somewhereoutth

It is a common trope that tech people have a good understanding of computers, but a bad understanding of people. I see no evidence here to dispel that trope.

The best way to understand a loan is as the right to a future income stream (principal repayments and interest). The original debtor (the person/entity taking out the loan) establishes the credibility of that future income stream (based on income, expected returns on a project, etc) and sells it to the lender (usually a bank) for cash up front. Thus the loan is an asset on the bank's balance sheet, that is generating returns (assuming all goes to plan). Banks can and usually do sell on that asset to other parties.

Conversely, when you deposit cash to a bank, you are actually creating a liability on the bank's balance sheet - as you might want your money back one day!

I've often said (at least half jokingly) - C for systems programming, Javascript for everything else!

This artical makes the case for C as a low level language, I then offer Javascript as a functional-ish language for manipulating domain model data structures on the top.

Well, absent further interventions - possibly even a real invasion! - there is no reason for the current regime in Venezuela to change its policy very much (aside from beefing up its air defense maybe)

Probably we can distinguish between military invasion, and the use (arguably miss use) of the word colloquially in the US for home intruders.

Per the Cambridge Dictionary: "an occasion when an army or country uses force to enter and take control of another country"

In our case no 'taking control' has occurred, so this is not an invasion.

Not really - an invasion implies holding ground, which isn't the case here.

Of course that would be great, but pretty unlikely with just a decapitation strike. Like most dictators, Maduro was not holding the country in a superhuman iron grip, but instead the representative of various elites and factions that kept him in power for their own interests. However given how easy this operation has been, there is a suspicion that one or more factions colluded with the US, and may now be consolidating control - and then maybe a peaceful transition back to democracy? We shall see.

As far as I can ascertain, there is no invasion. Just a special ops kidnapping.

It is unclear what will happen next, but likely the regime or large elements of it will survive. Perhaps a more moderate faction will take control? That would be the best case scenario.

Presumably Venezuela has a continuity of government process, so the VP becomes head of the government? I don't see any indications that the regime has or would relinquish power, unless (more amenable) factions within it take advantage of this opportunity (perhaps this has been already coordinated).

Maybe the Venezuelan government remains intact, and now we have the embarrassing situation of a sovereign country asking for its head of state back.

depends if they can monetize that spot. So either ads or subscription. It is as yet unclear whether ads/subscription can generate sufficient revenue to cover costs and return a profit. Perhaps 'enough ads' will be too much for users to bear, perhaps 'enough subscription' will be too much for users to afford.

I spent a couple of years in the States before returning to Europe. I really had no idea what happened to people - what they became - if they weren't fed and watered properly. I had always assumed that the way people were was just how people are. Scary how much we take for granted.

> generated code can be verified by compiling and running it

I think this gets to the crux of the issue with LLMs for coding (and indeed 'test orientated development'). For anything beyond a most basic level of complexity (i.e. anything actually useful), code cannot be verified by compiling and running it. It can only be verified - to a point - by skilled human inspection/comprehension. That is the essence of code really, a definition of action, given by humans, to a machine for running with /a prior/ unenumerated inputs. Otherwise it is just a fancy lookup table. By definition then not all inputs and expected outputs can be tabulated, tested for, or rewarded for.

Not just their usefulness, but LLMs themselves are worse than an illusion, they are illusions that people often believe in unquestioningly - perhaps are being forced to believe in unquestionably (because of mandates, or short term time pressures as kind of race to the bottom).

When the ROI in training the next model is realised to be zero or even negative, then yes the money will run out. Existing models will soldier on for a while as (bankrupt) operators attempt to squeeze out the last few cents/pennies, but they will become more and more out of date, and so the 'age of LLMs' will draw to a close.

I confess my skeptic-addled brain initially (in hope?) misread the title of the post as 'Reflections on the end of LLMs in 2025'. Maybe we'll get that for 2026!

Yes - in the UK we had a strong social housing sector, with award winning architecture (and some terrible mistakes too).

Then along came 'right-to-buy', allowing tenants to buy their social housing for knock down prices (and so become a natural Tory [right of centre party] voter).

If councils had been allowed to use the money to build more social housing, then maybe this was fine. But they were not. So now we have affordability issues in the UK too.

