Why is losing the ability/interest in navigating through a paper map by hand bad, though?
Humanity has adopted and then discarded skills many times in its history. There were once many master archers, nobody outside of one crazy Danish guy has mastered archery for hundreds of years. That isn't bad, nobody cares, nothing of value was lost.
You can still use pencil and paper for the difficult things. In fact, you'll have more time for doing so, because you don't have to use pencil and paper for the simple things.
Indeed, the researchers undermined their own argument, even:
> ... it is notable that contrary to their main results, Fort et al. find that the stock market value of the average patent has actually fallen over time.
Their methodologies are very indirect and yield contradictory results.
Trying to decide if a patent is important by looking at the evolution of word use doesn't sound robust, nor does looking at the stock market. When Google invented the transformer algorithm, I don't think there was a sudden jump in their stock price. There are lots of papers and people can't evaluate their value immediately like that. Stock prices move in response to earnings, not patents or papers. I don't remember ever hearing about a sudden stock price jump because a patent was filed.
There's lots of other questionable stuff in this argument. How are they defining researcher, for one? For US tax purposes it's common to define all software development as R&D. If they're using similar data then the huge growth of the software industry would make it appear like research productivity has fallen.
APEX is probably just as widely used now as Access was. Access likely had higher market share but of a much smaller market. There are gazillions of APEX apps out there.
Are you being deliberately obtuse? In case you aren't, the answer is roads, bridges, public transportation, electrified rail, grid modernization, utility-scale storage and solar. We need these things desperately, and instead we're going to get sheds full of video cards from here to the horizon.
If you want to spend your money and time building bridges for electrified rail, go ahead. Nobody is stopping you. Other people clearly feel they have enough of that and would rather invest in datacenters. Who are you to say they're wrong?
This kind of absolutist individualist argument just rings more and more hollow as we see the very real consequences of that philosophy for our society.
Who am I to say they're wrong? A human being, that's who. A human being who lives in a modern society that does not have to prioritize the whims of the wealthy few over the needs of the many. We can choose to set stringent requirements on people who have that much money, and therefore power, and that is not evil. Indeed, it is the furthest thing from it.
And what happens when those people don't want to have your requirements "set" on them? Do you force those peaceful people to do your bidding with violence? Would that not make you the evil ones?
Look at the reply from the guy I was questioning. It took just two or three mild questions for him to go full Hitler, talking about how his comrades will have to "discipline" a whole generation of "oligarchs" (i.e. anyone who makes things he doesn't personally prioritize).
Collectivist thinking always leads to violence, and eventually societal failure.
There's nothing violent about using elections to make the decision to tax rich people so that we can spend (formerly) their money building roads and bridges. The idea that this is the road to Hitlerism is absurd, and thankfully this rhetorical stance no longer rings the slightest bit true to anyone within earshot of the working class.
Also, as I'm sure you're aware, I was using "discipline" as a term of art to mean "withhold our labor until their profits suffer and they are willing to negotiate". This was the strategy employed the last time we seriously dealt with concentrated capital getting high on its own supply. It is also not a form of violence. What's the alternative? Capital using force to require us to work against our will? Would you call that slavery? Or just serfdom? Which do you advocate?
> There's nothing violent about using elections to make the decision to tax rich people so that we can spend (formerly) their money building roads and bridges
The results of votes are enforced on the losers using the police, who will do so violently if required.
You mentioned the Fordist truce. The unions the auto industry dealt with weren't just a bunch of people refusing to work. They were frequently violent, and they also used stealing other people's property as a standard tactic to prevent anyone else from working also. Those were violent times.
> The results of votes are enforced on the losers using the police, who will do so violently if required.
The whims of the dictator are also enforced on the public using police.
All human rules, laws, customs, and edicts are enforced, ultimately, with violence of one sort or another. There is no way to avoid the threat of violence being the bedrock of the power of the state, and in the absence of formal states the strong would use violence to enforce their desires until they became states.
So if you're an anti-statist, just say so. (So we can all dismiss everything you have to say as coming from a place of absolute ignorance of what's needed to live and operate in the real world. If we were to abolish all states tomorrow, and erase the very memory of their existence from every human alive, by Sunday new ones would have arisen to replace them, one way or another, because they are how humans organize themselves.)
This thread started with a false dilemma: do we spend money on datacenters or "real infrastructure". It's only a dilemma if you assume governments should decide the answer, as ElevenLathe did.
