence decisions with a mere 3% or so stake because the likes of BlackRock and Vanguard support them or abstain.
It's possible to treat software close to a first-class citizen even technically as a cost center. Probably the key is to have a lean team that is effective and management trusts, in a business whose profit margin or stability can be greatly enhanced by good software. If you're relatively lean then nobody sane would be looking to cut fat in your team. On top of that, the attitude of management and the culture is really important, and this CEO's attitude is at least a good start.
I hear Meta Ads has one of the more toxic environments for SWEs, even though Ads make money.
You can learn to take good risks and handle hardship without taking stupid unnecessary risks. One important life lesson is that the only risks worth taking are those that offer corresponding upside - else the expected outcome is ruin. Education wise this means you give them necessary or low-impact risks to take - and let them endure outcomes such as failing a difficult but important project, failing to find love, losing a basketball match, or losing friendships. Of course, sometimes you may need to do something even if there is no upside for yourself directly, but that is outside the scope of this topic.