The UK, unlike continental Europe, doesn't have either domestic registration or even exit passport number tracking. The EU has a much better idea of how many people are legally in Schengen than the UK does about EU nationals.
The DM isn’t the source and there are dozens of articles that will report the same finding. I’m not going to spend the extra time to find an ideologically neutral “news” org for an HN post when the topic being discussed is generally agreed upon.
To be less dramatic and more realistic than this headline: there is a small set of economics papers which make use of an inaccurate estimate of populations way in the distant past.
Seriously the fraction of papers in which you see these data used is in my personal experience 0. As a fraction of the total amount of economics research produced annually is probably 10^-6 or smaller. It just isn’t necessary in 99.99999% of papers to make any reference to any national or regional populations 2,000 years ago or more.
Also maybe ask the authors of these papers how seriously they take the estimates OR how much weight they ask the estimates to bear in their papers?? “Not very” and “not much” are the most likely answers.
I was expecting though a more comprehensive result on the "bad use" of the data, the section "How Economists Use MJ", mentions ~14 papers (out of how many? out of how many checked? any quantitative analysis?) that will be affected (and anyhow in various ways).
This seems to have some educational value to some (economics) authors that rely too much on the fact population is known, but there are many bad assumptions done in many papers, so doubt this (some researchers relying too much on inexact data) is such an outlier.
In fairness, Maddison's time series which use MJ are pretty much the canonical source for the "look how much things have changed" graphs, and data errors make "look, France was much richer than Poland" sort of comparisons" more dubious[1]
But it's also not as relevant or surprising as it sounds, since the numerator of those fractions is also guesswork. They're not dividing accurate figures for the total wealth of European countries to get an inaccurate estimate of how rich the "average" person was, they're actually tending to work the opposite way from scant data on wages, prices and what "average" people owned multiplied by scant data on population to get a wealth of the nation figure[2]. Needless to say, there's a lot of margin for error subjectivity and scope for the introduction of personal biases in those estimates too
(not to mention the inequalities and completely different access to stuff that makes adjusting for purchasing power and comparing the wealth of Inca subjects, the average medieval Englishman, the average modern Equatorial Guinean and the average modern Briton from a single GDP per capita PPP metric pretty dubious)
Essentially no economic theory (as opposed to economic history) depends on the accuracy pre-Industrial Revolution time series for inspiration or validation, and if you double the values for the period before the exponential growth curve, you still get an exponential growth curve.
[1]except to the extent that the figures were guesstimated based on the consensus of historians that country X is much richer than country Y
[2] cf modern Nigeria, where we have reasonably accurate census data and an "oops, using our new method of calculating GDP, it seems to have doubled previous estimates"
Some economies, like New Zealand/Aeoteoroa were not populated until 1200/1300 CE which would be interesting to see in the stats. Other economies populated as translocation/transmigration of the population of other places wholesale. Indonesia had this, at scale. Millions of people moving. Hong Kong. 1949 partition in India caused massive shifts between two economies. Do we discount this happening in past times?
Estimates of population based on the number of individuals reported killed in battle would be very culturally informed. I have no idea how many camp followers a hoarde has, facing off the Roman army in Farther Gaul or Spain. How do you estimate population under migratory communities in steppes, or climate change effects in northern Europe?
It feels very guesswork-y
I liked the circularity reference. Easy to buttress yourself in a chain of references to peer review legitemacy, so economy estimates inform population and vice versa? use economic indicators when you don't have pop stats and don't ask too many questions where they came from!
We don’t even know the current human population within more than a couple percent. Being at the right order of magnitude for things that happened centuries ago seems pretty good to me.
The opposite appears to be more the story - why did Gates-Meta-FB Research put millions of dollars into post-doc methods to quantify "every school on Earth" or "population density of every 10km area on Earth" ? Their detailed mapping world-wide is well known and documented.. the FB people had some .. excuse me.. cult-like phrases about "nobody is excluded" .. It seems like economic spying with a cover of social goals to motivate young coders, from here.
The notion of a pre Westphalian nation is itself a little flawed, so definitionally it doesn’t even really make sense to pose the metric as such, let alone measure it.
Well, there existed both nations and nation-states, fighting other nation states (often with conscious national sentiment and pride, like any modern era nation).
It was not all some blur, just because they didn't have modernity-style institutions.
But particularly the post-1648 notion of compact integrated territory just isn't there in premodern times except in a few exceptional cases, and population counting in the modern sense is bound up in that. But to take a particularly egregious case in the Holy Roman Empire a person might have three legal "existences" from the state's standpoint: one as a mediate vassal, one as an imperial knight, and one as a delegate of the church, and by the thought process of the HRE that was three different "persons" in one body.
