A Persian Cafe, Edward Lord Weeks

Tuesday, 27 May 2014

Summer Reading List

My blogging has been lighter than usual or recent, primarily because I've been going through various exams. The final one was this morning, which means that I now have a fair bit of time on my hands. I'm sorting out a list of books I intend to read and review over the summer; by review I mean chapter-by-chapter or section-by-section, over at my other blog (which has also not seen much use, albeit for rather longer than this one). At the moment, my reading list is as follows:

Thinking, Fast and Slow by Daniel Kahneman. I've already read most of this, but I'm intending to re-read it from the beginning and take notes.
Decisive: How to make better choices in life and work by Chip Heath and Dan Heath. I know next to nothing about this book, but it was on the CFAR recommended reading list and I feel like I've tended to focus rather more on Epistemic Rationality than Instrumental Rationality, so hopefully this will help my instrumental rationality.
How to Win Friends and Influence People by Dale Carnegie. This wasn't on the CFAR list, but I've seen recommendations elsewhere and I'm reading it for pretty much the same reason.
Pop Internationalism by Paul Krugman. I've been worried for a while that I read a lot of economics by libertarians, and very little by lefties. I don't tend to read Krugman because a) there's a paywall after your first ten New York Times articles each month, and b) the relentlessly, aggressively political nature of his writing nowadays. This is a collection of his essays from (I believe) the late nineties, before he went all political.
Poor Economics by Abhijit Banerjee and Esther Duflo. A friend of mine who founded the Manchester branch of GWWC recommended this to me. I would like to have read this before the Effective Altruists Weekend Retreat in late June.
Anarchy, State and Utopia by Robert Nozick. I've been intending to read this in full for a quite considerable while, and indeed I already have notes on most of the first section. Most likely I will re-read those sections but will not worry about writing new notes on those sections.
An Economist Gets Lunch by Tyler Cowen. My mother has spent much of the last five years working towards a qualification which has boosted her pay; now that it's over, we're having a holiday - only our second foreign family holiday - in Canada. This will involve a fair amount of eating out, so I'm intending to finish this book before heading across. (I wouldn't take it across with me - my copy is in hardback, and there are weight/size limits regarding baggage on the plane).
The Strategy of Conflict by Thomas Schelling. I've seen this recommended in various places, and again hopefully it will contribute towards my instrumental rationality.

This list may be added to, although I want to avoid making it too large in order to keep my goal realistic. The main things I am considering adding are (a) a work of fiction, and (b) a textbook on computer science.

If anyone happens to read this, I would love to receive your recommendations.

One other project I am considering for this summer is that, having been through pretty much all the Sequences concerned with epistemic rationality and noting that there is rather less concerning instrumental rationality, making a list of all the important LW posts regarding instrumental rationality and recording audio versions of them. I am definitely intending to make a poster of the key points of Humans Are Not Automatically Strategic, which I shall hang up in my room. This should give me greater recall of the effective habits and increase the likelihood that I will actually act strategically in practice.

Addendum: I am also aiming to read Be Still, My Soul (editor: Nancy Guthrie) and The Bible Unearthed by Israel Finkelstein and Neil Asher Silberman, for reasons explained here.

Saturday, 24 May 2014

Under Anarchism, Who Will Build the Roads?

Previously most of my writing here has been a) as a record of my thoughts and b) for anyone who stumbles across it. I intend to link to this post elsewhere, so it is somewhat different from many of my other essays here in its formatting.

But if there's no government, who will build the roads? This is apparently a question which genuinely gets asked, so I'm going to attempt to provide my answer here. [1] Or rather, I'm going to respond with a question of my own: why are roads different to the many other goods which we agree are best provided by the market?

Pictured: a list of things best provided by the free market.

One possible response for someone who hadn't really gone into the theory would be simply to suggest that a) roads are in practice provided almost universally by governments, and b) there is probably a good reason for this, hence c) we have reason to believe that roads are best provided by government. (It's true that there are private roads, but given that they are nearly all on private land I don't think this constitutes a real objection to the argument).

