Clearing out some old links

Solution here.

A neural net creating names for British towns. There's an associated Twitter account, though it does lose its charm after a while.

A lovely old profile of Derek Parfit (pbuh) and Janet Radcliffe-Richards. I don't really know the work of the Churchlands, but it's in my links folder so let's unload that too.

A brief discussion of the value of art, very interesting in the light of Hanson and Simler.

HAIL SATAN!

Some wholesomeness for you

Photos of a real war fought with bows and arrows. (Content warning: unpleasant photos of wounded people).

This is the future liberals want.

At least it's not brutalism.

The Pinkerian Case for Campus SJWism

Last week, courtesy of a commercial offer which I am shamelessly and ruthlessly abusing, I was able to attend a talk by Steven Pinker discussing his new book Enlightenment Now. I haven't yet had the time to look beyond the opening pages, so if you want a review on the book you should go to the one written by his ultimate fangirl. However, after the talk I was able to ask him the question:

"Many people who accept the trends you point to argue that due to the decline of religion and of thick communities, it is harder for individuals to find meaning and purpose in their lives. Do you agree with this assessment, and either (a) why not? or (b) do you expect it to continue?"

He disagreed with this assessment, giving two counterarguments. The first, which I don't find especially compelling (although IIRC I found it rather more compelling when Peter Singer said the same thing in a book I was otherwise disappointed by) was that people can find meaning in making a better world in general. People are not, in general, motivated strongly by the prospect of making the universe better. (Ctrl-f "charity"). There definitely are some people who are, and more power to them, but I don't think universalism can play the role in people's lives that, for many years, deities did.

His second, more convincing response was that people are finding new ways to build meaning in their lives. The example he himself gave was social justice movements on campuses - a purpose which many people choose for themselves as a purpose to which they can dedicate themselves. People may no longer identify as Christians, but they are very happy to identify as feminists.

Review: The Most Good You Can Do

I originally wrote this review in June 2015 for what was intended to be a collection of reviews of books with interesting and/or provocative these. Unfortunately, the person who was organising the collection did not manage to publish it before they left the ASI; I was reminded of this book by another discussion, and so am making the review generally available. This is the review as I submitted it, without any changes.

Peter Singer achieved prominence as a moral philosopher in the 1970s with a series of books and articles arguing for controversial positions in impeccably logical fashion. One article in particular, Famine, Affluence and Morality (1972) argued that as members of rich, developed nations, we have strong duties of rescue to people living in less developed countries. This line of thinking has spawned the Effective Altruism movement, a set of groups whose members are pledged to ending poverty, saving the world, and in general averting suffering wherever they see it. Effective altruists, due to their focus upon concrete impact, think and act very differently from members of other charitable movements. The Most Good You Can Do functions as an introduction to this movement, presenting an introduction to and defence of its main beliefs and practices.

The opening chapters give a brief description of the movement and of how it came about. This includes some of the controversial claims to which effective altruists tend to subscribe – notably, that one is unlikely to achieve a great deal of good by working for a typical charity. When one is employed by a charity, this is likely to fill a role in the charity which could equally well have been done by any other volunteer. If one instead finds a well-paid job and donates money to the charity, the net positive impact of one’s career is likely to be far greater. This has led to some effective altruists seeking out employment in financial trading, despite the rather poor reputations held by financial firms regarding the morality of their practices.

The second section of the book deals with some of the specific actions taken by effective altruists. These include reducing one’s consumption in order to give more, seeking high-earning jobs, and donating organs. The chapter on earning to give contains the first seriously philosophical sections of the book, a response to objections made by David Brooks and by the ghost of Bernard Williams. In response to the idea that earning to give sacrifices one’s integrity and alienates a person from their personal goals and projects, Singer claims (without much in the way of argument) that merely “doing good” is a perfectly adequate goal for one’s life – in which case earning to give, far from representing the subjugation of one’s aims to an imperative to maximise global utility, can be the ultimate expression of authenticity.

In response to the idea that going into finance upholds and strengthens the system of capitalism which impoverishes many and drives inequality, Singer engages in a brief defence of capitalism, pointing to the fact that it has “lifted hundreds of millions out of extreme poverty”. Finally he considers the idea that going into finance harms people, and that ‘do not harm’ ought to be prior to ‘do the most good’ as a principle of morality. Singer questions this priority with an example drawn from the London Blitz, but seems to devote more attention to attacking the account of harm upon which the objection rests. It is unclear that Singer needs to defend earning to give against these specific objections – while finance is one career path for someone who earns to give, there are after all a range of alternatives including law, consultancy, and entrepreneurship.

