Economics Nerd Central

[How bad boy]

The issue is it thought a millimetre cubed was a millilitre.

You don’t need the right training data sets for that, that’s just objectively wrong.

Crude but I’d just discount any technology getting something basic wrong like that.

The cell count itself is first year biology. When I was in first year I’d open a first year biology book and find that out using the index. I thought Google had solved finding that kind of thing without going into Boole over 20 years ago - this is a step backwards.

The terrifying bit is this is where I know the answer is wrong. What if it’s something I ask it about which I know nothing about it? Which is most things.

It’s a good tool for creating more convincing bullshit artists. Like the world hasn’t enough of those already.

The model has no sense of what is objectively right. At the core of it is an LLM system that's just chunking words up into tokens, pre-processing them, then pushing them through a neural net to predict the next most likely token.

It does, without a shadow of a doubt, exhibit emergent intelligence.

And the speed of improvement is phenomenal, so it really matters which model you used that gave that answer.


You can make them look stupid easily.
The easiest one until about 6 months ago was to ask how many Rs are in the word strawberry. Almost all used to tell you there were 2, because the chunking up into tokens would break it into something like "Straw" as a single token and "berry" as another token. The exact composition of those words is effectively compressed out, so it would then only count 2 tokens associated with R.
They get it nowadays as they've put systems around the models to compensate for that sort of case. And the systems around the likes of ChatGPT Pro are a lot better than those around ChatGPT free. Same with Copilot free vs paid enterprise with ChatGPT5 enabled.

The benchmarks that were previously created, such as MMLU are now barely relevant, most good models get it 85% correct.


Humanity's Last Exam is the main benchmark people are tracking (albeit one might argue it's waaay too focused on science * maths), but even the progress there is amazing, GPT-4o could only achieve 2.7, GPT-5 is now up to 25.3%


I'm sceptical whether or not it'll actually go into human type of intelligence, I don't see how a token prediction engine can be particularly creative when it's trying to output the most probable answer*, but with the current rate of progress, would not rule it out entirely. Additionally, few people are actually creative for the vast, vast majority of their life, so if they can improve the accuracy and mitigate hallucinations, then there's still plenty improving to be done

It's amazing that simply using something that predicts the next word is actually as intelligent as it is, for all its flaws.



*Philosophers such as Wittgenstein have argued that while lanugage is not everything, it is one of the fundamental basis of our knowledge and understanding. Dictators such as the Soviets (and as a result, Orwell) understood this well enough, that if you ban the expression of an idea then you suppress the ability to express that idea long term, and thus you suppress the idea, the inverse also being true, that ideas arise from language. That if you can capture our language, you can capture our intelligence. Clearly, it's only partially true...

 

[How bad boy]

Like I said, the main source of enlightenment has been people think putting words together in a pleasing order constitutes intelligence. It explains a lot actually.

That's basically the entirety of literature

 

[How bad boy]

Literature is creative.

I’m talking documentation/technical writing/legal.

That’s the area I’ve been most impressed by AI using it.

Which makes sense given it’s a word calculator.

Yep, a very, very impressive autocomplete.

 

[Bennyton]

Back to economics.

Prompted by the BOE keeping rates on hold, been keeping an eye on the UK bond yields, because right now, the UK is paying a lot higher in interest payments than most equivalent countries.

Ireland's 30 year bond yield right now is 3.64%. The UK's is 5.2%

Ireland's debt to GDP ratio is now down to 40%, so debt interest payments are about 3.3% of total government spending.

For the UK, that's 8.3%.

To put it in perspective, if the UK had Ireland's level of debt interest spending, then the UK government would have about £60 billion more to spend than it does now, which would solve all of Rachel Reeves' problems.