Otherwise there's no need to choose between dictatorship or majority rule via democracy: everyone can spend money on the infrastructure they feel is more important, and there doesn't need to be any losers. Which is mostly how we try to do things in reality.
We can talk about how violent taxing the rich is once we have the first instance ever in history of the police locking up a rich guy for refusing to pay their taxes. Even then, sure, I'm fine with that level of violence. We would live in a utopia if that were the worst kind of state violence we had to deal with.
Go ahead and twist my normal, non-radical politics into whatever shape you want. You're the wing nut, not me. Normal people want normal stuff out of politics: functioning infrastructure, upward mobility, a future. Only the most warped, unreachable paint huffers are willing to throw away all possibility of a normal country for the "freedom" of a few dozen rapacious sociopaths. This means that we will ultimately win. Unfortunately normal people have been asleep at the switch for at least a generation, so you're probably going to be able to drag us through several hellish decades, maybe centuries, until we can right the ship.
I'm sure I'll see you in the camps, so at least we'll have that in common. Have a nice day.
Falsifying your tax return statements is not the same as a simple refusal to pay. By doing that, you are indicating that you agree to the legitimacy of the taxes in general, but would prefer to lie about whether you should personally pay them or not. These people were also all given their day in court, and convicted of actual crimes in fair trials where they had adequate representation. If this is your idea of "violence", then I don't know what to say.
He was "was sentenced to 38 years in prison". That's why most tax evaders try falsification rather than refusal.
> At the sentencing hearing in Denver District Court, Doucette fidgeted in his spinnable chair, while chained up in a green jumpsuit. He sat alone because he has insisted on representing himself in this case. Before the hearing, the judge asked him if it was OK to proceed and he said, “I do not consent and never have.”
Note the photo of him wearing handcuffs, surrounded by police.
All law is implemented through using violence or the threat of using it. You can't resolve that conundrum by claiming that holding a vote to tax rich people is somehow apart from using violence. It's just an abstraction over it.
These are basic facts, but a lot of people struggle to understand them because our society likes to pretend that there's nothing underneath the abstraction - that courts and rules is all there is. It helps them believe that if they vote to take other people's stuff, it's white and pure, that nobody is getting hurt. It's a "might makes right" argument pushed at every level of society, because it enables what you're doing here: claiming that "we" should be able to choose what is done with the fruits of other people's labour.
I say they're wrong, and I do so in my capacity as a citizen. These large pools of capital should not be allowed to follow the whims of a handful of unelected oligarchs who have clearly lost the plot. In a functioning society, this scale of decision would not be left to the whims of international finance capital, but decided via democratic means. It's unfortunate that the last scraps of the Fordist labor truce are unraveling, because it means that I and my comrades are going to have to discipline this generation of oligarchs just like our grandparents did the last really nasty one.
I would say it should belong to voters (or "society", or "the people", or whatever formula you want to use to express it), in a functioning society. Unfortunately, we're not in a functioning society, and it doesn't. On the other hand, property is socially constructed, so this political economy can be changed, though how exactly is left as an exercise to the reader -- I don't a foolproof answer.
I have savings, sure. I need them, because in the current system the alternative is to starve in the street if anything at all goes wrong with my employment, my health, etc. Many others are not so lucky. If you mean this to be a "gotcha" because I wouldn't want my savings "confiscated" to build roads and bridges, then save it. There is a difference between 1) taxing billionaires so that they're merely hundreds of times richer than the average citizen, and 2) stealing my retirement savings and emergency fund without providing any equivalent public safety net.
No, that's been a common proposal from economists pretty much since people started examining how economies work. Some places do have a capital gains tax of zero. Switzerland is one of them, and Switzerland is economically more successful than the rest of Europe.
Tax is one of those issues where there are actually correct and incorrect answers, thanks to many hundreds of years of active experimentation and relatively simple/robust theory. But people ignore the correct answers for social reasons.
The correct answer on tax is:
1. Figure out how much money the state needs to supply the services that are in-scope for it to an acceptable level of quality.
2. Aim to raise that much in taxes.
3. Optimize deadweight costs. That is, configure taxes to minimize the level to which the activities being taxed are discouraged and driven either out of existence or abroad.
If you do this sort of thing then you get Georgeism, you get zero capital gains, I think you get zero taxes on businesses, and a bunch of other policies I can't remember right now. The results can be economically very efficient i.e. they make everyone better off. However, almost nowhere uses them because there's nothing in the above three items about social engineering, and governments use taxation largely as a tool of social engineering. And in particular to please leftist voters who use the tax system to penalize wealth for its own sake, and to reward groups of client voters. Many governments also have a lot of trouble defining what's in scope for them and then working backwards to needed tax revenues; they prefer to raise as much tax as they can manage without totally crushing their economies and then find ways to spend it.