Just as I am wary of Austrian economics de to the general disregard of its adherents for empirical testing, I am wary of this kind of argument due to its lack of a theoretical explanation - one may as well label the proposed mechanism by which government intervention improves welfare as "magic happens". (This isn't intended as a strawman - it is better epistemic practice, when one is confused about a mechanism, to label it as "magic" than to dream up some believable but probably untrue explanation or to attach a meaningful-sounding buzzword which may convince you that you understand it). That said, this doesn't mean it is wrong.

There are two counterarguments which occur to me. The first is simply that there are many industries which are systematically dominated by the public sector when it is far from clear that this is a good thing - the best examples being currency, law and education. The second would be to note that the industries which are controlled by governments are often very important ones for the purpose of controlling a society. A society would be worse off for the loss of its pasta industry, but such a loss would be survivable; the same could not be said for currency, rights enforcement, or transportation. (Education - at least beyond a certain level - is probably not on the same level of importance as these, but let's not pretend that the primary reason for state intervention in education is anything other than a desire to indoctrinate children).

With that out of the way, let's look at the actual theoretical objections. The Wikipedia page on Free-Market Roads gives two objections: that roads are a natural monopoly, and that road privatisation would adversely affect the poor. I'll explain why I don't take either of these objections particularly seriously, why I might have taken the first one seriously a few decades ago, and I shall raise a (to my knowledge) completely original consideration which I suspect may provide a significant tendency towards less competition, but has only just occurred to me while writing this and I need to think about more.

Are Roads a Natural Monopoly?

"In many parts of the world land use patterns mean that building two or more highways in parallel isn't practicable." Hence, the market cannot be contested and a monopoly arises. This is quite obviously nonsense, and to see why let's look at some of the most valuable land on earth: the Upper West and East Sides of New York City:

Do you see all of those roads in parallel? If they can get that many parallel roads on estate like that, I find it completely implausible that they can't do it pretty much anywhere else. After all, it's not like roads take up all that much space - if you turn a road into a tunnel you can always build on top of it. No, the idea that the roads cannot be built in parallel is rather silly.

It is true that there are significant costs to building a road so there may well be examples of local monopolies in formerly state-controlled areas with only one road. Suppose, however, that you were a private developer of houses or retail outlets. You would want to avoid large road fees nearby: retail and transport to retail outlets have joint demand, hence a rise in the price of one reduces demand for the other; similarly, houses and transport to/from those houses have joint demand. For this reason you would be keen to set up a competitive market for transportation to the places you were building.

A stronger argument would be that, rather than individual roads being examples of natural monopolies, networks of roads would be natural monopolies. It is not feasible to have a toll booth every time you move onto a new road, at least within an urban setting: having to show your card (or windscreen sticker) to someone at the end of every road would massively slow down your commute to work. Moreover, if one person doesn't have the relevant permission then this creates potential for a massive backlog which would hold up everyone behind him.

This would have been a good argument until the development of road cameras. In the modern state, these cameras are used primarily to catch people speeding; in Ancapistan, they would be used to check who was using a particular road so that they could be charged after the fact. There would of course be issues of people using someone else's number plate, but there's no reason why that problem should be any less rare in Ancapistan than it is in current societies. This would remove the need for people to stop or slow down excessively when moving between roads owned by different people.

This, my friends, is what freedom looks like.

There is still one possible issue: houses tend to be accessible to only one road, so we might expect these fees to be excessively high. The most likely answer is that residential roads would be owned by homeowner associations. These don't tend to exist in the UK and I've heard bad things about them from the US so I'm not entirely happy with this; all I can say is that monopolies cause deadweight loss, so if there is an alternative allocation of goods which leads to greater social welfare this will tend to be realised within the market - goods go to those who value them the most.

Would a free market in roads hurt the poor?