If it had Ireland's bond yield, then I suspect the immediate pain would be somewhat lower. That interest rate is not necessarily representative of what the UK pays overall for its interest rate, it's the 30 year rate. The 10 year rate is 4.45%, but Ireland's is 2.89%

That 30 year rate has gotten a lot worse over recent years though, going from ~ 4.5% when labour took power to its current level:
View attachment 45576

Its a hell of a lot worse than 5 years ago:

View attachment 45577



As a result, the UK is in a bit of a bind. If it doesn't get its house in order, then that rate will climb, costing taxpayers more.

This is the bind they're in.

15 years of slow gradual austerity, very little investment for much of that time, besides Johnson spending a few pounds on hospitals, and crossrail.

You have the obvious contributing factors:

Not taking their licks like they should have in 2008, and dragging it out for too long.

Brexit caused lots of productive Europeans to leave, to be replaced with economically lower educated, and less productive people. (I'm speaking economically here, not trying to call out race or whatever.)

The pandemic has left a lot of people not arsed to work as much, long COVID, trauma from the pandemic etc, and mental health is now a ticket to decent social welfare, so quite a few people are taking that offer up.

HS2 was an absolute disaster too, just executed so badly at every step, unnecessary bat tunnels, trains to nowhere , etc, they tied themselves in knots with British regulations, and flip flopping, only to spend 60bn on something that is of no value.

But anyway, you've got an economy running "dry", capital flight, talent flight, and a growing deficit.

The only way to reduce the deficit is growth, or tax more.

If they tax more, then there isn't much fat there to trim, so it'll probably reduce the size of the economy.

If they try to borrow, then it might send interest rates soaring, like it did with Liz Truss.

Perhaps, 6 months ago, Reeves could have tried to grow the economy with borrowing, but I think the writing is on the wall for Labour now, to do one term.

All of the above is largely self inflicted, or exacerbated by British exceptionalism.

Going forward, they need to get off the cheap immigrant labour drug.

Education, particularly in secondary, narrows too soon, leaving people financially illiterate and not very well prepared for the world ahead of them, it also sends a lot of kids who aren't suited to college in, only to fail and walk away with debt and no prospects.

They need to raise the basic rate of tax to 23%, cut mental health PIP, cut motability.

Reduce legal and illegal immigration to about 100k per year for a decade or so (this reduces rent pressure, and causes a correction in the price of housing, reducing talent flight, leaving the economy on a better footing, long term, again, this isn't some mad racist rant, I'm an immigrant there myself).

Invest heavily in primary and secondary education.

Invest heavily in energy infrastructure and green energy.

But it'll be politically too difficult to do basically any of that, so they'll scratch around the edges, dump any savings into the NHS, and hand things over to reform next time around.

The term George Soros took from social science and applied to economics was Reflexivity (and a great example of why I don't think economics is a science).

Labour have set fiscal rules so that in theory, they end the "economic cycle" at the end of their time in parliament in surplus. These rules are supposed to give the bond market confidence.
Right now, the deficit is higher than they thought it would be at this point in the cycle. Forecasts of GDP growth are shite, spending is up, and those higher interest payments are making things much worse.
If they cut spending and/or increase taxes, then the budget deficit decreases, the bond market is happier, and interest rates reduce, which also reduces interest rates, improving the budget deficit.
Or maybe, if you cut spending and increase taxes, you damage growth prospects through negative fiscal multipliers, in which cases, your projections for future tax take drop, your budget deficit widens and the bond market punishes you for failing to create growth.

Or you say that in order to have growth, you accept a wider deficit and break the fiscal rules, the bond market goes along with it, interest rates decrease and your spending decreases, narrowing the deficit
Or the bond market doesn't, raises interest rates and makes your already worse deficit much wider, which will also knock business confidence, result in even steeper cuts to spending and taxes later on which further damages growth.


The fact that the attitude of the bond market to fiscal policy makes such a dramatic difference to the outcomes is why I think economics isn't a science. That's not really a mathematical model thing with falsifiability, it's far more of a vibes game. The maths is handy for analysing situations, but at the end of the day, it's all vibes that pulls at the big levers.