In addition to these, Singer discusses a range of other careers in which one’s impact might be directly through the work – among others effective altruist advocacy, jobs in aid organisations, and medical research. Finally, he discusses the good one can achieve by donating parts of one’s body. Since many people are unwilling to donate kidneys except in exchange for kidneys to save the lives of their own friends and family, someone who is willing to donate a kidney without attaching conditions can start a “kidney chain” of multiple donations, perhaps saving five or six lives through a single donation. Unfortunately the number of such donors is currently small (117 in the UK in 2013; the US figure, adjusted for population, is worse), not helped by the fact that until 2006 such donations were in fact illegal in the UK.

The third section of the book discusses the factors which motivate effective altruists to undertake apparently sacrificial actions purely in order to help others. Singer suggests that the emergence of effective altruism represents a triumph of reasoning over emotion, and presents a range of evidence to show that members of the wider population are usually moved to act altruistically more out of instinct than out of reasoned consideration. He also argues that we tend to overestimate how much happiness we will lose out on by giving away money and to fail to recognise the sense of purpose and self-esteem which many people gain from helping others.

The final section of the book presents perhaps the most controversial claims which effective altruists universally take for granted: that some charities and causes are simply better than others. Singer observes that, while poverty and suffering exist the whole world round, it is generally a lot easier to relieve them in the third world than in the first world. Singer compares a program of Rubella vaccination by philanthropist Ted Turner, estimated to have prevented around 13.8 million deaths between 2000-2012 at an average cost of $80 per life saved, with a 2007 operation which separated two conjoined twins from Costa Rica at a cost running into millions of dollars.

After sharply criticising the practice of spending megabucks on improving museums while there are starving children in Africa, Singer turns to some issues which are not universally accepted even by effective altruists. The first is animal rights; the second, the perhaps less familiar subject of existential risk. Given that (hopefully) the vast majority of humans have yet to exist, one of the biggest threats to the sum of human wellbeing is the risk of becoming extinct. Efforts to reduce the risks of nuclear war, asteroid impacts, and unfriendly artificial intelligence, then, could be a remarkably effective form of charitable giving.

All in all, The Most Good You Can Do is very readable and serves well as an introduction to the effective altruist movement. Even as someone who has been involved with effective altruism for almost two years, I learned things from reading it. Since the book is more a summary of existing arguments than an attempt to break new ground, the arguments made are perhaps not as strong as one might expect, with an often unnecessary reliance upon utilitarianism.

Review: Jorge Luis Borges, Ficciones

Borges is a writer who I had been somewhat aware of for a while, read a few passages of and enjoyed, but never got around to reading deliberately. So when I set out a reading list for 2018, his collection Ficciones, generally regarded as the most accessible starting point in reading him, seemed an obvious inclusion.

Ficciones is a set of seventeen short stories, originally published in two separate volumes in the 1940s and then later collated; they first appeared in English in 1962. Borges wrote in Spanish, though he was heavily influenced by English writers, in particular G. K. Chesterton. There is a tremendous playfulness in many of Borges' stories, exemplified by my favourite story from the collection: Pierre Menard, Author of the Quixote. Pierre Menard is a deceased author and the story is an appreciation of his work, in particular of his greatest project: an attempt to rewrite Don Quixote in the exact same words used by Miguel de Cervantes. The narrator of this story therefore takes Menard to have made a conscious decision to write not in his own native land and time, but instead in "the land of Carmen during the century of Lepanto and Lope". His Don Quixote is a wild and romantic figure, in contrast to de Cervantes' more pedestrian protagonist.

There are other wonderful stories. The Library of Babel is an excellent counterpoint and companion to Pierre Menard, Three Versions of Judas is another masterful piece of intellectual trolling, and The Form of the Sword and Theme of the Traitor and the Hero provide scintillating plots which blast by in only a few pages. Borges' writing style goes after my own heart, with numerous allusions to both the real and the imaginary. But the quality is distinctly uneven. The Circular Ruins is eminently forgettable. The End will seem quite pointless to anyone who is not already familiar with the Argentine national epic Martín Fierro. Perhaps the biggest disappointment is The Secret Miracle, a story about a Jewish playwright in 1940s Prague struggling to complete his masterwork before his execution by the Nazis. By a miracle he is allowed to unfurl it all to its conclusion, to put each word into place - but only in his head, and it dies with him. There's a fantastic basis for a story there, but it seems so incomplete. One might argue that the point would be spoiled if we were to know what this play is about, but I'm not buying that - we already know that he was granted this miracle to complete it, something which no-one else inside the story would have been privy to. So the content of this play seems like a massive missed opportunity to draw parallels with the greater story, to exude some moral about life, or to draw some dramatic irony with the situation in which the playwright finds himself.