The description of this on Wikipedia is very vague. I'm not certain whether the objection is that markets cause poverty - in which case a) no they don't, and b) even if they did it would be far better to carry out redistribution purely in cash rather than through a single market - or that the tendency of pricing would be to make roads relatively more expensive for the poor. 

I'll come back to that suggestion, but first let's think for a moment: how much would road access actually cost? I can think of four costs associated with people using roads: the cost of building them, the cost of maintenance, congestion caused for other road users, and the rent on the land taken up by a road.

The cost of building a road will affect market structure due to sunk costs, but since it does not affect the marginal cost of a road it will not affect the price of a road for a given market structure. As established above, we should expect the market for roads to be competitive.

The cost of maintenance cannot be all that big considering the amount a road gets used. This document is unhelpful, which is hardly surprising given its origin; this report gives Highways Agency maintenance spending as £663 million in 2014-15, which even assuming that government does it just as efficiently as private actors would, adds up to about £10 per person in the UK - perhaps £20 once we filter out non-drivers. Remember that this is an annual cost; hardly breaking the bank.

Of course, the government is known for the efficiency of its road maintenance.

What about congestion? I have no idea how to measure it, but my intuition is that it is pretty large relative to maintenance. Why? When you contribute to congestion for others, you also suffer it yourself. Suppose that by spending twenty minutes in a traffic jam, you slow down 120 other people's journey's down by ten seconds; in that case, you will inflict congestion equal to that which you suffer. If anything, I would expect you to slow down a greater number of people by a similar amount of time, in which case it only takes two hours of being in a traffic jam in a year to make this exceed your share of road maintenance cost. 

Finally, rent. This is again hard to measure, to a large extent because of the extensive existing government intervention in housing and land causing the relative prices to be all out of whack. Furthermore, it's difficult to know exactly where roads would be built (affecting the cost of land) and whether they would be turned into tunnels (affecting the amount of land used).

Will any of these bias prices against the poor? The cost of road maintenance shouldn't, since that is determined to a large extent by weather patterns; rent should work in their favour, since they would a) tend to live in and therefore drive on cheaper land, and b) it would be more worthwhile for them to take a long-cut through cheaper land. The costs of congestion are largely non-monetary, but the biggest determinant of their value will be the value one places upon one's time - which will of course be smaller for poorer people. Hence, in a free market we should expect poorer people to face generally lower fees for road usage than rich people. This is because ultimately they are buying a different product - they are paying for access to a different set of roads, and so just as one pays less for a lower-quality car, a poor person in Ancapistan would face a lower charge to use the roads than a wealthier person.

A new consideration

Roads aren't just roads. The land underneath them tends to contain various pipelines and cables, because it's far easier to dig up a road than a house. It seems likely that in Ancapistan, ownership of the pipelines would be divorced from ownership of the roads above. But the owners of these will obviously have to interact - you can't just dig up someone's road whenever you need to.

"What's this about? Well, I just felt like digging up a road of a
Sunday afternoon. Not bothering you, am I?"

Furthermore, I would expect there to be a substantial difference between the average length of pipeline owned by a single firm and the average length of road owned by a single firm. If a firm owns all the pipeline between (say) the water treatment plant or reservoir on one end and people homes on the other, this will most likely go under a considerable number of roads.

I need to sleep before I think all the way through the implications of this, but it seems like there could be problems resulting from this. Suppose the existence of a pipeline from location A to location B would create value for pipeline builder Dana of £x. There are n road-owners, each of whom has the power to veto the pipeline. Dana must reach an agreement with every individual road-owner for them not to veto the project, and it seems like a sensible solution might be for Dana to pay each of them £(x/n). But suppose one of them decides to insist on receiving £((x/n)+y), where y > 0. Then it would be in the collective interest of the other road owners to reduce the amount they charge Dana by £y, but it would not be in the interest of any individual firm. Moreover, if one firm can raise its price and other firms will reduce their prices to match, then it is in the interest of every firm to raise its price and the pipeline will not be built because poor Dana has no way of making a profit.