Anyway, one absolutely fun outcome of what you might think of as the lowest risk option, cut spending and raising taxes now, absolutely opens the door for populists and will be absolute poison at the next election, making a Reform government far more likely. Which is a far, far worse thing for growth and stability in the long term.

Which is nice.

Kyla Scanlon recently coined the term “vibecession” , vibe-recession, to describe the mismatch between strong macroeconomic data and weak public sentiment, that definitely sits with the feeling in Ireland and the US.

The money isn't trickling down, it's not hitting our pockets, and if it is, it's being swallowed up by expensive food, energy, rents and mortgage rates.

I think it's fair to say the Vibes are off, for most of the western world at the moment.

You have that feeling in the UK too, only the GDP is also fucked.

 

[How bad boy]

This is the bind they're in.

15 years of slow gradual austerity, very little investment for much of that time, besides Johnson spending a few pounds on hospitals, and crossrail.

You have the obvious contributing factors:

Not taking their licks like they should have in 2008, and dragging it out for too long.

Brexit caused lots of productive Europeans to leave, to be replaced with economically lower educated, and less productive people. (I'm speaking economically here, not trying to call out race or whatever.)

The pandemic has left a lot of people not arsed to work as much, long COVID, trauma from the pandemic etc, and mental health is now a ticket to decent social welfare, so quite a few people are taking that offer up.

HS2 was an absolute disaster too, just executed so badly at every step, unnecessary bat tunnels, trains to nowhere , etc, they tied themselves in knots with British regulations, and flip flopping, only to spend 60bn on something that is of no value.

But anyway, you've got an economy running "dry", capital flight, talent flight, and a growing deficit.

The only way to reduce the deficit is growth, or tax more.

If they tax more, then there isn't much fat there to trim, so it'll probably reduce the size of the economy.

If they try to borrow, then it might send interest rates soaring, like it did with Liz Truss.

Perhaps, 6 months ago, Reeves could have tried to grow the economy with borrowing, but I think the writing is on the wall for Labour now, to do one term.

All of the above is largely self inflicted, or exacerbated by British exceptionalism.

Going forward, they need to get off the cheap immigrant labour drug.

Education, particularly in secondary, narrows too soon, leaving people financially illiterate and not very well prepared for the world ahead of them, it also sends a lot of kids who aren't suited to college in, only to fail and walk away with debt and no prospects.

They need to raise the basic rate of tax to 23%, cut mental health PIP, cut motability.

Reduce legal and illegal immigration to about 100k per year for a decade or so (this reduces rent pressure, and causes a correction in the price of housing, reducing talent flight, leaving the economy on a better footing, long term, again, this isn't some mad racist rant, I'm an immigrant there myself).

Invest heavily in primary and secondary education.

Invest heavily in energy infrastructure and green energy.

But it'll be politically too difficult to do basically any of that, so they'll scratch around the edges, dump any savings into the NHS, and hand things over to reform next time around.




Kyla Scanlon recently coined the term “vibecession” , vibe-recession, to describe the mismatch between strong macroeconomic data and weak public sentiment, that definitely sits with the feeling in Ireland and the US.

The money isn't trickling down, it's not hitting our pockets, and if it is, it's being swallowed up by expensive food, energy, rents and mortgage rates.

I think it's fair to say the Vibes are off, for most of the western world at the moment.

You have that feeling in the UK too, only the GDP is also fucked.

This is where I agree on the principle of what Gary Stephenson discusses.
Government is heavily indebted (I'll come back to that)
The middle class in the west is going bust, the next generation of middle class adults simply can't afford to live independently. The lower classes are fucked with cost of living, but being a little better protected by the state, e.g. the minimum wage has gone up by almost 50% in the past 4 years.
The upper middle class are being squeezed pretty tightly, because that's where the easy money to tax is, and the rest of the population gives zero fucks about them.
So if the government is increasingly in debt, household disposable income has stagnated since 2007, average household debt has gone from £164k in 2015 to £211k in 2024, where has that money gone? Who is benefiting, because the net sum of debt has to be zero (because debt is an IOU to someone else...)

