> Types are groups of related values.

This is an operational perspective which misses one of the most important aspects of types: in a statically typed language, types are a way to classify terms, i.e. they're not just about values.

As an example of why this distinction matters, consider this expression: 1 + "hello". In most statically typed languages, this will be rejected at compile time. It doesn't have to be evaluated (i.e. at runtime) to be rejected - it's invalid because the types of the terms - 1 and "hello" - aren't compatible.

This compile-time rejection is only possibly because types apply to terms, and terms aren't values. This is critical to the usefulness of statically-typed languages - without this, they'd have to run the program to detect type errors.

Benjamin Pierce's book "Types and Programming Languages" has some great coverage of this topic.

I’ve done some NGO-related work and in that context, they talk about “appropriate technology” - technology that’s suited to the local population, among other things. I’ve started to see tools like Helm and languages like Go as appropriate technology for many teams, but that doesn’t mean I’m going to switch to a pedal-powered water pump in my own house.

This is not correct. Terms are syntactic expressions in a program. For example, 1+2 is a term, but it’s not a value. At runtime, it evaluates to 3. As such, some terms correspond directly to values - like 1 - but other terms represent computations that will be performed at runtime.

tl;dr: terms are a syntactic concept, values are a runtime concept. Terms evaluate to values.

For example, here's one that's described as "the first operating algebraic compiler": https://en.wikipedia.org/wiki/Laning_and_Zierler_system

I had a similar college experience, but there was a mainframe at our university. We submitted Fortran code either written on paper designed to be OCRd (write each letter in a separate box), or later on punch cards and got a printout back a few days later, sometimes as long as a week. It was a strong incentive to make sure there were no bugs.

At this point it's not clear that there's anything other than a 14 year old in him. A greedy, sociopathic 14 year old.

At this point, I don’t agree. I’m an immigrant living in the US currently. From an infosec and real-world consequences perspective, I have much less reason to fear what the Chinese government or corporations could do with my information than what either the current US government or its major tech corporations might do.

> Israel still have the reputation of being part of the ‘free world’ in the collective mind.

That collective mind is cognitively challenged and extremely ignorant, but it doesn’t mean I have to be.

If someone comes up with an interpretability result based on analyzing a neural net algorithm, they'd want to provide some evidence that the result applies to actual executing models.

Happily, we already have good evidence that inference non-determinism doesn't generally invalidate results obtained through analysis of the algorithm. That's not to say there couldn't be pathological cases, but in general, this isn't something keeping researchers up at night.

It's the highly non-uniform nature of the distribution before sampling that allows the LLM to produce coherent results. That non-uniformity is the result of complex dependencies in the model weights, which allow it to produce context-dependent sequences.

Since you're being snarky about asking an LLM, I'll note that your comment above is closer to gibberish, for the reasons I've described, than anything a frontier LLM would produce. If we were to extrapolate from this one data point, we might conclude that an LLM is a more trustworthy choice to flash a router than a human that produces such comments.

> ... she was protesting the use of delivery robots because she believes they are replacing human workers.

I mean this with zero snark: if you’re interested, you’d be best off asking an LLM or doing a search. But the overall idea is that Brandom described how concepts and meaning are created socially by a kind of game in which people keep track of the claims that other people commit to.

If you accept Brandom’s framework, then it can be argued that LLMs can’t meaningfully participate in this process. This could be considered an explanation of why people have a negative reaction to LLM involvement in activities that are meaningful to humans - even though most people won’t be familiar with the theory, they would instinctively recognize that something important is lacking, such as commitment over time.

Its name stands for “Formula Translation,” after all.

> I guess this was revolutionary at the time.

Yes. There was some prior art, but Fortran was the first “mainstream” (for the time) language to offer anything like this.

You probably know that non-determinism in real-world inference comes from deliberately stochastic features like temperature, from implementation choices like distributed and parallel computation, and numerical effects like floating point precision.

The fact that a neural network's execution can be nondeterministic doesn't mean that the network stops being an algorithm.

The point you're making boils down to speculation that the accidental non-determinism involved in real-world inference is significant enough to undermine or negate interpretability findings based on the algorithms. That would need to be demonstrated.

The CGoL and BB examples demonstrate what might be possible in a worst-case scenario, but they're only analogies in this context. It's far from clear that real models suffer from such effects to a significant degree. You might actually be better off talking about the butterfly effect!

If that's what you took from the article, perhaps AI writing is just the thing for you.