I don't think a social expectation of not charging for pipelines to go under your road is in the least bit likely, since there are very real costs to their prescence. Tying ownership of pipes to the roads only creates the same problem in a slight disguise - you have a value £x being somehow split between homeowner Dan and his water-supplier, with n pipeline-section-owners having an ability to veto the pipeline.

Note that this problem disappears if n equals one: the road owner charges £x and the pipeline gets built. If there is only one possible route by which the pipeline could go and n is large (with "large" defined as "too many for them to effectively coordinate and impose discipline upon each other"), then it seems like it will not be built. If there are multiple routes then this improves the chances of one of them not having a large value of n, but note that essentially there is a significant tendency towards monopoly in that this gives the best chance of this water (or sewage, or whatever) pipeline being built.

This may be a more fundamental objection that the standard "roads must be a monopoly" arguments, since they rely upon ignoring the fact that roads must compete with trains, footpaths, cyclepaths and other forms of transportation, whereas this objection, having nothing to do with their use for transportation, evades those sources of competition.

If you don't have a pipeline to take away the sewage,
you end up with a house completely full of shit!


I have no doubt that roads could exist in an anarcho-capitalist society. However, I have serious concerns regarding their interaction with other utilities, which may lead to a monopoly situation and hence inefficiency. I have not put nearly enough thought into this to conclude that there is no solution, but I do not have a solution ready to hand either.

The author of this essay wishes to thank David D. Friedman for making generally available his paper "A Theory of the Size and Shape of Nations"; much of the thinking in this essay was influenced by his discussion of the effect of trade routes on state revenue.

Tuesday, 13 May 2014

Bailout guarantees as Newcomb's Problem

Suppose a government has the option of having a policy of bailing out the major businesses within a certain market if there is a major crash within that market, which will limit the damage caused by the crash and improve confidence in the absence of a crash. If the firms within this market become aware of this guarantee, however, they will take increased risk and so make a crash more likely. Averaging out probabilities and utilities, the result is something like this:

Average Social Utility                  Firms take risks               Firms do not take risks

Bailout guarantee                        -2                                    5

No bailout guarantee                   -10                                 0

For a given behaviour on the part of the firms (who are all assumed to act in the same way) the government is always better to operate a bailout guarantee; however, if firms are able to predict that the government will operate a bailout guarantee, then society is worse off for this. Does this situation sound familiar? It is Newcomb's problem in a very slight disguise. The view on Newcomb's problem which I espouse is that if it would have been useful for one to have pre-committed to acting in a particular way, one should act in that way even if it is not in one's short-term interest. (This, at least, is my understanding of Timeless Decision Theory - that the decision to perform action X and the decision to pre-commit to performing action X should be the same decision). In slightly more layperson's terms, one should aim to be the kind of person who one-boxes. Philosophers as a whole lean more against this view than in favour of it, but by far the most popular view is "other", which I take to mean "don't know".

Sunday, 11 May 2014

An actual defence of homophobes

A (himself gay) friend of mine wrote a blog post entitled "In Defence of Homophobes". I have a problem with is his post: it's not an actual defence of them, merely an argument that we should not be getting them fired from jobs (outside of politics) for their views. As it happens, I pretty much completely agree with him, but for the record, an actual defence of homophobes, one that actually attempts to defend the thing which distinguishes them from the population at large - their homophobia - instead of merely treating it as an unfortunate flaw, would look rather more like the following:

Any minority group within society which cannot interact romantically with non-members of that group places its members at a severe disadvantage. Such groups include homosexuals, most Christians, and probably a fair few other groups which I have less exposure to. In a society of 100 men and 100 women, where exactly 8 members of each sex are homosexual, each straight person has 92 potential partners whereas each homosexual has only 7. Straights are therefore able to be more selective about their chosen partner, and so their relationships are likely to be better optimised for happiness. Moreover, the assumption when meeting someone in whom you have a potential romantic interest is that they are straight, which means that in order to find a partner a homosexual will generally have to go to some specialised place. For example, this is why there are specialised nightclubs for gays - it's not that gays have systematically different tastes in music to straight people, nor is it that they are more fussy about the décor - it's to facilitate a hookup culture in which gays do not face a 90%+ chance of being rejected without any consideration.