So billionaire wealth has almost tripled over the past decade or so. Which, IMO, solves that puzzle, especially in a stagnant economy.

There is a problem with the pie not growing, and yeah, the Tories really fucked things. They knew the game was up, stopped bothering to govern and then salted the earth, e.g. https://www.ft.com/content/fef202b8-576d-429d-8136-cbb6db52f0f0
“There’s very little going on day to day,” conceded one Conservative minister. A Labour insider argued some prime debating slots in the chamber were being dedicated to “niche” issues to fill time.

On Tuesday last week the Commons adjourned at 3.53pm, while on one Monday earlier this month almost five hours were dedicated to an unfocused “general debate on farming”, critics pointed out.

The FT analysis used parliamentary data going back 27 years — the data is available online — as well as Commons Library research.
...

It is an assessment shared by some Conservatives. One former minister said there was a relative lack of hefty legislation at present: “Other than Rwanda and the Budget, the legislative agenda seems quite thin at the moment.”"


Those two cuts to National Insurance and the fraud around not sharing details with the OBR were insanely reckless.
It's nuts that there's just been a poll putting the Tories ahead of Labour.

But Britain isn't massively over indebted, It's right in the middle of the (admittedly too high overall) the pack when it comes to indebtedness, well below France, Italy and the US. And it's not like it's forecast to rise massively, the budget crunch is due to the fiscal rules, not actually debt out of control:


As for your policy proposals:

Going forward, they need to get off the cheap immigrant labour drug.

Yeah, it's the only thing that has been making GDP go up, if you actually look at GDP growth per capita, it was negative in 2023 and zero in 2024.

Education, particularly in secondary, narrows too soon, leaving people financially illiterate and not very well prepared for the world ahead of them, it also sends a lot of kids who aren't suited to college in, only to fail and walk away with debt and no prospects.

I'm not a fan of the A-level system, on equivalent PISA scores, the UK has been in decline too, with Wales in particular doing badly.

They need to raise the basic rate of tax to 23%, cut mental health PIP, cut motability.

I would disagree with raising the basic rate, it's a blunt weapon and you're primarily taking money from people who will actually spend it in the real economy.
The systems for both PIP and motability need major reform, but mental health is massively underfunded, so cutting PIP just makes a shit system worse. The SEND process around schools is insane, bureaucratic and massively wasteful. The mental health provision in the NHS is a mess and care in the community has been used as a cost saving measure. The police are having to deal with far more mental health situations than they should, because the other services don't deal with it properly.

Reduce legal and illegal immigration to about 100k per year for a decade or so (this reduces rent pressure, and causes a correction in the price of housing, reducing talent flight, leaving the economy on a better footing, long term, again, this isn't some mad racist rant, I'm an immigrant there myself).

I mean, the exact level isn't important to me but 1 million back in 2023 was insane. Forecast for this year is down to about 200k, give or take 50k. That's within the bounds of acceptability for me, especially when house building is lagging at about 150k a year.

Invest heavily in primary and secondary education.

Only partially agree, I don't think the malaise the UK is suffering comes from mediocre primary and seconday education. There has been a lot of damage to the university sector, especially with the clampdown on migration, I think that needs to be fixed before it becomes a bigger issue. The UK has an excellent university sector for its size, but it's declining rapidly.

Invest heavily in energy infrastructure and green energy.

+startups, reshoring automated manufacturing and housing.

Fundamentally, industrial and overall business policy has been nonexistent for a couple of decades. I think the UK government shouldn't be afraid of founding or investing in companies, which it is kinda doing with Great British Energy & Great British Railways.
Just stop calling everything Great British.

