fl7305

> See the recent deals where it’s been used as a political bargaining chip; it still ended up being a very desirable and capable platform from my understanding.

From a european perspective, I can tell you that the mood has shifted 180 degrees from "buy American fighters to solidify our ties with the US" to "can't rely on the US for anything which we'll need when the war comes".

> It is impossible for a simulink model to accidentally type `i > 0` when they meant `i >= 0`

The Simulink Coder tool is a piece of software. It is designed and implemented by humans. It will have bugs.

Autogenerated code is different from human written code. It hits soft spots in the C/C++ compilers.

For example, autogenerated code can have really huge switch statements. You know, larger than the 15-bit branch offset the compiler implementer thought was big enough to handle any switch-statement any sane human would ever write? So now the switch jumps backwards instead when trying to get the the correct case-statement.

I'm not saying that Simulink Coder + a C/C++ compiler is bad. It might be better than the "manual coding" options available. But it's not 100% bug free either.

> you guidance inevitably falls out of the context window

Yes, for a pure LLM session.

But using GitHub Copilot, the agent server picks out what it thinks is your most important rules, and inserts them as part of the context before running each new prompt.

So it becomes a sort of "long-term memory" outside of the context window.

You can also write your own copilot-instructions.md which is also inserted into the context window.

> The order defines Class M1 vehicles as “those falling within class M1(a) and class M1(b) as specified in Schedule 1 of the Vehicle Classes Regulations, which refers to another bit of UK legislation.

> Oxford’s congestion charge is almost certainly enforced by cameras that scan your number plate. An ox-drawn cart doesn’t have a number plate, so it won’t be charged. Other vehicles like a Renault Twizy or Reliant Robin do have number plates, so they’ll be charged even though they’re technically exempt.

So there's not much to it: Plates are scanned -> the number is checked in the vehicle registration database -> not class M1(a) or class M1(b) -> no charge.

This goes for ox carts, Twizys and Reliant Robins.

In 1971 there where ships with almost twice the displacement of the Dali.

They weren't freight ships destined for Baltimore, but it wasn't hard to imagine future freight ship sizes when designing the bridge in the early 1970s.

The London sewer system was designed in the 1850s, when the population was around two million people.

It was so overdesigned that it held up to the 1950s, when the population was over 8 million. It didn't start to become a big problem until the 1990s.

I think it would be well enough for Nethack.

Would be interesting if even a Z80 could run Nethack in a CPU emulator.

You could easily write a RISC-V CPU emulator for your older/smaller machines, and run the original Nethack code compiled to RISC-V.

> In particular, at the time that it was written, it meant arm themselves with military weapons for the purposes of military action. That's what the contemporary use of the term "bear arms" was understood to mean. Try to find any mention of self-defense from back then. It wasn't what they were thinking about.

That's what I meant too. I didn't bring up self-defense, did I?

The 2nd amendment protects the states' right to build up their own state militias by allowing their citizens to arm themselves with military weapons. It forbids the federal government from interfering with this.

> The modern interpretation of the second amendment is very different.

Yes. The federal "assault weapons ban" is completely incompatible with the 2nd amendment.

This was pushed through without a new amendment. Instead people used linguistic acrobatics to re-interpret the meaning of the 2nd amendment.

It would have been a lot easier today to shut down any attempts to re-interpret the 14th amendment if we hadn't started down this path with the 2nd.

> we cannot function as a nation on the basis of, "Aw, everyone knows what they meant...".

I guess that's fine when it comes to the 2nd, but not as fun when your opponents tries the same for the 14th?

> Effectively no legal scholars or judges of merit support that belief

And how did the "legal scholars or judges of merit" interpret the 2nd amendment in 1800?

The same way as today?

The constitution seems to have become a lot more flexible today than people should be comfortable with.

>> Executive orders cannot overrule the Constitution.

> I would hope this is a fairly universally held position, not so partisan.

I agree. I think the constitution limits both the executive and the legislative branches.

> how will they feel if the next administration decides to pull the same stunt with the 2nd?

The 2nd amendment has already been overridden by federal laws without a constutional amendment.

The 2nd used to mean that the states has a right to let their citizens arm themselves privately with military weapons. The federal government was forbidden by the 2nd to interfere with this.

I'm from Europe and fine with the very restrictive licensing we have here.

But it looks very shortsighted to wildly re-interpret the constitution far outside of the original meaning, instead of passing new amendments.

Some of us are ancient enough to have been born before the last person walked on the moon.

>> the presumption that reasoning skills cannot evolve

> I don’t know how to address the evolve part. LLMs don’t directly mutate and have selective pressure like living organisms.

Sorry, that was poorly worded. I meant "can reasoning skills not be evolved through the neural net training phase?"

Sure, once you deploy an LLM, it does not evolve any more.

But let's say you have a person Tom with 5-minute short term memory loss, meaning he can't ever remember more than 5 minutes back. His reasoning skills are completely static, just based on his previous education before the memory loss accident, and the last 5 minutes.

Is "5-minute Tom" incapable of reasoning because he can't learn new things?

> They appear completely brain dead at times

Yes, definitely. But they also manage to produce what looks like actual reasoning in other cases. Meaning, "reasoning, not pattern matching".