I wonder if there's religious symbolism attached to this, i.e., I wonder if people in the HRE made an allusion between the trinity you mentioned and that of the Father, Son, and Holy Spirit.
No, three is just an arbitrary number for their example. The point is not the exact number of roles but the fact that the basic legal unit is the relationship, not the person. For instance a prince-bishop might in principle be subject to an archbishop as a bishop, subject to the Emperor as an imperial prince, subject to the King of Bohemia (who was in turn subject to the Emperor) as the holder of a secular fief, and part of the Polish nobility, all at once.
yes this is true today - some researchers have identified the LEGAL definition of a person taking over from the biological definition of a person, in modern times. If you consider the realms of finance, things are definitely not consistent nor fully resolved.
Yeah, in retrospect I realized the three might be distracting but it was just the example I thought of at the moment. (It gets even worse if the guy also has burgher status in a free city!)
Not really. While there was a Holy Roman Empire of the German Nation, it was neither an empire nor a national one; people identified much more with their local land or town than some nation. There was no Italy. Some nations currently residing in the former territory of Austria-Hungary did not exist at all - sure there were local languages and dialects and culture and so on, but there was no national identity of eg Slovaks.
It's hard to understand that from our perspective, but bringing the current notion of a nation or a nation state into 16th century is an anachronism.
Well, duh! We hardly know the population of any country more that a few hundred years ago. Even now, some numbers are guesswork - how many Russians have died in the war in Ukraine? That's already an uncertainty of tens of thousands.
For older societies, who even counted as persons? Slaves? Babies? Prisoners of war? Outlaws living in the woods?
And that's even assuming there _is_ a census. The Domesday Book was quite exceptional a thousand years ago, and even that only gives us a population estimate between 1.2 and 1.6 million (see Wikipedia). Any place with a poorer census or written record (i.e most of the world at the time) would be even more of a guesswork.
> For older societies, who even counted as persons?
Even more importantly, who is writing. Julius Caesar defeating seven bajillion Gauls is not a figure that should be taken at face value if the source of that number is Julius Caesar.
Sorry for the offtopicness but could you please email me at hn@ycombinator.com? There's an issue with your account (nothing you did wrong! but something you need to know about).
Because the traditional reason for census is taxation, voting, or military service, and in those cases you effectively only care about free adult males and count only those.
For example, the original Roman census, from which we take word 'census', registered and counted adult male citizens. They did list their wife, children and slaves as part of extra information about their household, but those dependents weren't part of the census itself, just an "attachment" to the recorded citizen just as the notes about their land and other porperty. Other people like single women or orphans would get reported by their guardians, but they as well explicitly weren't included in the "count of heads", which counts only adult male citizens i.e. those who can run households and own property, be recruited to fight, pay taxes, and vote.
In effect, the ancient world doesn't really consider persons or individuals, they care about households (represented by a single individual) and so register and count those.
Historically, census was typically done for taxation purposes. Kids, homeless etc was just not that relevant since households were taxed. Slaves counted as property, not persons.
The question is what those older societies included in their censuses. The reason for the census and the definition of a person can vary wildly.
Remember that not that long ago at this scale, the USA considered slaves as 3/5 of a person when determining states' populations for representation and taxation purposes.
To be more precise in a few more words, the article makes two points:
* some papers rely on McEvedy and Jones’ Atlas of World Population History data as the ground truth, when a lot of it is the best educated guess (sometimes the guess is based on basically almost nothing).
* the difference between the "guessed" population and the real population (which do not observe) is not random, which would lead not only to less precise results but also to bias.
Thanks. This is one of those cases where the article is more interesting than its title.
I've taken a crack at revising the title, though not a very good crack - it's better to use language from the article itself and I didn't do that in this case. If anyone wants to comb through the article and find a representative phrase that would be more suitable, we'll happily edit it again.
I read this on repec a few months ago. Interesting seeing it here.
A better title:
Economist points out some economists had been using a dodgy source; cites other economists who had pointed this out adding to literature of checking citations and a growing body of literature on checking claims in economic history in particular [editorial: as these feed into pop science years down the road]
Original title was far more accurate than the editorialised one which is extremely inaccurate ("economists" outside the subfield of economic history do not "typically" "rely" on population data for the pre-Industrial Revolution period at all) and seems like clickbait for people to add their middlebrow-at-best dismissals of the field.
"Widely used historical population time series mostly guesswork" would be a more accurate title, though you might get a few "water is wet" replies :)
I think hard sciences are amenable to more precision, but it’s fine to be imprecise as long as you’re cognisant of the level of precision you’re using. The problem with the studies being criticised is their conclusions aren’t supported by the level of precision.