Indeed, with several of the less allusive stories one begins to wonder why one does not simply read the Wikipedia page for each of the stories. Perhaps one does not gain so much intellectually from reading Pierre Menard that one could not also learn from the Wikipedia page, but Borges' charming voice makes the extra reading time well worth the investment. Some of the better stories combine abstract theorising and an actual story, again making them worth the time to read properly. But unless one enjoys all of the writing styles which Borges employs, one is liable to find some of the stories to be distinctly full of air and little else.

Overall, I definitely recommend the book - if nothing else, most of the stories are pretty short and there's a pdf of Borges' collected works to be found on Google for free, so the costs of trying him and not enjoying it are trivial. More importantly, while there are some dull stories the greatest stories are magnificent, and the good significantly outweighs the mediocre. But if, after a couple of pages into a story, you still have no idea what it's actually about, take that as a sign that it may be worth skipping ahead to the next one.

The Elephant in the Brain: some notes

The Elephant in the Brain, written by Kevin Simler and Robin Hanson and recently out in paperback, is best viewed as two books on connected topics. The first is a convincing argument that "the elephant" exists: that we consistently engage in self-deceptive behaviour for purposes of social gain. The second is a serious of arguments, ranging from the highly plausible to the outrageous, that this explains various the function (or malfunction) of various human social behaviours.

The Elephant

The first section of the book presents multiple lines of argument leading inexorably to the conclusion that many of our behaviours are inexplicable in the first person but are on some level intended, in a way that a third party might easily observe, towards attaining social advancement. This is made possible by the modular structure of the brain, in which sections of the brain may have the ability to make decisions but not to communicate them or defend them. Crucially, we are unable to distinguish between those actions caused by the parts of our brain which also control what we say and those caused by other parts of the brain: hence, we will typically invent justifications for such actions which will own nothing to the actual motivations behind them.

The upshot of this is that one section of the brain can engage in devious, cynical scheming, and we are free to act upon this advice while having no conscious awareness of it, and therefore being able to honestly protest complete innocence when accused of holding these devious and cynical motives.

Hanson and Simler present a range of evidence for this, which I won't reiterate partly because other reviews will cover it and partly because I didn't take very good notes and really need to reread this section of the book. What I do remember finding illuminating, however, is the way they placed features of humans in the wider context of nature. Why is the American Redwood tree so tall? On clear and flat ground, being taller doesn't allow a tree to get any more sunlight but it does mean that the tree has to acquire more nutrients and transport them further upwards. The answer, of course, is that redwoods don't originate from clear and flat ground: they have to be as tall as, or taller than, the trees around them in order to have access to sunlight. The redwoods become so tall because of competition with other redwoods.

Similarly, how did humans become as smart as demonstrated by the graph above (taken from the book)? The answer lies in not in the abilities it grants over nature, but in competition against other people. This thesis is not new to Hanson and Simler, of course, but their presentation of it is especially clear.

I have some further thoughts following from the discussion of norms and how we subvert them, but they are not developed enough to appear even in this miserable excuse for a book review.

The Elephant in Practice

There then follow ten chapters, each discussing a different phenomenon from a Hansonian perspective. I don't want to go over all of these, so will briefly look at two that I found especially interesting. Firstly, they argue that laughter - which we often struggle to explain, of course, so looking for hidden motives may well be the way to go - serves the function of signalling that we are "at play". When one laughs, this indicates to those around oneself that one is not in a serious mood, which can allow one to say or do things that would normally be taken as threatening.

This theory is fascinating, and for lack of a better theory has changed my view on at least one issue: rape jokes. The ability to laugh at something is an indication that one is not concerned about it - if this theory is true, then, we should probably consider dark humour to be indicative of a lack of virtue, and indeed to actively discourage such a lack of caring in others. Perhaps this doesn't merit an absolute prohibition on such jokes - humour is a value which can weigh against other considerations - but it does suggest that we should be very cautious with such jokes and should never consider rape in itself to be suitable for a punchline.

There's also a defence of canned laughter, which I don't remember well enough to faithfully pass on.

