I'd rather pay $9.99 per month for what Netflix used to be before all the companies said, "I can milk these properties for more money if I create my own streaming service and cut out the middleman."

There may or may not be too many streaming services, but there are WAY too many streaming services owned by content distributors. You can't have any sort of meaningful free market among streaming providers if they're all just providing their own content. You still have competition among content providers at that point, but zero competition on the streaming itself.

I wouldn't call plant-based meat alternatives "healthy" unless your idea of healthy is dying of salt poisoning.

Apple was one of the cofounders of ARM (the company) in 1990. It did not create the architecture, though it likely had an impact on ARM6 (ARMv3 architecture) and later. Either way, the ARM architecture probably would not still exist if ARM (the company) hadn't been founded. The ability for multiple companies to design and manufacture chips turned out to be critical for its long-term survival and viability in the cell phone market and others.

ARM (short for Acorn RISC Machine or Advanced RISC Machine) is an acronym, and all letters are capitalized. Arm is something attached to your torso.

I'm not saying Apple Silicon isn't better than the competition — it is — but that's not a fair comparison. Raspberry Pi's performance is largely because they use Broadcom chips, which stay several generations behind the state of the art. For example, the Raspberry Pi 5 (released in 2023) was designed around the Cortex A76 CPU (released in 2018).

Disagree. They put blazingly fast single-core performance and roughly half the speed of a workstation-class laptop in the power envelope of a netbook.

• M5 PassMark CPU Mark: (28561 multi / 6001 single)
• Intel Core Ultra 9 275HX top-end laptop chip (56007 multi / 4745 single)
• i9-14900KS top-end desktop chip (60511 multi / 4828 single)

They're nowhere near the top overall, but their single-core performance (which affects perceived speed more than multi-core performance, typically) is at the top.

You really can make the comparison. Which one is best depends on the workload.

Yeah, that's about right. But Apple also uses those chips in desktop, where the comparison is not nearly as rosy.

Don't get me wrong, I love my M1 MacBook Pro. The battery life is spectacular, and performance is good enough. But I'd be lying if I said there weren't workflows for which Intel would be better. :-)

They are no longer required to fully fund pensions:
https://en.wikipedia.org/wiki/...

No, it was no different with the US. This is an irrelevant story for the whole world outside of the US, because they have been living in this reality for 75+ years. The only nation who is feeling a new pain is the US.

Correct, the problem is that China is acting more like the US now. Which is more of a problem for the US than for other nations, because we have been taking advantage of our unique status to increase our standard of living at other nations' expense, and that will now be harder. You have it backwards about who should be worried about it. My guess is out of some misplaced feeling of hypocrisy.

A CEO doesn't get paid for any of the work AI does. CEOs collect information from other executives, peers, consultants, and the media and make decisions. LLMs can disrupt the work of consultants, the media, and the employees feeding information to executives, but it's horrible at making good decisions that can be trusted.

In fairness, since some of the claims are that AI will replace all jobs, even massive disruption such as replacing 10% of the workforce is still a very big deal. I'll be surprised if we don't reduce our call center staff by at least 50% in the next 3 years, and AI chat/voice bots is a small portion of that projection. That is mostly from AI agents assisting call center agents and assisting product managers to find ways to improve human agent UX.

LLMs were capable of doing all of this in early 2024, and have only gotten better since then. We weren't having success with nano/flash models in 2024 but we have been moving to those models for most use cases in late 2025 (reducing LLM costs by 80%).

That is usually true, but we don't always use AI to replace employees (although it usually does).

I am working on something now that reads all of our transcripts and identifies what part of each call takes the most time to help product management prioritize call center improvements. Traditional NLP couldn't do as good of a job at this as early testing is showing LLMs can do. We would have to more than triple our call center staff to have a human listen to every single call and identify opportunities to improve call center agent UX, but a nano/flash LLM can do this for around 1 cent per call. For $250k we can do this for our 25M annual annual calls. That isn't replacing a human. It is doing something we would have never paid humans to do and giving us information we never would have had.

This information will still be used to either decrease call center staff or increase the caller experience, but that is true of every product enhancement we do for this business function. Not just AI.

No, it's how you get mens rea for the felony murder rule. You didn't carry the gun with the intent to kill, only to intimidate, but you still had a guilty mind, and if you then used the gun to kill someone in the heat of the moment, there's your mens rea.

And remember that actual cause does not mean literally pulling the trigger. At least in the U.S., the courts apply a "but for" test. If the event would not have happened without the previous event, then the previous event is considered the actual, not proximate cause. The police would not have shot the other person but for the perpetrator pointing a gun at someone (and possibly shooting at the police).

IMO, that's not meaningfully different than involuntary manslaughter convictions for allowing unsafe working conditions at a construction site or leaving your loaded gun out where a child can take it, both of which have happened.

True, but only momentarily. At the end of the first depreciation cycle, assuming purchasing of hardware is not accelerating, you're depreciating 5x as much hardware over 5x the time, and your momentary bubble in the stock price is gone.

And even if hardware purchasing is growing right now, eventually, that will flatten out, and the above will be true.

The only real question should be whether the depreciation rate is reasonable. If you're still getting substantial use out of the hardware after five years, then depreciating it over 3 years is questionable.

Also, the more slowly you depreciate it, the less you save on taxes each year. Faster depreciation is beneficial if you think the tax rate will go down and you will lose the benefit of that depreciation. Slower depreciation is beneficial if you think the tax rate will go up and you will benefit more from depreciating it later. So this may also mean that these companies are expecting corporate income taxes to go up. Make of that what you will.

In my experience, with only rare exceptions, the limit should be three. More than three, and you are likely involving people working on multiple projects who don't really need to know what the people on other projects are doing beyond what an email every few months would provide.

Those rare exceptions are situations where you have a meeting of managers in an org or similar with each other, where everybody is working towards the same goals, and they're planning towards those goals.

Or the way I usually describe it is that the usefulness of a meeting with n participants is one over the square of n minus 2 for all values of n greater than 2.

Depends on what the person was doing at the time. If the person who didn't pull the trigger was holding up a liquor store and the police shot the wrong person, there's at least arguably mens rea, which is how we get things like the felony murder rule. Extending that to involuntary manslaughter when the person didn't actually pull the trigger but directly created a situation where the police did seems like not that much of a stretch to me.

To my knowledge, you are pedantically correct, but that doesn't mean you aren't in legal jeopardy for the death; you just won't face murder charges. You could still very easily be hit with civil wrongful death claims, and maybe negligent homicide or involuntary manslaughter charges for creating the situation that led to that death.