I highly recommend reading some of AE Housmann’s prose. He lays out a vision of _English Literature_ that intellectually rigorous.
the correct answer to all of these is yes, like obviously? all of those countries are countries. If you control your own territory you are a country. Transnistria, western sahara, somalialand, also. Sure there are nonsense political reasons countries' governments dont recognise these but as a regular citizen in a free country the answer is pretty clear right?
> Since 1922 the United Kingdom has been made up of four countries: England, Scotland, Wales (which collectively make up Great Britain) and Northern Ireland (variously described as a country,[1] province,[2][3][4][5] jurisdiction[6] or region[7][8]). The UK Prime Minister's website has used the phrase "countries within a country" to describe the United Kingdom.[9] Some statistical summaries, such as those for the twelve NUTS 1 regions of the UK, refer to Northern Ireland, Scotland, and Wales as "regions".[10][11] With regard to Northern Ireland, Scotland and Wales particularly, the descriptive name one uses "can be controversial, with the choice often revealing one's political preferences".
What about the Isle of Man? It's not part of the UK, but is a self-governing possession of the British Crown.
There weren't many "countries", per se. There are artifacts from most major empires but untold numbers of city-states, villages, and small-medium settlements are lost to time. There are some finds like Los Millares in the Eneolithic, but there isn't much left of mesolithic settlements apart from about a dozen examples. In terms of proof of political organization, that requires evidence like writing or inscriptions.
> 2. presence of settlements is an indicator of political organisation, but political organisation is not a requirement for settlements
I disagree. People living next to each other need to reach a political order (organisation) even if it is tacit. A common understanding of a shared living space where you can trust your neighbour to not try to steal you or enslave you is a form of political organisation, even if no interaction between the settlers ever occur. Likewise, a settlement where the strongest people enslave the others is also a political organisation. Also, seasonal settlements where people only spend a few weeks a year also require political organisations to coordinate on the dates and logistics.
Monarchy and anarchy are forms of political organisation, just like families, unions and corporations.
I would agree if we were talking about _Institutionalised_ political organisation, but this statement would still be wrong:
> In terms of proof of political organization, that requires evidence like writing or inscriptions.
Other forms of evidence can prove institutionalised politics. For instance, the presence of shared infrastructure, palaces, temples, monuments, royal funerary sites, battles, etc.
As a physicist, it might just be my layman's impression, but it does seem like economics people are just putting a list of points in Excel, no error bars or anything. I don't think I've ever seen some error propagation stuff in some Economics calculation.
As a physicist, I think you probably don’t have enough experience with economics research. (Which is fine! It’s not your field!)
Quantification of uncertainty is bread and butter for us! Criticize the field fo relying on null hypothesis significance testing, not for failing to provide measures of uncertainty!!
I've seen errors being considered while fitting a model, but not in the source data or when using the fitted data in another calculation.
But yeah, I hope it's just a simplification for laymen like me. But the fact that apparently there weren't errors bars in a dataset so important such as population estimatives doesn't give me the right impression
I can feel that twitter account draining my IQ just by looking at it.
"One of Britain's forgotten issues is that the government does not know who is currently in the United Kingdom, in any sense."
True if one ignoring the census, or the many registers, national and regional, the 1 surveillence camera for every 10 people watching over us or the many corporate holders of everything from your physical location to your browsing history that they are compelled to report to the UK government.
That thread is not substantially incorrect; the data really isn't very well correlated. You can estimate using things like "number of mobile phone users" or "internet users", but that leaves out potentially large numbers of people. The census and the electoral register are basically voluntary (yes, the electoral roll is mandatory, but not meaningfully enforced at all).
Agree that the account is sus, starting from the anime pfp.
It is however trivial to correlate - if there was a desire (or significant error) deemed it to be so.
Of course basing your population estimate off the number of tax returns would "leave out some people" (>90% of the population of the uk for that example). Likewise basing it off the register of all children in schools would ignore most people over the age of 18.
Ultiately the issue is that these people claim is rife is - all these undocumented people are getting all the jobs, benefits, healthcare, housing, government grants and school places which ignores the fact that to get any of those you have to be documented and registered on a myriad of systems to access that.
> all these undocumented people are getting all the jobs, benefits, healthcare, housing, government grants and school places which ignores the fact that to get any of those you have to be documented and registered on a myriad of systems to access that
This is correct. There are people operating outside of all of those - there was a number of illegal schools in Birmingham that OFSTED spotted and shut down - but to get into the legal system and especially to claim money you need identification.
Isn't the larger issue that the UK, like the US, doesn't have passport exit controls so neither government can actually say who is and isn't in the country at a given moment?