The second section has already gained some attention when I shared a page from it on Twitter: their theory of art. This theory, originally developed by Geoffrey Miller, is that art developed primarily as a way to show off various attractive traits - in particular intelligence, creativity, and conscientiousness. They draw a distinction, which I assume must have been drawn many times before, between the intrinsic and extrinsic properties of an artwork. Intrinsic properties are those that we perceive in an artwork, extrinsic are those that which cannot be known - primarily facts about how it was created. Quoting directly:

The conventional view locates the vast majority of art's value in its intrinsic properties, along with the experiences that result from perceiving and contemplating those properties... In contrast, in the fitness-display theory, extrinsic properties are crucial to our experience of art. As a fitness display, art is largely a statement about the artist... If a work of art is physically (intrinsically) beautiful, but was made too easily (like if a painting was copied from a photograph), we're likely to judge it as much less valuable than a similar work that required greater skill to produce.

This has the consequence that as our ability to produce things has improved, artists have had to find new ways to make art difficult for themselves. They offer this as an explanation for why theatre continues to be popular, despite the various capabilities (camera angles, numerous takes, vast amounts of post-production editing) that film offers: it has the chance to go wrong, and so demands greater skill of the performers. I think this is not the whole story (and nor, for that matter, is Michael Story's theory that theatre serves to make lowbrow comedy acceptable for the middle and upper classes) - theatre offers advantages in terms of one's ability to focus on whichever section of the stage one prefers (regardless of whether or not, artistically speaking, it is the best), and the ability to tailor to particular performances (theatre actors can wait for laughs to subside, film actors can't). But it's a fascinating view on the topic.

As I suggested on Twitter, I am only partially sold on this. How good are audiences at realising that mistakes have been made? Sometimes it's clear - for example, a playgoer may see an actor requesting a line from the stage manager (I didn't see this happen when I saw Twelfth Night at the RSC the other day, but it happened very obviously a couple of months ago when I saw an amateur production of Arcadia) - but much modern art is highly abstract. If one of the lines on Jackson Pollock's No. 5 is out of place, how shall we know? If someone gets the timing wrong or plays the wrong note in some atonal piece of music, will anyone without a score be in a position to check?

I have some other thoughts on this in regard to popular music, which will be a post of their own because they're worth actually developing. For now I'm just going to raise three questions which I think are worth asking of the authors:

How sophisticated is the elephant, anyway?

Some of the signalling stories which Simler and Hanson tell are very complicated. For example, they argue that much advertising works not by influencing us as individuals, but by causing us to expect others to be influenced by it:

When Corona runs its "Find Your Beach" ad campaign, it's not necessarily targeting you directly - because you, naturally, are too savvy to be manipulated by this kind of ad. But it might be targeting you indirectly, by way of your peers. If you think the ad will change other people's perceptions of Corona, then it might make sense for you to buy it, even if you know that a beer is just a beer, not a lifestyle.

The classic strawman of evolutionary psychology is that almost no-one has a conscious aim of maximising their genetic footprint. The chain of reasoning "I will do X, because X will make me more attractive, which will allow me to attract a higher quality mate or to attract more mates, which will increase my genetic footprint" will almost never include the less clause, and may not even go beyond "I will do X" if X is something we are inherently motivated to do. The answer, of course, is that we don't need to think everything through - so long as a category of action reliably leads to higher fertility, we may well find ourselves inherently motivated to do it. This explains desires to eat and drink, to have sex, to parent our children well, and many other things. But these things which we are inherently motivated to do are fairly broad classes of action, with no particular cultural knowledge required. The Corona example is actually highly sophisticated cognition, involving not only instrumental rationality but also a theory of other minds. Do Hanson and Simler think this is all being done non-verbally, by evolved instincts - or is there a portion of the brain thinking thoughts, in a verbal fashion, but entirely detached from our stream of consciousness?

How far do signals rely on common knowledge?

Another example from their chapter on consumption:

Blue jeans, for example, are a symbol of egalitarian values, in part because denim is a cheap, durable, low-maintenance fabric that makes wealth and class distinctions hard to detect.

I had no idea about any of that. Indeed, I doubt most people consciously pick up on most of the signals which Simler and Hanson allege we send. So how far can we actually be expected to react to them?

Signalling vs. Creating Meaning

Depending on what kind of story we tell, the same product can send different messages about its owner. Consider three people buying the same pair of running shoes. Alice might explain that she bought them because they got excellent reviews from Runner's World magazine, signaling her conscientiousness as well as her concern for athletic performance. Bob might explain that they were manufactured without child labour, showing his concern for the welfare of others. Carol, meanwhile, might brag about how she got them at a discount, demonstrating her thrift and nose for finding a good deal.

