agentcoops

It can be a mix of both. There's no reason, after all, that humans no longer being the most intelligent systems on the planet has to mean giving up such a passion, especially since it means living in a world with more and not less mathematics. One has to separate the joy of doing mathematics from a capacity (or, more importantly, need) to obtain a living doing it from the pride of being the best in your field.

It's interesting to recall Turing's own position: "There would be great opposition [to AI] from the intellectuals who were afraid of being put out of a job. It is probable though that the intellectuals would be mistaken about this. There would be plenty to do, i.e. in trying to keep one’s intelligence up to the standard set by the machines, for it seems probable that once the machine thinking method had started, it would not take long to outstrip our feeble powers. There would be no question of the machines dying, and they would be able to converse with each other to sharpen their wits."

On the other hand, Thomas Bernhard's The Loser is a horrifying account of the opposite standpoint---of an essentially second best piano student who cannot stand existing in a world with Glenn Gould.

I’ve been doing really heavy text analysis work with LLMs where false negatives/misses are important to minimize and my god did I hit cost thresholds quickly with 5.6 Luna — it was the first time I felt motivated to seriously work with local open models, even if inference was degraded for the task. Cheaper and much better inference now brings me back to the closed models for better or worse.

It’s more than that, I think. I read through the various government statements as I was curious about rationale and precedent: the other two groups impacted by the order are rather unambiguous extremists, so understandable. In the extended justification for this a/I group, the US gov lists a large number of US and EU “Antifa” groups that variously leverage their services, including one that the individual recently arrested for wiping his phone is meant to be connected to. It suggests to me this was the result of some considerable network analysis and that the intent is to now have stronger legal tools to go after those groups.

I think that second (speculative, of course) step would be the more worrying precedent: if the use of technical infrastructure also used by designated groups was formally made to be incriminating/right-waiving. Why wouldn’t that apply to Signal, for example? The OP suggests the dividing line should be between true infrastructure open to all, including unsavory actors, versus tools that advertise services to a particular faction. It’s not an entirely disagreeable distinction, but I think in practice it’s rather fuzzy (Moxie of signal is/was rather far left himself, if I recall).

EDIT: Re-reading the document, the state department emphasizes that a/I “manually vets all potential users.” OP’s distinction stands for now.

I don’t think the effects on the hardware market of open weight models triumphing has been reflected upon enough. It’s not clear that it will be less impactful if every enterprise decides to build for predominantly on-premise inference. In fact, before the question is ultimately decided, we’re probably heading towards a few years where hobbyists, enterprises and ‘hyper scalers’ are all competing for certain parts in common. Ram booked through 2027 sounds about right.

Professionally, Scala was always my favorite language to work in and I was lucky to get to use it most of my career. It is, however, probably the worst language I’ve experienced using with LLMs. Next worst is any dynamic language: it’s just so hard to not introduce strange bugs after iterating on a large-ish project across multiple agent sessions.

I’ve had good enough experiences with Rust, but actually OCaml has been hands down the language I’ve seen best results with. The quality (and performance) of code is just phenomenal — and the main issue when working as a solo human with the language, namely smaller pool of community libraries, just isn’t an issue any more. Jane Street has really done tremendous work modernizing the language and tooling.

This is fascinating. I haven't really kept up with SEO for years, but was recently helping an academic publisher setup their web presence and ran into exactly this issue: Google was/is refusing to index the actual journal articles on the new site and they were moved into this liminal "Crawled, not indexed" state before vanishing entirely. Comically, it's now easier to get your content indexed and served with a visible link by ChatGPT than it is Google...

In finance and so the world we live in, the value of an equity is less related to the merit of the product than the firm's capacity to generate future free cash flow. Software companies, B2B SaaS in particular, have been basically unbeatable in this regard, hence the state of the market (and our salaries) for the past few decades. Industrial firms have to use metrics like "EBITDA" to show how much cash they'd have to potentially pay to investors if they didn't have to pay so much in interest on their debt, taxes, and fixing up decaying equipment...

I used to work in this field myself long ago (academically and practically). The difficulty that OP didn't suggest much appreciation for is that the interesting spaces for automated verification -- especially of distributed systems, say -- require exploring the space of formalisms between the trivial proofs of propositional logic (SAT solvers) and the undecidable proofs of first or second-order logic (interactive theorem proving environments such as Isabelle and Lean). This is the incredibly large space occupied most importantly by the various modal logics, which allow for quantification over graph traversals. There are remarkably expressive logics where the decision problem remains tractable, such as the modal mu-calculus.

That is, precisely if one wants automated tools for verification the seemingly obtuse work on novel formal systems is necessary.

Yeah, it's been a relatively rational decision for a while now, but one I personally hadn't really seen taken until recently. Again, all anecdata, but I am curious to track how much of a trend it really is.

Yes, it's never a trivial process, so a lot of work is certainly being performed by that "relatively": I have extensive personal experience with the UK immigration process and know of the US equivalent through the experiences of former colleagues back home. France, for example, is five years to a passport/naturalization. Germany is three years of skilled work to indefinite leave to remain. The Netherlands is five years to indefinite leave to remain. None of those examples require European ancestors, EU spouse etc, but generally it's easier if you have a university degree and work in the various fields most readers of hacker news do.

Generally continental Europe -- except the Scandinavian countries -- makes it relatively easy to get long-term residency and even a passport. The UK is considerably more difficult, but very easy to work in for an extended period of time (intra-company transfer visas etc).

If you're living and working in California or New York, as I suspect a large number of hacker news readers are, EU taxes on income are generally not prohibitively more expensive, especially relative to increase in quality of life. 'Native' salaries are considerably lower, however, and tax treatment of equity-based compensation is very much not in favor of employees...