Sharlin

25%-80%.

Certain (mostly AI) workloads. Depends on how hot you're comfortable running your card. The real-world improvement is considerably less due to other things being bottlenecks.

That applies to NA English. And even there, as the reference says, notional agreement (verb/pronouns in plural) can be used when it’s more about the individual members of the collective doing something. So GP’s rule that only grammatical agreement (verb/pronouns in singular) is ever correct is wrong even in NA. But in BrE notional agreement is the most idiomatic choice while grammatical agreement is certainly also fine, just not standard style.

I’ve no idea what you refer to with "laziness". The "rule" that you claim you were taught simply does not always apply.

A smart writing assistant will give advice that’s appropriate to the context, register, the recipient’s culture, communication style and tastes, and the author’s relationship with the recipient. Sometimes writing very normatively and by the book is part of the message. Sometimes the exact opposite is true. If the assistant is unaware of all the context relevant to the communication, it’s likely to be a bad assistant.

Dropping articles is wrong because English as natively spoken doesn’t work like that. Except when it does, like news headlines, condensed "telegram" or SMS style, and so on.

In British English it’s entirely standard (and common) for grammatically singular collective nouns to take a plural verb and be referred to with a plural pronoun:

"The ManU have won three out of their last four matches."

"The Parliament have had heated debates over the NIH budget."

"Tesla introduced their newest EV model last week."

Any competent English teacher should know this.

Also, there’s no "correct" language outside of how it’s actually used. Linguists are the first to explain that their job is to describe, not to prescribe.

I’m not sure most physicists would make that specific distinction between "theory" and "model". We know QM is wrong or at the very least incomplete as long as there’s no theory of quantum gravity. Every theory, like every model, has a domain of applicability and gives wrong or even nonsensical answers when applied outside its domain.

What a theory is, is more like a model (or maybe meta-model) with a particularly broad scope or high explanatory value, often a mathematical framework that gives a common foundation and theoretical justification for multiple experimentally-derived models.

And whether it’s a positron or an electron (beta+ or beta- decay) doesn’t seem to matter that much as far as equivalent dose is concerned, the energy deposited is only fractionally higher in the case of a positron. (But PET can only work with positrons because it’s exactly the pairs of 0.511 MeV annihilation gammas that it’s designed to detect!)

I wish it were more often even generalizing from one case, that would be a big leap in epistemic hygiene compared to the more common generalizing from zero cases!

You mean they don't accrue knowledge during the training?

No, either you misunderstood or there was misinformation in the other thread.

There are two things that Google would’ve been allowed to legally do without an environmental review:

* Log the trees for economic use, then plant new ones on the same property, definitely not do terrain alterations. Like your average forest owner. But obviously that’s not what they wanted to do, they wanted to build a data center.

* Log less than 150 hectares, do any earthmoving etc needed within that area, build a data center.

These are mutually exclusive operations.

There’s no world in which Google would’ve been allowed to log the 300 hectares and do all the other preparative work on the property that they’ve done. They were told that they need a YVA (environmental impact review) process before doing anything beyond some very light reversible preparations because the area they wanted to modify exceeded 150 ha.

Somehow, even with the ability to hire the best lawyers in the country, they either fucked up and misunderstood the relevant legislation and the quite clear communication from LVV, the govt office responsible, or they just ignored all that and reasoned that they can get away with it because they’re big and – unfortunately – there’s no sanctions specified in the law for skipping a required env review.

There’s this thing called rendering, pretty important for maps.

Honestly this is starting to feel like a "God in the gaps" argument.

Not nearly enough time has passed to make any conclusions about long-term maintainability of LLM-written code.

People nowhere pay for the full external and opportunity costs of cars, particularly ICE cars, in the purchase price (or even in fuel price), including carbon emissions, particulate emissions, noise, waste of space, perceived and actual safety, and so on.

Cars are literally an econ 101 textbook example of negative externalities.

Two-wheel vehicles are much more reasonable in all ways, I’m not arguing against them (although two-stroke ICEs are definitely terrible things).

Because recent events have abundantly shown that dependence of any US SaaS is a big liability.

The point is, obviously, that we should be decoupling from all gas. The past four years have made it very clear how risky it is to depend on foreign fossil fuels, economically and geopolitically.

Like Putin in Ukraine. But decoupling from Russian gas is still slow going and controversial in Europe, and the "moderate" right is more and more inclined to collaborate with Putin’s little fringe-right helper parties, likely partly because they’re massively bleeding votes and partly out of real sympathies for fringe policies that just used to be unmentionable in polite company.

Yes. Which is exactly the problem. If cars cause traffic jams - never mind their other externalities only some of which are fixed by EVs - then that just means cars are too cheap and their true costs aren't reflected in their price. Cars are a classic tragedy of the commons, everyone in traffic would prefer that there be less traffic so they could get to places faster.

Cars make themselves not-fast by their sheer inefficiency unless huge amounts of taxpayer money is continually poured to the infrastructure. What’s incredible is that cars have been getting even larger, an example of astonishingly perverse incentive structures.

Thanks for your opinion, we’ll give it all the attention that it deserves.

How can you say that? Even if their future livelihood were 100% guaranteed at the same or even higher level, the loss they feel would be the same. If you don't understand that, you probably just aren't empathetic enough.

Your employer probably would, and Anthropic as well.

—

In any case, these Netflix and LLM subscription comparisons are utterly beside the point, the equation is hilariously different when it’s a surveillance tool used by the state violence monopoly that we’re talking about. The standard should be absurdly higher than "no big deal, everyone shares their Netflix".

Honestly I’m not even surprised.

That’s exactly the fear: that math becomes just a meaningless game. Us mere mortals can get enjoyment from solving math puzzles whose solutions are well known, but it’s rather different being in the frontier, discovering entirely novel mathematical structures and proof techniques – and answers to questions nobody knows how to answer yet, sometimes even questions nobody has thought to ask yet!

> If what you say was true, many of them would be recluses up in the mountains, but none of them are.

Except, for instance, Wiles and Perelman? And I’m pretty sure that mathematicians are rather more introverted and solitary on average than most people.

In any case, that just doesn’t follow. Mathematicians realize that collaboration is important in math just as it is in most other fields. Humans are also social animals and many of us enjoy collaboration and social interaction for its own sake. And yes, that means social status is a thing that motivates many humans. But nobody decides to study math in order to gain status! Any status stuff is entirely contextual to and conditional on having already decided to study math.

I don’t think it’s about job security in the case of professional mathematicians. It’s about the fear of giving up the very thing they’re passionate about. It’s not like math researchers work on pure math because of the pay. If anything, they do it despite the pay.

Defining "agent" as just the harness makes about as much sense as defining "human" as just the rest of the body without a brain.

Price of chicken dropped steeply (and consumption exploded) in the second half of the 20th century after the introduction of broiler breeds and what’s essentially muscle tissue factories. Before that chicken was neither particularly cheap nor eaten that much.

We’re not closely related to dogs as far as mammals go. Indeed we’re closer to rodents and lagomorphs, and our last common ancestor with dogs and other carnivorans likely lived before the K–Pg extinction event.

To be fair, sealed classes were introduced in Java 15 back in 2020. But granted pattern matching is what makes them really useful.

The Finnish high/secondary school (= years 10 to 12) gives a choice of two alternative math curricula, the more comprehensive option giving a reasonable foundation for university-level math and is usually very much recommended to anyone interested in STEM. It includes complex numbers (at least superficially) as well as a lot of other stuff that the narrow version doesn’t, like limits, integration, basic linear algebra, and analytic geometry.