> Some places do have a capital gains tax of zero. Switzerland is one of them, and Switzerland is economically more successful than the rest of Europe.
Tax
I live in Switzerland. It's an excellent example of a place without a capital gains tax, because it doesn't have one. I didn't say it doesn't have other taxes!
The type of tax matters a lot. The reason capital gains taxes are bad is that they discourage investment, but investment is how you create wealth. "Creating wealth" is ultimately a synonym for creating material progress. Voters like progress, and so this is a very simple and direct argument, which is why most countries that have capital gains tax it at a lower rate than income. Wealth taxes have different incidence and change incentives in different ways. Basically, they discourage having wealth rather than creating it.
It can create its own problems. Switzerland has had big problems in the past with the wealth tax discouraging the creation of tech startups. The reason is that if you create a company then sell some equity in it to investors, that creates a valuation of your company which is then considered wealth, even though it's theoretical wealth and not liquid. In other words, doing a big VC raise can land the company founders with an unpayably massive tax bill: they literally don't have the money to send the government because it's only paper wealth.
To fix that the Swiss tax authorities had to introduce a new rule that says if you have ownership of a startup, this doesn't count towards the wealth tax. What exactly is a "startup" and what differentiates it from other kinds of business? Whether it is "innovative". What counts as innovative? The taxman decides. That means creating a startup in Switzerland is quite risky as if some random bureaucrat decides your product isn't truly innovative and you do a big VC raise you could be personally bankrupted (or you have to use some of the investors money to pay yourself out each year, which is then taxed as income too pushing you into a much higher tax bracket, etc). There are lots of other practical problems with the wealth tax.
Tax incidence is complicated!
In practice the Swiss approach works because:
- The wealth tax is quite low
- This "innovative startup" hack seems to work out in practice even if it's concerning in theory (tech startups aren't the only way to create a lot of wealth)
- Wealth taxes discourage all kinds of wealth equally, so the effects are diffuse and they don't specifically discourage e.g. getting promoted over company formation over inheritances, which is a distortion a lot of other approaches do create.
mike_hearn: Switzerland has no capital gains taxes and it's great.
triceratops: Ok but it still taxes capital.
mike_hearn: I live in Switzerland. No capital gains taxes are great and everywhere other than Switzerland has a lower tax rate for them than income because we want more capital gains. Also wealth taxes can cause startup founders to be taxed heavily.
There's a bit of a disconnect here. You're arguing against multiple strawmen IMO.
Outside Switzerland the current situation is: regular people pay high income taxes while they work, then somewhat lower capital gains taxes in retirement. Ultrawealthy people pay far less of both because they have ways to avoid them (keep employment income low, borrow against wealth instead of selling it).
In Switzerland, since the wealth is straight up taxed, even if at a lower rate (I ran the Swiss wealth tax numbers myself a while ago and you're right it really is a very small amount. I pay way more in capital gains taxes) there are fewer games. Everyone pays taxes on what they make or own.
The startup wealth tax problem has another solution: allow payment in non-voting startup shares, instead of liquid cash. The shares go into a sovereign wealth fund. The government either reaps a windfall eventually alongside the founder, or it misses out on tax revenue it shouldn't have collected anyway (if you look at it from the fairness point of view).
You're right, the origin of this thread was making an argument about all taxes on capital, not just capital gains. I missed that, I guess because nobody mentioned wealth taxes specifically and it's fairly rare for taxes on capital to mean anything other than capital gains tax. Mea culpa.
> The startup wealth tax problem has another solution: allow payment in non-voting startup shares, instead of liquid cash
This is an excellent idea! Did you come up with this yourself or have you heard of others proposing it?
I came up with it myself. It's possible there's prior art but nothing that I've read personally.
I don't think it's a particularly revolutionary idea because sovereign wealth funds already exist. Improving productivity means using less labor which means lower income tax revenues as time goes on (and that's what you want - higher productivity, fewer labor inputs).
And yet, the government needs revenue. What's growing? Wealth. Liquidating wealth to pay taxes is problematic. Hence the sovereign wealth fund. You can apply this to most forms of wealth - even publicly traded stock, real estate, crypto, and artwork.