My point is that socially beneficial capital intensive projects require, well, a large lump of capital. This can be organised by the state (command economy), or through allowing private accumulation of capital - but here I do include corporations, and indeed financing through loans. It would be quite possible for individuals to have much the same wealth (perhaps modulo time to retirement?), and to be pooling it as share owners of said corporations. This would still be a capitalist system - but a fairer one maybe!

Unfortunately a lot of the hype around LLMs is that their capability is humanness, specifically that they are (much) cheaper humans for replacing your expensive and annoying current humans.

"the accumulation of capital by some to the exclusion of others"

This allows decentralised decision making for large grained resource allocation - for example should we build a factory for shoes, or for toothbrushes? - and is a good thing, as central planning has been demonstrated to not work if applied to the whole economy. (the converse, no central planning to any of the economy, has also been demonstrated to not work!)

However that accumulation can be (and nowadays usually is) orchestrated by a corporate entity, which in an ideal world would be almost entirely beneficially owned by retirees on an equitable basis.

What has gone wrong, is that the benefits of productivity enhancements (since 1970?) have flowed to capital more so than to workers - which not least prevents them from forming capital themselves (savings/pensions), hence rising wealth inequality.

> repay the cost of training

Key point. Once people realize that no money can be made from LLMs, they will stop training new ones. Eventually the old ones will become hopelessly out-of-date, and LLMs will fade into history.

and would it be true to say that regenerative braking on electric cars reduces significantly this dust?

However the world we live in is not ideal, and wages are being systematically crushed by capital.

The AI bubble is causing large scale resource mis-allocation. That means that resources that could and should have been allocated to useful things have instead gone into data centre build outs and the panoply of marketing, executive hot air, forced/frivolous use of LLMs etc etc - the list goes on.

This means that society as a whole is perhaps significantly poorer than if LLMs had been properly valued (i.e. not a bubble), or had simply never happened at all.

Unfortunately it will likely be the poorest and most vulnerable in our societies that will bear the brunt. 'Twas ever thus.

C for systems programming, javascript for everything else!

The most commonly used/accepted foundation for mathematics is set theory, specifically ZFC. Relations are modeled as sets [of pairs, which are in turn modeled as sets].

A logician / formalist would argue that mathematics is principally (entirely?) about proving derivations from axioms - theorems. A game of logic with finite strings of symbols drawn from a finite alphabet.

An intuitionist might argue that there is something more behind this, and we are describing some deeper truth with this symbolic logic.

Actually it kind of is, in as much as it expands the money supply.

When a bank issues debt, the money is created 'out of thin air'. When the debt is paid off, that money is destroyed. However usually more debt is being created than redeemed as things go on, so the total money supply increases (this is a good thing, as it allows the economy to expand).

Various regulations and central bank market interventions (quantitative tightening/easing) control this process, which thus can be induced to 'print money' if the government wishes - assuming they have a sovereign currency.

Yet another [1] AI winter [2]

[1] https://en.wikipedia.org/wiki/AI_winter#The_setbacks_of_1974

[2] https://en.wikipedia.org/wiki/AI_winter#AI_winter_of_the_199...

However 'normal people' buying into the stock market via 401ks or otherwise likely (and arguably sensibly) will be in index funds, that of course are exposed to the bubble via (grossly?) inflated tech stocks. Effectively their current pension/savings contributions are being clipped by whatever the delta is between now and post bubble. Time in the market and all that, but still it might be a hefty haircut.

LLMs: The chemical weapons of public discourse.

The cleanup is going to be a grim task.

Although INTERCAL's "come from" provided the opportunity to implement a gloriously concise multi-threading syntax/mechanism [1]!

[1] https://esolangs.org/wiki/Threaded_INTERCAL

1. Understand that code that has been wholly or partly LLM generated is tainted - it has (in at least some part) been created neither by humans nor by a deterministic, verifiable, process. Any representations to its quality are therefore void.

2. Ban tainted code.

Consider code that (in the old days) had been copy pasted from elsewhere. Is that any better than LLM generated code? Why yes - to make it work a human had to comb through it, tweaking as necessary, and if they did not then stylistic cues make the copy pasta quite evident. LLMs effectively originate and disguise copy pasta (including mimicking house styles), making it harder/impossible to validate the code without stepping through every single statement. The process can no longer be validated, so the output has to be. Which does not scale.

Is this about freedom to vs freedom from?

The first freedom of course is freedom from fear (e.g. from fear of being snatched from the street just because you are brown and speak with an accent)