If so many different messages could be sent by the same purchase, then none of them will be sent. I think these are far better explained as the stories we tell ourselves in order to create a sense of meaning and purpose in our lives. Once one raises this spectre, one wonders how much of their theory it could take over. Is the extrinsic value of art not that it could go wrong and is therefore a display of fitness, but that the process of creation is a way of creating meaning? Perhaps creating meaning is just another form of signalling, but this is something that has to be actually argued for.

One piece of evidence in favour of signalling over meaning-creation theories of fashion is a dog that hasn't barked - decorating the inside of clothing. The underside of a shirt could have many messages, verbal or pictorial, that would be understood by the owner but not by observers. The fact that we worry greatly about the outside of clothing but not the inside suggests that it the impression given to observers that we care about.


Conclusion

The book is very readable, and if you like Robin Hanson's other writings you'll like this. That said, it didn't quite live up to the praise given to it by other sources (e.g. Tyler Cowen) - there are some excellent passages, and some wonderful ideas, but there are also many ideas which are in sore need of greater defence. It's worth reading, quite possibly more than once, but it is not - in my view - Book-of-the-year level good, which is the level I feel it has been hyped to.

We Already Have A Voting Lottery

When, on that fateful day last year, the UK voted to leave the European Union, there was a great deal left as yet undecided. There were a great many paths we could have pursued, ranging from the Norway+ options that would have removed us from the European parliament and little else, to the economists' nightmare No Deal scenario. None of these options could at that point be declared "undemocratic", since the referendum gave us an answer to only a single question. Theresa May - or whoever else might have become Prime Minister - could have, if they so wished, declared that "52% to 48% is no mandate for radical change. We will leave the EU, but smoothly and cautiously" and changed very little.

Instead, the narrative very quickly became that the referendum had ultimately been about immigration, and that the British People had ultimately voted to Take Back Control Of The Borders. It's not hard to see why the notoriously anti-immigration Theresa May wished for this narrative to prevail. Moreover, it's not totally absurd - the campaign for Brexit did, after all, emphasise this as a reason in favour of Brexit (though by no means the only one, or even the main one - remember that bus about the money which could go to the NHS?). What is puzzling, however, is the lack of pushback against this narrative. Theresa May may not have wanted to suggest the vote was anything less than an endorsement of radical change, but why have so many other actors, including many who are in principle in favour of immigration, colluded in this narrative and not challenged this interpretation of the vote?

The answer, of course, is that it is a correct interpretation. Not that you could tell this from the Brexit vote alone, of course - referenda are, much like general elections, a quite incredible effort to extract the minimal possible sliver of information from voters. But we have a great many other surveys and polls of public opinion, conducted with great regularity and on a much richer array of questions than the usual choices offered at the polling station. We know that Brits want less immigration, but this is not because given a choice of two highly uncertain prospects, they chose the one likely to involve less immigration: rather, it is because people from YouGov have asked them exactly what their views of immigration are.



An alternative to universal suffrage that almost only ever appears in the academic literature is the "voting lottery". The idea of this is that rather than collect votes from every single person, we select a certain much smaller number of citizens - say, 1000 - and only ask their votes. This would have three key advantages: firstly it would be cheaper ("But you can't put a financial value on democracy!" "Sure you can. In 2011 we rejected AV because it would be too expensive.") Secondly, by increasing the power of those who actually get to vote, it would give them more of an incentive to seriously consider their vote and its impact. Perhaps most importantly, it would provide an opportunity to stratify the sample of voters. Currently certain groups - in particular, the young and various ethnic minorities - are grossly underrepresented by democracy because of their lower turnout. A voting lottery would allow us to ensure that these groups are counted in accordance with their proportion of the total population, not merely the their proportion of the population that turns up to vote.

Now of course many people who encounter this idea have a strong aversion to it. The point of democracy, they say, is in the mass participation. But the fact that our assessment of public opinion comes not from five-yearly general elections but from weekly polls rather pulls the rug out from under this. Voting in an election is screaming into the void: real political participation is happening to be selected for a YouGov survey, and giving your opinion there.