I've proposed it on this site several times in the past.
Why does it matter whether the value of the shares is observable every day, like a public stock? The value of the shares is quite defined at the time the tax is due. We know this because the government has a specific number in mind for valuation for tax purposes.
The shares are illiquid and that poses a problem for the taxpayer because the government only accepts cash. If instead they could sign over an equivalent number of shares then morally (and arithmetically), they've paid what they owed.
The government may subsequently choose to dispose of the shares on a secondary market, if one is available. Or it may hold on to the shares until there's a liquid, public market for them. Or it may never sell. It all depends on how the sovereign wealth fund is managed and structured. Way smarter and more knowledgeable people than me would have to design how the fund actually works and prevent market manipulation and insider trading.
Sure, but that doesn't stop them being taxed under whatever their most recent valuation was under a wealth tax. Just not taxing non-liquid assets would also be an improvement.
Investment is not how you "create wealth". An actual worker somewhere performing their job is what creates wealth. Yet when that worker is paid for the wealth actually produce, we tax that heavily. So if you want to encourage productivity, regular income ought to be taxed higher than passive investment.
The argument for low capital tax is that if it's high, the people with the capital - who, crucially, need someone else to use it to make money from it - will just hoard it. For one thing, the obvious glaring issue with it is that however high the capital gains tax is, so long as the owner of capital in question still gets to pocket some of the wealth produced using it, they still have an incentive to continue - something is better than nothing. The actual, real world threat is that some other jurisdiction sets the tax rate lower than you will, and capital will then move there. But this same threat applies to many other taxes, capital gains aren't special in that regard.
This is the kind of semantic argument about words that makes anything other than flat personal taxation an endless rabbit hole.
When people talk about wealth creation they mean the creation of new wealth. Filling potholes isn't normally described as wealth creation because it's sustaining activity. You can choose to define wealth creation differently, that's fine, but it makes the term useless because it'd become synonymous with any kind of work.
Additionally, there's no real world difference between investors and workers. The idea you can separate capital as a class of people from workers is a Marxist concept that doesn't make any sense outside that broken ideological framework. The classical example: if someone owns a food stall, are they capital or a worker? If they pick up that stall and cart it to a bigger town down the road, is the act of them hauling their cart along the road work or an investment? You could argue equally well both ways, which makes the distinction just a distraction.
> however high the capital gains tax is so long as the owner of capital in question still gets to pocket some of the wealth produced using it, they still have an incentive to continue
Not at all! This is the kind of weird prediction that false distinctions between capitalists vs workers causes. It's why Marxist economies always fail. Investment is work and it also requires taking a lot of risk. If you confiscate 99% of someone's ROI nobody is going to say oh well, at least I got 1%. They're going to give up investing at all because the act of making the investment not only took effort, but also meant they could have lost the whole shebang.
People aren't clearly separatable into "owners of capital" vs "workers", you're right, but they don't have to be. You just need to recognize that the role that they play at any given moment can be so categorized. And sometimes they play many roles at once - for example, a company owner who is also its CEO is both a capitalist and a worker, and fair wages that he receives as the latter (fair here meaning that an equally capable manager hired from the side would ask for this much on average) is not a problem.
If there was no difference between capital and labor, then capital gains and labor income would be taxed at the same rate. That's just the empirical argument. The theoretical is left as an exercise to the reader.
I feel like you have only a cursory understanding of finance, economics, and taxation. If you didn't, you would't ask questions such as
if someone owns a food stall, are they capital or a worker?
It reads like you're trying to find evidence that reinforces your priors while dismissing whole swaths of empirical and theoretical work that would immediately challenge it.
For context, I spent a decade as an M&A banker, so as far from a Marxist as one can be.
> If there was no difference between capital and labor, then capital gains and labor income would be taxed at the same rate.
There's a distinction between capital and labor when the terms are used in an accounting sense but when "capital" is used as a shorthand for a class of people, there isn't. Once someone starts talking about "actual workers" vs "owners of capital" they're drawing that distinction.
Isn't a wealth tax just an expanded capital gains tax?
If last year I had wealth X and this year I have wealth X+Y, I have to pay a wealth tax on the gains, in addition to the the tax on the amount I had previously.
- Wealth tax is much lower, think a percent of your wealth or less vs 20% of your gains.
- You can avoid wealth tax by spending. If you sell a bunch of shares to earn $100k then take a year off to see the world, you pay no tax on that (other than sales taxes etc).