Census: the estimated return rate of the last census was 97%, i.e. 3% of the population didn't give it back. That's millions of people even by the official numbers. Yet there are only ever a few hundred people who get convicted for failure to complete the census. Meaning, the chances of anything bad happening to you for not filling out the census is on the order of 0.01%. That's a very weak incentive for honesty. If you're already living on the outskirts of the law (illegal immigrant, trafficked, etc), you can just ignore it, and it'll be about as risky as ignoring a TV license threat letter. The accuracy of the census depends on the bulk of the population's honesty and willingness to do this civic duty. Most people think "oh that's something you just have to do", like stopping for traffic lights on an empty road at 2am. It's an illusion, you can just ignore it.
The census is meant to be the single source of truth from which all other population-related metrics are derived, but since it only comes around once every 10 years it's usually out of date. In practice you always need a bunch of fudge factors to account for population growth, immigration, emigration, people moving around, etc in the years since the last census was taken. And that's not even accounting for tourists and so on. I've seen how this sausage is made; you just get used to the idea that the census info is merely a rough guess that you have to finesse until the numbers look reasonable. Those fudge factors are themselves often taken from ONS projections. So if the ONS itself has a small systematic undercount because of an overly-optimistic methodology, that poisons all calculations downstream.
As for the registers: the point in that thread is that the registers disagree with each other. Covid-era NHS patient counts, for example, because it's obviously in people's interest to get healthcare in the way it isn't to fill out a census form. EU settlement applications is another one.
Surveillance cameras: do you know what the "CC" in "CCTV" means? It's not all fed into a giant database. There's nobody combing through all that video and counting how many people are there. In almost all cases the data is retained for a few days and then gets destroyed with no human eye ever looking at them.
For all the talk of mass surveillance, there is still an enormous world that lives hidden outside the notice of the state. You only see the outlines of it during unusual events like Covid or Grenfell, like a lightning bolt suddenly illuminating the landscape for a split second. There were dozens and dozens of people living in that tower that the local authority had no idea existed. We only know about it because it went up in flames -- it can't be unique.
Edit: Economists are the modern first estate: the statistical clerics preaching the benefits of greed. Now with alternative facts!
Old title about 200 years claim: While the Middle East, Egypt, and Greece contain the longest continuously occupied cities, the nations and people who compromise them changed and lost generational memory over the centuries. And few continuous civilizations survive in any meaningful form for more than 250 years / 10 generations on average.
> the statistical clerics preaching the benefits of greed
You have to define greed very broadly to make this statement true. Broadly enough that you should tell people to discard any negative connotations of the word when interpreting what you're saying.
I work for money I can spend; that's greed only in the sense that I choose what I do and I get rewarded for it, as opposed to working to avoid being whipped or killed or sent to Siberia or to hell. Self-interest is so far the best motivation we've come up with.
That means that, as much as they try to stuff it with math, economics remains fundamentally an application of loose sociologic principles, which are themselves extremely debatable.
The genius is that that bit of casual math instantly insulate economists from their peers (i.e. historians, sociologists, etc), no matter how bogus such math might be.
The problem is economists pretend they are a hard science, while their founding works are largely just assertions from randos in the 1600s. There's no scientific basis for any of it, and plenty of economics still stand on the stilts of things like "rational actors" and other handwaves that they use to pretend their terrible toy models can represent reality. They also do things like define market efficiency in circular ways.
Even game theory sits on top of the throne of "rational actors", meanwhile human beings are explicitly IRRATIONAL actors.
If economics was a functional science, the richest people in the world would be economists. That it isn't is strong evidence that no economics yet has produced strong predictions, which is a rather significant problem for a "science".
It's a question of predictive value. Economics definitely has predictive value. Eg supply and demand, the emergence of money, price elasticity etc.
Economies in video games fascinate me, because it's another 'world' where you can see whether some economics principles apply or not. Eg the game Path of Exile doesn't have 'money', but various crafting materials ended up being treated as money by the players. The exchange rates between them are set by supply and demand as well.
The problem is lots of economists at least seem to think that what makes something a science is "using math" rather than "validating or disproving theories via observation". I've even seen people sneer "mere heuristics" at suggestions to try that...
It's quite surprising because ever since the "credibility revolution" [0], a major critique of economics is that it focuses too much on issues that are well suited to "validating via observation" as you call it, rather than issues that are actually important. For instance, more looking at the effect of soccer victories on inter-ethnic trust [1], fewer grand theories of development.
Papua New Guinea’s population was significantly larger than thought - closer to 18 million than 9 million. (https://www.dailymail.co.uk/news/article-11507443/amp/Papua-...)
Suffice to say, this is a problem us Economic Historians and many other social scientists have been aware of for some time.