 

[Bennyton]

Great response.

RE: SEND, it's a mess everywhere tbh, hundreds of SEND kids in Cork still have no transport to school, and Bus Eireann have stopped responding to the schools or the parents.

My primary concern with british education is that it's too narrow at A Levels, and frankly, not challenging enough when you compare it to the Leaving Cert.

RE: PISA - I'd be very interested in let's say, seeing the financial literacy results for the UK, test scores will be fucked for a decade across the board, until the covid babies grow up, and that's expected.

However, under educated kids become adults, and financially illiterate adults buy silly financial products.

Take this for example, I'm not having a go at the FCA, per se, but we need to make sure that people understand compound interest, it's like a dumb tax, preying on the vulnerable:


In fairness to FG, while they fucked everyone on tax and austerity, and fucked my generation under the bus, they did have a national industrial strategy, and there are long term projects ongoing always when it comes to infrastructure, and even cross party support for slainte care, universial healthcare, and hot school meals, the fact that each party buys in is critical, it means the wheels stay on, no matter who gets voted in.

Really Ireland gets some of the easy stuff hilariously wrong, but some of the hard stuff, they do remarkably well with, it's pretty puzzling.

 

[How bad boy]

Great response.

RE: SEND, it's a mess everywhere tbh, hundreds of SEND kids in Cork still have no transport to school, and Bus Eireann have stopped responding to the schools or the parents.

My primary concern with british education is that it's too narrow at A Levels, and frankly, not challenging enough when you compare it to the Leaving Cert.

RE: PISA - I'd be very interested in let's say, seeing the financial literacy results for the UK, test scores will be fucked for a decade across the board, until the covid babies grow up, and that's expected.

However, under educated kids become adults, and financially illiterate adults buy silly financial products.

Take this for example, I'm not having a go at the FCA, per se, but we need to make sure that people understand compound interest, it's like a dumb tax, preying on the vulnerable:

View attachment 45595
In fairness to FG, while they fucked everyone on tax and austerity, and fucked my generation under the bus, they did have a national industrial strategy, and there are long term projects ongoing always when it comes to infrastructure, and even cross party support for slainte care, universial healthcare, and hot school meals, the fact that each party buys in is critical, it means the wheels stay on, no matter who gets voted in.

Really Ireland gets some of the easy stuff hilariously wrong, but some of the hard stuff, they do remarkably well with, it's pretty puzzling.

Agreed. The other thing is Ireland is starting from a much, much lower level. Britain has the history of being the richest country on earth and a massive empire until the 50s. There's a lot of physical and institutional capital that's built up as a result. Ireland was pillaged pretty thoroughly and broke until the mid 90s. It's the difference between income and wealth, Britain is like the old worker nearing retirement who's just plodding along but they paid off their house long ago, Ireland is like the young graduate who's doing incredibly well but owns nothing and has a shitton of massive costs in front to catch up.


The school meals is a good example, every British primary school has had a kitchen since forever, Irish primary schools just didn't have them.


Both countries have massive problems with just how hard it is to get stuff done. The children's hospital is Ireland's HS2. Both emblematic of a level of institutional inertia that I think ends up pitting "good enough" against some theoretical perfect state* which makes everyone happy, but doesn't actually exist. And because they do these things infrequently, they're generally not very good at them (and get fucked by the private sector for their naivety and incompetence).
So in the time/cost/quality trade off, they aim for the latter but inexperience makes it that they don't truly get any of the three.
Which means, the next time a big project is proposed, it gets killed off or torn to pieces politically.

The solutions aren't to turn it over to the private sector or not do it though. It's to do more, but being more systematic about it.


*the expensive term would be a pareto optimal...