- In practice a lot of things aren't covered by wealth tax. If you spend on a fancy new TV it's not measured. Only the big ticket items are wealth taxed (houses, financial assets, art, cash piles, etc).
The Switzerland model is unique in several ways, both in its history, which cannot be replicated, and in embracing of...questionable financial services.
It's unclear that the model can be replicated generally, let alone whether it should. Importantly, there may not be sufficient demand for banking services like the Swiss provide.
Your three step plan says nothing about how much should be taxed at the personal vs corporate income level, or on the gap between capital gains and labor income taxes.
I'm not arguing for higher tax revenue overall. I don't believe in that, but I also wouldn't even need to make the argument even if I believed in it.
The simpler, more defensible argument is that taxes on capital gains must be much closer to income taxes. Historically they were, even in the US, and we seemed to be fine.
The idea Switzerland's economy is dodgy or dependent on banking is an urban legend. Only 10% of Swiss GDP comes from finance at all and that includes everything, including insurance and pensions. Private banking is only a fraction of that, and private banking with anonymous accounts - which is what people tend to mean by this - was a tiny fraction of that again.
Meanwhile, financial privacy isn't inherently questionable. The USA did a big push in the 1970s to strip privacy from the financial system which until that point had been the default. That was the birth of the concept of money laundering, created as part of the war on drugs. The approach failed as drug cartels found ways to launder money cheaply enough that it wasn't a big friction for them (normal estimate, it adds ~10% to their costs). Not everyone thought that was a great tradeoff, and the Swiss numbered accounts had been used by people trying to hide from the Nazis.
At any rate, the USA forced their concept of anti-money laundering on the world (not that most countries needed the arm twisting) and Switzerland has implemented exactly the same policies as everywhere else for decades. It has no special rules with respect to banking for a long time now.
> Your three step plan says nothing about how much should be taxed at the personal vs corporate income level
It's a set of principles for answering those questions, not the full set of answers.
It's been years since I looked at this but IIRC the general agreement is that you shouldn't bother with corporate/business taxes, because they're both an indirect/inefficient way to collect tax (all taxes are paid by people in the end), and easily avoided.
It was for this reason that the designers of the EU's taxation system originally configured corporate taxation to be collected wherever the nameplate was (i.e. an arbitrary location chosen by the company). The assumption was that with time individual countries would compete the corporate tax rate to zero, fixing the underlying inefficiencies. Of course what's actually happened is some of the countries try to gang up on the others to try and force them to stop lowering taxes. It's not a stable outcome, politically.
In practice business taxes are popular because politicians view them as a way to tax citizens of foreign countries. That has bad effects too but schools either don't teach economics or don't teach it properly in most places, so there are lots of weird hacks like this where something that creates more harm than the alternative gets preferred because people can't resolve the harm to the root cause.
An ideal LVT would tax the full economic rent of the land, but that's unlikely to happen. We don't want to overshoot 100% because that would cause land abandonment.
So in theory, LVT could collect more tax than the state needs to fund services. If that happen, it would be distributed as a Citizen's Dividend.
I am skeptical that we wouldn't be able to find a productive use for government spending, but that's a discussion for citizens of a Georgist state to have.
Also, Georgist policies would discourage the existence billionaires and other people with extreme wealth simply because a lot of their wealth came out of economic rent.
I never understood this Georgist argument. The richest people in the world today require very little land. Remote working is easily possible and plenty of companies use it, even if managers don't always like it. This feels like a medieval perspective.
Georgists aren't frozen in time, nor had they ever been limited to just taxing "land". We consider any economic "land" fair game. We even discussed network effects that allow companies like instagram retain a monopoly.
In any case, California are where some of the most powerful tech monopoly are located, and not coincidentally it's also where some of the most expensive land there is.
Just my opinion as a Georgist amongst many, I would categorize copyright and patents as non-reproducible privilege rather than economic land though non-reproducible privilege also describes private ownership of land. It's very clear that it's artificial, as ideas do not suffer from the exclusivity problem that comes with owning physical land. What IP has in common with owning land is the extraction of economic rent.
Economic land is anything that's fixed, finite, and not man-made, such as land, the electromagnetic spectrum, and orbitals.
Services like amazon and instagram are something of a puzzle to Georgists, but it's at least clear that Amazon and instagram benefits from labor and effort of the platform users. Without people selling on Amazon, there's no amazon. Without users, there's no reasons to be on instagram. To be perfectly clear, platform companies obviously put in labor to build their services, but the network effect isn't entirely of their own making.