Of course, homosexuality is not something that is chosen, so it is unfair to blame homosexual people for the position in which they find themselves. But that doesn't mean we can't try to shift people's preferences towards being more straight. If there is less talk about homosexuality, then it is at least plausible that reduced exposure to the possibility of finding someone of the same sex attractive will cause people to think less about being homosexual, act according to a "straighter" set of sexual preferences, and hopefully find more satisfying relationships as a result.

It should be noted that the argument would apply the other way if homosexuals were the majority and straights the minority. Indeed, the ideal would be for everyone to be bisexual, which would raise each person's number of potential partners in the example above from a mean of 85.2 to 199. But because a) currently the majority is straight, which provides as easy coordination point, and b) it's easier to reproduce (if not necessarily to actually raise children) within the context of a straight relationship, it just so happens that barring a major shift in favour of bisexuality, it makes sense to push towards a greater homogeneity around heterosexuality.

There are counter-arguments, of course. Given that there are significant similarities between members of the same sex which do not exist across the sexes - e.g. men supposedly having a greater appetite for sex than women, and tending to be more interested than women in things like sport and computer games - homosexual relationships could have a natural hedonic advantage over straight relationships. Without doing empirical research I really don't know which effect would be larger; the point is that an actual defence of homophobes - as opposed to an argument that they should be tolerated - would treat them as reasonable people with reasonable grounds for their beliefs and actions. Of course, no actual homophobe holds his/her position because of the reasoning I have laid out above.

Saturday, 10 May 2014

Too much healthcare?

In one of my political philosophy tutorials last semester, I was asked to defend the libertarian position on healthcare - that it should be private with minimal if any government intervention. I immediately brought up the issue of overconsumption: that, when someone is not bearing the cost of their consumption, they will tend to consume more than the socially optimal amount. It failed to convince the people playing Rawlsian liberals and religious conservatives who were also in the debate, but it was at least a sensible argument that people of many different should take seriously. (I'd suggest it also gave the lie to the whole point of the exercise, which was to develop an Overlapping Consensus: that is, an agreement on policy between all "reasonable and rational" people. In order to attain this for healthcare, a Rawlsian would denigrate the libertarian position as "unreasonable", a word which here means "unwilling to consider the interests of others", despite the thoroughly reasonable nature of my argument).

The reason I bring this up is because I'm now wondering it the problem is actually far worse. I'll present a simplified theoretical model first, then show the things I have been reading which led to this chain of thought. I haven't done any independent empirical research and so would be wary of the conclusion, but it certainly seems plausible. The articles I have been reading all concern the US; my model would predict that the problem should be worse in the UK, so that's something that really needs to be investigated as a test of the theory.

In general in an economic model, we would assume consumers to be perfectly-informed, perfectly rational utility maximisers operating within a budget constraint. Within medicine, the budget constraint is very weak due to subsidies, nationalisation, insurance etc, and so we will treat it as non-existent. There is also very limited consumer information; in particular, I suspect that consumers will systematically overestimate the benefits of more healthcare. Furthermore, I model the actual utility of increasing healthcare increasing at first, but then actually falling. Essentially, I am rejecting the axiom of non-satiation: the marginal private benefit of consumption would look something like this:

Given the absence of a budget constraint, we would expect the consumer to consume up until the point where Marginal Private Benefit (better health and longevity) equals Marginal Private Cost (inconvenience of treatment, pain). However, this is where we introduce the problem of lack of information, by which I suggest consumers systematically overestimate the benefits of more healthcare.
We clearly see overconsumption here, even compared to what the individual would ideally choose in a post-scarcity economy. I'll add in another cost line to show the cost to society as a whole, given that individual is not bearing the financial cost of healthcare:
And to be fair, I should probably include a benefits line which takes account of positive externalities from healthcare - to concerned friends and family who have less to worry about, bosses who have less time off work to account for. These are likely to taper off with the higher levels of treatment, of course, since the increased treatment may restrict mobility and/or activity on the part of the patient.
The intersection labelled A is where, in an ideal world, consumption would occur. The intersection labelled B is where, in my model of the actual world, it will happen. Between these two points, the person subsidising the medical bill is largely paying for suffering to be inflicted upon the patient.