 

[Ogre Moran]

Agreed. The other thing is Ireland is starting from a much, much lower level. Britain has the history of being the richest country on earth and a massive empire until the 50s. There's a lot of physical and institutional capital that's built up as a result. Ireland was pillaged pretty thoroughly and broke until the mid 90s. It's the difference between income and wealth, Britain is like the old worker nearing retirement who's just plodding along but they paid off their house long ago, Ireland is like the young graduate who's doing incredibly well but owns nothing and has a shitton of massive costs in front to catch up.


The school meals is a good example, every British primary school has had a kitchen since forever, Irish primary schools just didn't have them.


Both countries have massive problems with just how hard it is to get stuff done. The children's hospital is Ireland's HS2. Both emblematic of a level of institutional inertia that I think ends up pitting "good enough" against some theoretical perfect state* which makes everyone happy, but doesn't actually exist. And because they do these things infrequently, they're generally not very good at them (and get fucked by the private sector for their naivety and incompetence).
So in the time/cost/quality trade off, they aim for the latter but inexperience makes it that they don't truly get any of the three.
Which means, the next time a big project is proposed, it gets killed off or torn to pieces politically.

The solutions aren't to turn it over to the private sector or not do it though. It's to do more, but being more systematic about it.


*the expensive term would be a pareto optimal...

Speaking of Pareto optimal outcomes, is this one?
Jack Sprat would eat no fat. His wife would eat no lean. And so between them both, you see, they licked the platter clean.

 

[How bad boy]

Speaking of Pareto optimal outcomes, is this one?
Jack Sprat would eat no fat. His wife would eat no lean. And so between them both, you see, they licked the platter clean.

Basically covers it, yeah

 

[omar]

In a break from some of the standard macro news around things like growth an unemployment, wanted to do some thinking on the AI boom.

I'm starting from the position that, despite being neck deep in AI professionally, I think we're in a massive boom on the infrastructure side that's going to pop pretty soon.


NVIDIA just topped $5 trillion market capitalisation. That's more than the GDP of every country in the world besides the US and China. Which is a lot.

This is a graph of its value, it was worth only $400 billion this time 3 years ago:
View attachment 45551


That should set off alarm bells. In that time their revenue has gone from just under $30bn in 22 to $130bn. Which is a pretty impressive improvement.

There's a bit of a circular economy growing up around NVIDIA that I think is fucking with their valuations:

View attachment 45552

That's not easy to understand, but as an example, NVIDIA are investing $100bn in OpenAI. OpenAI are then going to pay NVIDIA for chips. The way data centres are specced are by power, they're commiting to 10GW of chips, which in today's prices would be about $200 billion in spend.

Combined with other deals like that with Microsoft, Oracle, Coreweave and AWS, they're committing to over half a trillion dollars in spend.
They've done another deal with AMD, similar in nature.

OpenAI's current revenue is reported to be about $13bn a year, but they're losing a lot of money. They're aiming at $100 billion by 2027.

Which is a lot. However an 8x increase in revenue in 2 years is a lot.

Meanwhile, a lot of companies are struggling to get a positive return on investment on AI, an MIT study has only 5% of enterprises integrating AI tools into production at scale (I don't agree with the headline, no production deployment is not zero ROI):

Generative AI does nothing for 95 percent of companies

: MIT NANDA study finds only 5 percent of organizations using AI tools in production at scale

www.theregister.com

Bringing it back to GDP, according to some analysis, you were to take AI related investment out of the US economy, the total US GDP growth in the first half of 2025 would have been 0.1%.

Without data centers, GDP growth was 0.1% in the first half of 2025, Harvard economist says | Fortune

Is a U.S. without data centers a country without GDP growth?

fortune.com

It all looks very dotcom boom. When will it pop? Dunno, maybe never. Or maybe tomorrow. I suspect it'll need something like an interest rate hike or some absolutely disastrous results announcements to trigger it.

Trader who inspired The Big Short sets off AI tech share sell-off

Shares of major tech companies have fallen in the US and Japan over fears about AI bubble

www.bbc.com