skew-aberration

How do you prune out the 'fails to prune out behaviors' behavior? If my model is failing to prune it out, can another model prune out that behavior?

On the other hand, symbolic reasoning easily gets superhuman results in narrow domains. The Robbins Conjecture was proved on hardware like that after 60 years. You could check out this theorem prover benchmark suite that has been maintained for over 30 years https://www.tptp.org/

I grew up with these machines around, and as a kid they seemed to so much more powerful than humans. It's kinda weird to think that I became 'smarter' than them eventually. I suspect they will be able to run superhuman AI one day too.

What's your setup? Flash in opencode requires far more oversight for me than Opus 5.5 in claude code.

> I don't want to relitigate Malthusianism

But you do want to assert that it is true, without being aware of its meaning?

Malthusianism is the notion that some people must be poor / hungry because humans choose to reproduce until they reach the limits of the environment to provide for them (especially those Irish potato farmers, according to Malthus), and that increases in productivity will not change this - they will just be absorbed by a growing population (which ought to be 'swept away').

In reality, growing populations have led to more than adequate improvements in food production efficiency and people are choosing to reproduce less rather than more.

There is an excellent summary and rebuttal of Malthusianism in the book 'Progress and Poverty' by Henry George, which offers an alternative view of how supply and demand for finite land (your words) rather than reproductive instinct (Malthusianism) creates poverty.

Deepseek 4.1 flash is so fast and cheap. For coding tasks, it only seems to lack the robustness of Opus 5.5 rather than 'intelligence' - it fails to double check critical assumptions, stops before the task is over, and doesn't do enough speculation or exploration on its own initiative. Yet it can do all these things when prompted.

There seems to be another magic ingredient for improving agents beyond model size & efficiency, and so far only OpenAI, Anthropic, and maybe Google know what it is.

It's ambiguous, because it lets the reader think for themselves about the implications of a straightforward fact?

If you do insist on being spoon fed - getting information about labor relations from Fortune magazine is like taking heath advice from a Luck Strike promotion. The article's cheery conclusion is: Yes, AI productivity is driving wages down - but we only need to be patient and wait for more jobs to replace them, not tax or regulate AI. Yes, unemployment is driving wages down - but more immigration will fix this, not make it worse. Yes, the deficit is high while corporate profits are up - but the solution is more jobs (how?) rather than more tax. Yes, smoking causes 'throat irritation' - but Lucky Strike is toasted.

I know nothing about fentanyl honestly, but my impression from heroin users at least is that it really is that good, and issues like addiction and withdrawal are somewhat overblown - many occasional users completely avoid them.

This seems obvious in current year, but extrapolating it I wonder where it takes us.

If society isn't there to keep us safe, free, and happy - what's the point? Accumulating lean certificates and energy production capacity?

There's already people kinda living like this. Fentanyl is abused because of the intense euphoria the users experience. As tech improves the downsides will go away and the highs will get even more powerful.

A quick search suggests that modern gallium extraction is not particularly toxic, and is no worse than the Bayer process itself. China uses ion exchange technology that has never been successfully commercialized in the west. i.e they invested in making clean methods competitive, about 20 years ago. In fact, they have monopolized much of the ion exchange supply chain.

https://payneinstitute.mines.edu/strategic-intervention-to-r...

> China produces approximately 98% of the global gallium supply and 88% of the global germanium supply; dominance has been fueled by government intervention. Beijing mandated its rapidly expanding aluminum producers to install the capacity to extract gallium. Between 2005 and 2015, China’s production of low-purity gallium surged from 22 metric tons to 444 metric tons, a nearly 2000 percent increase. This move flooded the market and forced producers in the United Kingdom, Germany, Hungary, and Kazakhstan to shutter their operations.

If it was marginally economic for a time while the price was 1/10th of the current price, it must be economic now. This appears to be yet another case of China undercutting, creating a monopoly, and then jacking the prices up.

Anyone care to explain the downvotes? I am simply pointing out a glaring misuse of a word in the title, which made me curious about the author's background.

Capital historically meant 'that part of production set aside for future production'. Machines, land improvements, infrastructure, intellectual property. Now it tends to include all assets used to generate wealth - including things like land which were once in their own category.

Wages are just money and may or may not become capital according to the discretion and opportunities of the employee.

Capital != money. Confirmed that author does not have finance, economics or business background.

Investment in decarbonization is vast and has been for decades. Even in Trump's America.

The US, a nation in huge debt, is committed to around $1T in tax subsidies and $500B in loans and other funding for clean energy in the next decade. Private capital investment in that same time will account for another several trillion - driven by these subsidies and the regulations which force industry to adapt (and because clean solutions are genuinely cheaper in many applications).

The government tax subsidies and direct spending (depending on tax projections) are on the order of 5-10% of the discretionary budget (around 2-3 times the annual 'extra' cost above baseline of assisting Ukraine or bombing Iran Feb-Oct which are each on the order of $50B). It's hard to get an exact number of course, since 'funding which decreases carbon emissions' extends way beyond clean energy and electrification. For example, Defense Production Act may be used to fund manufacturing or refining capacity in key areas.