If I were aiming to provide an even better model of medical consumption, I would be likely to include a role for hyperbolic discounting. The effect of this would be to cause the consumer to underestimate the marginal private benefit of healthcare when young. However, my model is intended primarily as a model of healthcare among older people, where the effects are more immediate and so hyperbolic discounting plays less of a role.

In any case, is this such a problem? Some money is wasted on too much healthcare, antibiotics become that bit less effective - a pity, but surely it's far better to err on this side than to err on the side of underprovision?

This brings me to the reason I was thinking about the economics of healthcare. After reading Scott Alexander's half-hilarious-half-depressing parody of Piano Man by Billy Joel a couple of times, I went back to looking over some of his older posts on working in medicine. The first post I read at Slate Star Codex (as an actual follower of the blog, rather than merely as someone who followed a link from Marginal Revolution*) was Who by Slow Decay, which discusses the horrors which are to be found in hospitals. It links to an article entitled How Doctors Die, which talks about how most doctors would not go through the treatment to which their patients are subjected. From reading those, I would suspect that a substantial portion of healthcare spending on terminal patients is not even mere waste - rather, it is destructive. It puts people through painful, disgusting, repugnant deaths in an uncontrolled manner. Euthanasia would be a mercy, but even better would be to just stop f***ing CAUSING these horrible, miserable endings: let people die.

So how much of medical spending, then is not merely wasteful but actively damaging? Somewhere between a fifth and a quarter of lifetime medical expenses occur in the last year of life; of course, not all of this is on terminal patients and not all that is spent on terminal patients is destructive; that said, you don't need to be Robin Hanson to think a substantial portion of healthcare spending is counterproductive.

*Speaking of Robin Hanson, the MR link was to this article describing his talk on effective altruism. Reading over this again, I notice the following claim attributed to Hanson:
the real rate of return on investment has been higher than the growth rate for 3000 years and this pattern shows no signs of changing.
Robin argues that for this reason, it is better to save money and donate it all at the end of your life than to donate it immediately. There are rebuttals of this conclusion (e.g. opportunities for efficient charity dry up over time, we can do more good now than we will be able to do in future) but the quoted claim goes unchallenged. Compare to the current debate over Thomas Piketty's  Capital in the 21st Century, where right-wing and libertarian critics of the book have been eager to dispute the claim. There are many trained economists within the effective altruism movement, so surely the statement could have been challenged when Hanson made it? This leads me to suspect that challenges to the idea that r > g are frequently more about ideology than economics.

Monday, 5 May 2014


I suffer from extreme procrastination. It's a serious problem, I'll realise at half-past nine in the evening that I need to go for a run and then shower, all before I get round to cooking and eating tea. When I finally get around to cooking, I'll be working in darkness because I didn't change the lightbulbs in the kitchen when one or three of them had blown. (To be fair my housemates should take some of the rap for that, but then again they're both so tiny as to be unable to touch the ceiling, even when standing on a chair, so I'd be needed anyway). I'll put this off further by writing a blog post about my procrastination, and get distracted half-way through writing the post by listening to We Know Better. Intersperse a few games of online chess, conversations with at least one flatmate, and probably half a dozen other distractions. Washing up will be put off to the next day due to the aforementioned lighting problems.

PS. The internet needs a Beginner's Guide to Buying Lightbulbs. Like, really basic. Different shapes, sizes, wattages, etc and other basic considerations and how to work them out from the package-less lightbulb sitting next to you on the settee. Also, which shops to buy them from, how much you should expect to pay, and other considerations which may seem obvious once you know how to buy them.