How much debate about this stuff still happens, outside of academics who need to publish or perish? From a quick check the only papers being published on this stuff are all variants of "if we add another variable <X> to some model we can conclude the difference in tax incidence is <Y>" where Y is some minor variant of pre-existing beliefs that may or may not hold in reality. Outside of that there are occasional flareups caused by old-guard left wing politicians or activists getting into fights about it with economists, and not much else.
A lot of the debate on this topic is tedious anyway because it revolves around semantic distinctions that only exist in specific kinds of ideological discourse.
It might have been human operated, but it also might have just been copying its training data.
A robot that properly supports being teleoperated wouldn't immediately fall over the moment someone deactivates a headset. Falling over is almost the worst thing a robot can do, you would trash a lot of prototypes and expensive lab equipment that way if they fell over every time an operator needed the toilet or to speak to someone. If you had such a bug that would be the very first thing you would fix. And it's not like making robots stay still whilst standing is a hard problem these days - there's no reason removing a headset should cause the robot to immediately deactivate.
You'd also have to hypothesize about why the supposed Tesla teleoperator takes the headset off with people in front of him/her during a public demonstration, despite knowing that this would cause the robot to die on camera and for them to immediately get fired.
I think it's just as plausible that the underlying VLA model is trained using teleoperation data generated by headset wearers, and just like LLMs it has some notion of a "stop token" intended for cases where it completed its mission. We've all seen LLMs try a few times to solve a problem, give up and declare victory even though it obviously didn't succeed. Presumably they learned that behavior from humans somewhere along the line. If VLA models have a similar issue then we would expect to see cases where it gets frustrated or mistakes failure for success, copies the "I am done with my mission" motion it saw from its trainers and then issues a stop token, meaning it stops sending signals to the motors and as a consequence immediately falls over.
This would be expected for Tesla given that they've always been all-in on purely neural end-to-end operation. It would be most un-Tesla-like for there to be lots of hand crafted logic in these things. And as VLA models are pretty new, and partly based on LLM backbones, we would expect robotic VLA models to have the same flaws as LLMs do.
There are other issues. In January they claimed that a US health report contained "fabricated" and "AI generated" citations with the headline being a claim from a Cigna Group report. Their claim it's fabricated is based on nothing more than the URL now being a redirect of the type common in corporate website reorgs.
I did some checking and found the report does exist, but the citation is still not quite correct. Then I discovered someone is running some LLM based citation checker already, which already fact checked this claim and did a correct writeup that seems a lot better than what this GPTZero tool does.
The mistakes in the citation are the sort of mistake that could have been made by both a human or an AI, really. The visualization in the report is confusing and does contain the 73% number (rounded up), but it's unclear how to interpret the numbers because it's some sort of "vitality index" and not what you'd expect based on how it's introduced. At first glance I actually mis-interpreted it the same way the report does, so it's hard to view this is as clear evidence of AI misuse. Yet the GPTZero folks do make very strong claims based on nothing more than a URL scraper script.
It's not, there's lots of ways to resolve citations without even using AI.
I experimented a couple of years ago with getting LLMs to check citations but stopped working on it because there's no incentive. You could run a fancy expensive pipeline burning scarce GPU hours and find a bunch of bad citations. Then what? Nobody cares. No journal is going to retract any of these papers, the academics themselves won't care or even respond to your emails, nobody is willing to pay for this stuff, least of all the universities, journals or governments themselves.
For example, there's a guy in France who runs a pre-LLM pipeline to discover bad papers using hand-coded heuristics like regexs or metadata analysis e.g. checking if a citation has been retracted. Many of the things it detects are plagiarism, paper mills (i.e. companies that sell fake papers to academics for a profit), or the result of joke paper creators like SciGen.
Sometimes this kind of problem can be fixed by adjusting the prompt.
You don't say "here's a paper, find me invalid citations". You put less pressure on the model by chunking the text into sentences or paragraphs, extracting the citations for that chunk, and presenting both with a prompt like:
The following claim may be evidenced by the text of the article that follows. Please invoke the found_claim tool with a list of the specific sentence(s) in the text that support the claim, or an empty list indicating you could not find support for it in the text.
In other words you make it a needle-in-a-haystack problem, which models are much better at.
Humanity has adopted and then discarded skills many times in its history. There were once many master archers, nobody outside of one crazy Danish guy has mastered archery for hundreds of years. That isn't bad, nobody cares, nothing of value was lost.
reply