That's a long way from 'refusal' and represents only one country.

Public and private combined, the EU is investing at about the same rate as US, China is investing on the order of about double the rate of the US.

These are very hard problems to solve.

I'm all for decarbonization, but 'refuse' is doing a lot of heavy lifting. These are very hard problems with a lot of people working on them. Dropping fossil fuels today would kill people too.

Submit a wish and get a video of the next locally visible shooting star

Ok if you define message passing that broadly then I would agree, but I think that definition is also broad enough to bring in the downsides of shared memory such as deadlocking.

We're also now talking about patterns which are common in other contexts, but I don't have a problem with that. io_uring for example - is it shared memory, asynchronous, message passing, or all three (probably yes)?

I'm no stranger to this but there is nuance. Thoughts on my other comment below?

Yeah, there's plenty of high-profile academic politicians in Europe with debatable legacies including Thatcher and Merkel. Time will tell.

A quick search suggests the economists associated with this movement are fairly 'mainstream' institutionalists of the type that's popular with government these days in the west. I don't think there will be big changes soon beyond dealing with corruption/nepotism.

Later on, I imagine they will push for faster EU integration on possibly slightly worse terms. Serbia may become marginally less independent in terms of fiscal policy at least and the markets might become even more exposed/beholden to foreign capital. This process has been in motion for years now and I don't expect the student movement to care - the upsides are broadly popular with younger people globally. Personally, I think it is deeply ironic that in the age of tariffs and huge government spending by the superpowers, small countries like Serbia are still signing up to be bound by the 'free market' consensus which is rapidly disintegrating globally.

Every copy is a performance downside. There's plenty of applications where it is essential to use shared memory. You could not rewrite Linux to avoid shared memory between threads.

The line is fuzzy though.

You might say that each RAM access is really a message over a bus, and so is each cache access. Conversely, you could write a message passing style key value store and create race conditions by using like shared memory.

Transactional memory is also 'message passing' like -> you bundle up access of shared memory into a single transaction 'message'.

A lot of shared memory code uses constructions like FIFO/queues which are really not too distinct from message passing except when ownership is shared or goes back and forth. Perhaps that's just an 'implementation' of message passing.

It's interesting that we've gone from cooperative to preemptive to cooperative multitasking again.

I agree that blocking threads is generally more natural / obvious. Cooperative multitasking was always done for performance reasons and more recently for simplicity (in single thread implementations with no locking) and/or scaling reasons.

Message passing concurrency is going even further though, and never really seems to have taken off wholistically. Again, primarily for performance reasons. When we do pass messages (across networks, too/from RAM or disk, etc) we usually add caching and other optimizations which make concurrency even harder to reason about.

My reading of the above was that at the time of LZB going down, PZB could not be enforced immediately, because its assumptions were violated (multiple trains in the same zone). Hence the trains are stopped.

The question is then what does manual operation entail? Do they just continue operations, driving slowly and communicating through other channels while allowing PZB rules to continue to be violated, or do they have a protocol to re-establish PZB as quickly and safely as possible? Asking as a layman.

Surely a better fallback is possible. Like if a train is known to be in a PZB with other trains ahead of it, then when LZB goes down it has to slow down/wait/follow some different protocol until PZB rules can be enforced again.

It's so platform dependent though, for many FreeRTOS targets I think you're just gonna be integrating against a vendor specific stack

Yes. Pre-o1, it didn't seem to be improving as rapidly as expected.

Yes, it is speculative and subjective - but so is all value. You can't factor that out, because it is inherent to every transaction that humans ever make. We make decisions based on our own valuations of expected future outcomes.

Even the value of the bridge is not realized until people cross it. It takes decades for such investments to break even, if they ever do - far longer than a GPU (which probably costs less than a mere boardwalk across a ditch).

Accountants are doing exactly what they should - every asset and liability is broken down by its market value (speculative) which investors then use to inform their speculation about the value of the company. If they think the company is more valuable than the current market price, they buy more stock. If you think the market value of a GPU is too high, don't buy it.

The market can be (and frequently is) wrong - but it's not an accountant's job to predict that.

Also, even GPT-3 (6 years ago) was a very powerful search tool for exactly these kinds of problems. You needed some result that would give a bound on some obscure property, and it instantly gave you five different applicable theorems from three different domains of math. Usually, the theorems even existed.

I was taken aback by the progress in 2025/2026 because late 2024 did seem to be a plateau, but 5-10 years seemed inevitable.

Even when I first studied automated theorem provers in around 2013, the 10-20 years to automating proofs (not theory building) seemed inevitable. They were already superhumanly fast and generating real results, and neural networks were making massive progress on translation, image processing, and games. Someone just had to work out how to stick a neural net inside an ATP. The surprise was that it happened the other way around, and that it solved even more than just mathematical proofs.

AI has only just begun to expose the depth of willful human ignorance.

Yes, values are inherently speculative - the S&P is just measuring current market values though - something like the market consensus of the net present value (profits further in the future are treated as less valuable than profits today). As this consensus changes, the S&P goes up and down.