Another common concern is that a sample of 1000 people cannot hope to fully capture the views of an electorate of millions. I'm not married to the 1000 number - in fact, I think it could stand to be more like 10,000. But the basis of all modern polling is the Law of Large Numbers, which in essence states that when you have a process consisting of many small things which are themselves error-prone, but whose errors can cancel each other out - the errors will tend to cancel themselves out. Hence a poll of 1000 people will be within 3% of the true values 95% of the time, and a poll of 2000 people will be within 2% 95% of the time, for example. Yes, the newspaper polls can be wrong, but this is more often due to bias in the way they have selected voters - asking by telephone, or at a particular time of day, or with certain incorrect assumptions about who is likely to vote - which we could eliminate by selecting voters directly from government records.

It ought to be clear where this is leading. How about, rather than maintaining our thin veneer of universal suffrage with all its attendant problems of unrepresentativeness, we acknowledge the fact that we already live in a political system dominated by the voting lottery, and adjust accordingly? Of course there are costs, but there are also real benefits, benefits which we would be much better equipped to realise if we were honest about our political system and learned to live with it.

People have every reason to worry about attempts to disenfranchise them. In the USA, very significant effort goes into attempts to disenfranchise black voters due to their tendency to vote for the Democrat party. But the voting lottery is different both in its intention and its effects: while we would stop even pretending to care about most voters as individuals (as though we ever could in a nation of 65 million!), we would give much greater weight to their views as members of groups. And voting is far from the only way to engage in politics: so long as we have free speech and a free press, those who are not randomly selected for the ballot will have the opportunity to influence the votes of those who do vote though force of persuasion.

There's a Chesterton-like paradox to the suggestion that we should improve our democracy by removing the votes of most citizens. But the idea, I maintain, is not ridiculous. Certainly no more ridiculous than the idea that rather than vote on every single decision, we should delegate this to some 650 people, mostly white men, all living primarily in central London. I urge you to consider being explicit about the voting lottery which we already have - and to consider how it might be put to better use.

A Retraction and An Apology

Several months ago I wrote a mostly-serious essay arguing the cosmopolitan case for #SpendTheSix. In one line I claimed that:

Typical practice during the days of the old Empire, as best we can tell, was to spend around 7% of GDP on the military.

I can't remember if I made any attempt to check this claim at the time, but it seems unlikely. It was half-remembered from a book I read sometime in my teenage years - most probably Niall Ferguson's Colossus: The Rise and Fall of the American Empire. As my memory records it, the book was briefly discussing whether continuing US military dominance across the planet was financially viable, and argued that the US spends around 3.5% of its GDP on its military compared to the 7% or so which the old British Empire spent. Hence whatever barriers there were to continued US hegemony would not be financial, etc etc. It should be noted that not only I am only about 70% confident that Colossus was the book in question, it is entirely possible that either I misread it at the time or that in the seven or eight years since my memory of the factoid has become confused. Certainly I do not wish either to accuse Mr. Ferguson of making this claim, or to suggest that my failure to properly check the claim when I wrote the essay was in any way excusable.

My attention was drawn again to this claim when, browsing Andrew Sabisky's Curious Cat, I discovered that he had cited this essay for the claim that "we did in fact historically spend the six, & not just during the cold war either". It occurred to me that I perhaps ought to check the veracity of this claim, so quickly googled "historical british military spending". From the results it seems clear that I could not have done this when I wrote the original essay. First, this article on the BBC website:

"It's often thought the British army in the 19th Century just mowed down natives with a machine gun. This is a myth," says military historian Nick Lloyd.

"The most remarkable thing is that they often had no technical advantages and we managed it by spending only 2.5% of GDP on defence, which is not much higher than we have today."

Second, ukpublicspending.co.uk:

Defence began in 1900 at 3.69 percent of GDP but quickly expanded during the Boer War to 6.47 percent. After the war it contracted down to about 3 percent of GDP.

Third, this fascinating graph from ourworldindata.org:

The first part of Sabisky's statement is supported - we have historically spent the six. Technically the second holds up in that we also spent the six during various wars, but I think that it would be fair to characterise this as misleading.

For what it's worth, I don't think the overall thrust of either Sabisky's or my argument is hurt to any great extent by this fact turning out to have been false - neither of us was arguing that we ought to spend the six because the Empire spent the six, merely trying to suggest that in historical perspective the claim would sound less absurd than it does to people who have only known the world of today. Nevertheless, it is entirely clear that I ought firstly to retract that claim, and secondly to apologise - to anyone who read my piece, to Andrew Sabisky, and to anyone else who encountered the claim indirectly through him or some other intermediary. It was not my intention to mislead, but I ought to have practised higher standards of scholarship - and hope that in the future I will do so.