So if a thing can reason at some times and in some cases, but not in other, what do we call that?

An LLM is a lot like a regular CPU. A CPU basically just operates step by step, where it takes inputs, a state memory, and stored read-only data, puts those into combinatorical logic to calculate new outputs and updates to the state memory.

An LLM does the same thing. It runs step by step, takes the user input+its own previous output tokens and stored readonly data, puts those into a huge neural network to perform processing to generate the next output token.

The "state memory" in an LLM (=the context window) is a lot more limited than a CPU RAM+disk, but it's still a state memory.

So I don't have a problem imagining that an LLM can perform some level of reasoning. Limited and flawed for sure, but still a different creature than "pattern matching".

> LLMs don’t really reason

Do you have a test for this?

Or is it based on the presumption that reasoning skills cannot evolve, it can only be the result of "intelligent design"?

It made me want to go back and check their other press releases to see where they did not include the "consenting" part.

It'd be interesting to try it as a regular keyboard too. Probably a bit too imprecise even for skilled users, but who knows.

And if you really get into them, maybe it's worth spending the time to learn chords instead, like for a court stenographer?

With that logic, we should put 100.00% of all public funding into protecting us against world ending asteroids.

You might be right that it wasn't GCC specifically. I think I was running a free/low cost version of OS-9 68K for the Atari ST, and managed to get hold of a proprietary C compiler for it. The combination of a "real" OS + C compiler made it possible to port the popular Unix text games of the time (empire, nethack, larn, etc).

My point was more that the 68K home computers like Atari ST and the Amiga of the mid-to-late 1980s were powerful enough to "run emacs / C compilers", even if it wasn't the GNU stuff yet.

Not apples to apples.

Just being near a potentially aggressive bear is a bad situation.

Being near a car or a nuclear power plant is not a bad siuation.

Of course you should run in the other direction if you're close to a potentially aggressive bear.

Same thing if you're in the path of an out of control car, or near a nuclear power plant accident.

But you've got to separate the "probability of bad situation occurring" from the "severity of the bad situation when it does occur".

By your logic, it's wrong to spend many billions per year on traffic safety. It should go into bear safety instead.

"Average deaths per TWh produced" is a good yardstick to me.

> What Stallman did a decade later was great if you happened to have access to the type of computer that could run Emacs. Even then, you probably didn't own the machine and maybe even had to pay for time on it by the hour. The small machines that ran Microsoft Basic were in people's homes.

No, a decade later was at the end of the 1980's. At that time, many middle class families could afford home computers like the Atari ST, which could run many of the GNU compilers/tools.

It was a great learning experience porting Unix apps and games to the ST.

Note that the inaccuracy comes from determining where the lightning strike starting point was from the waveform shape.

The raw position accuracy at 100 ns is better than 10 m.

The Swedish airforce doesn't have a "grouping of vehicles".

They are all parked individually at separate locations kilometers away from the road landing strip.

When the plane comes in for landing, they (3-5 pickup trucks and a tiny tanker) all scramble and meet up when the plane touches down. Refuel and rearm in 10 minutes, drive away again.

I've seen this done with my own eyes, it's very impressive.

> Does ChatGPT ever ask the user, like, anything?

Yes. At least when I just tried ChatGPT-5:

Can I replace sodium chloride with sodium bromide?

ChatGPT said: Yes, in some cases — but it depends on the application.

Chemistry/lab use: Both are salts of sodium and dissolve similarly, but bromide is more reactive in some contexts and heavier. It can change reaction outcomes, especially in halide-sensitive reactions (e.g., silver halide precipitation).

Food use: No — sodium bromide is toxic and not approved as a food additive.

Industrial processes: Sometimes interchangeable (e.g., certain brines, drilling fluids) if bromide’s different density, solubility, and cost are acceptable.

What’s your intended use?

An LLM is very much like a CPU. It takes inputs and performs processing on them based on its working memory and previous inputs and outputs, and then produces a new output and updates its working memory. It then loops back to do the same thing again and produce more outputs.

Sure, they were evolved using criteria based on next token prediction. But you were also evolved, only using critera for higher reproduction.

So are you really thinking, or just trying to reproduce?

Of course you can gerrymander your way to power. You can turn a fairly small minority in votes to a majority of the seats.

Professor Braunschweiger explains it well in this video: https://www.youtube.com/watch?v=E2EnuHsRJd4

CFR is usually higher than TFR

> it [...] suggested (assertively but correctly) where the problem was

> it was a decent, if probabilistic, rubber duck

How is it a rubber duck if it suggested where the problem was?

Isn't a rubber duck a mute object which you explain things to, and in the process you yourself figure out what the solution is?

(non-sarcastically from me this time)

Sure, I would agree with that wording.

In the same way, neural networks which are trained to do a task could be said to be "designed" to do something.

In my view, there's a big difference in what the training data is for a neural network, and what the neural network is "designed" for.

We can train a network using word completion examples, with the intent of designing it for intelligence.

> "The crazy thing is that people think that a model designed to"

It's even crazier that some people believe that humans "evolved" intelligence just by nature selecting the genes which were best at propagating.

Clearly, human intelligence is the product of a higher being designing it.

/s