Misc: May 2011 Archives


May 28, 2011

I've been listening to Gregory Clark's World Economic History -- Pre-Hsitory to the Industrial Revolution class on iTunes U. The class, taught out of Clark's popular A Farewell To Alms, is devoted to Clark's thesis that prior to the Industrial Revolution all humans mostly lived on the Malthusian frontier [*]. That is to say that the population level is maintained in equilibrium between population and resource levels/income. Anything that adjusts these factors temporarily removes the system from equilibrium, but homeostasis quickly reasserts itself. So, for instance, if there is some new technology that increases crop yields, people respond by having more children or by living longer (whether consciously or because being better fed makes you more fertile and live longer) and thus population increases with little or no likely impact on overall standard of living. Thus (Clark argues) there was nearly no net improvement in standard of living from the Neolithic to the Industrial Revolution.

As I understand it, Clark argues that two factors acted to change this state of affairs. First, technological change accelerated to the point where the the amount of resources that could be exploited was changing faster than the time scale on which birth and death rates responded, keeping the system permanently out of equilibrium. Second, people started practicing real fertility control, which acts to damp the population response to higher income levels (and of course there are natural limits on how much lifespan can increase solely on the basis of income.) [There's also a whole bunch of stuff about how these changes are due to cultural and biological evolution as a result of differential reproduction between the rich and the poor, but I don't want to talk about that just yet.]

Much of the first half of the course is devoted to explicating the model, and in classic counterintuitive economist style, Clark makes a big point about how in the Malthusian world, things that you ordinarily think of as good are bad, and vice versa. To take on example, if a horrible disease gets introduced into your society so that it increases the death rate by 10%, that leaves more resources for everyone else with the result that the people who don't die of plague have a higher overall standard of living. Here's Clark's Table 2.2 (page 37 of "A Farewall to Alms"):

Malthusian "Virtues" and "Vices"
Fertility limitationFecundity
Bad sanitationCleanliness
Harvest failuresPublic granaries
InfanticideParental solicitude
Income inequalityIncome equality
IndolenceHard work

Note the scare quotes around virtues and vices. Clark sort of equivocates between the view that societies with short life spans and high material standards of living are really better and the view that income levels are just one instrument. For instance, on page 36 he writes:

In summary table 2.2 shows Malthusian "virtues" and "vices." But virtue and vice here are measured with reference only to whether actions raised or lowered material income per person.

This sort of supports the "it's just an instrument" view but then on page 38 Clark writes:

The failure of settled agriculture to improve living conditions, and the possibility that living conditions fell with the arrival of agricultire, have led some economists, anthropologists, and archaeologists to puzzle over why mankind abandoned the superior hunter-gatherer lifestyle for inferior agrarian societies.

This argument isn't a new one. In fact, this specific form of the argument was famously made by Diamond's The Worst Mistake in the History of the Human Race. Clark comes back to the more general theme repeatedly, and suggests in a number of places during the class that people in fact would prefer to live in societies with a lot of disease and violence but correspondingly higher material living standards (though again, there is some ambiguity about the level to which he is actually endorsing this view.)

It seems to me that one ought to raise at least two objections to this general line of argument. First, it's not at all obvious that it's that useful to assess people's welfare by their income level (or as I once heard it put more crudely, by the number of calories they are able to consume per day.) [I wish I could remember where I read this.] First, to a great degree human sense of how happy they are is positional, so if everyone in society gets 100% richer, that doesn't make everyone 100% happier, it's just that instead of being jealous of the guy next to me at the stop light in his Audi RS4, I'm now jealous that he has a 911 Turbo; not much of a net win. Second, in the Malthusian model a lot of the "virtues" that result in a higher standard of living are things people find really unpleasant. In particular, random violence, crop failures, and disease (and uncertainty in general) is incredibly stressful (see Sapolsky's Why Zebras Don't Get Ulcers for a good primer on this.) It's not clear at all that people in general would rather trade much higher rates of catastrophic events for a somewhat higher level of expected welfare if you survive. Clark does offer some arguments that people make choices along these lines, for instance that people voluntarily joined the East India Company even though the risks were very high, but it's not clear that it's really that useful to use the risk/reward behaviors of 20-year old male adventure seekers as a stand-in for the entire society.

Even if we are to concede that people are individually happier in societies with high violence and disease rates (and hence lower populations) but high standards of living than they would be in societies with lower violence and disease rates (and hence higher populations) but correspondingly lower standards of living, that does not mean that those societies are truly more desirable. (See, for instance Parfit's "Only France Survives" in Reasons and Persons). Surely, most people in societies with a low standard of living would nevertheless prefer to be alive than dead, even if dead would mean that other people would live better, so it's difficult to say that the society with the lower population is better, especially when that equilibrium is obtained by high death rates (i.e., the killing of people who exist and have their own interests) rather than low birth rates (i.e., the nonexistence of people who might otherwise exist.)


May 9, 2011

MSNBC (from Reuters) reports that Sen. Charles Schumer wants Amtrak to create a "no-ride" list for trains (þ Volokh):
Schumer, citing U.S. intelligence analysts, said attacks were also considered on Christmas and New Year's Day and following the president's State of the Union address.

He called on the U.S. Department of Homeland Security to expand the Secure Flight monitoring program, which cross-checks air travelers with the terror watch list in an attempt to prevent anyone on the "no-fly list" from boarding, for use on Amtrak.

Such a procedure would create an Amtrak "no-ride list" to keep suspected terrorists off the U.S. rail system, he said.

This is one of those situations where reasoning by analogy can lead you seriously astray. Airplanes are an unusual case not so much because they are uniquely vulnerable, but because they are uniquely secure. It's true that planes are relatively fragile in that a small amount of explosive can kill a lot of people, and that even small accidends tend to kill everyone. [This is only a partially unique property, though, so a full account of why planes are such an attractive target surely needs to involve some social and psychological factors.] However, they are also ordinarily well protected, so it's hard to get access to them to do damage. At least in theory planes are kept in secure conditions on the ground so it's hard to place a bomb, and obviously once they're in the air it takes something like a surface-to-air missile (or least good luck with a gun) to cause catastrophic damage. This means that if you want to attack a plane, it's very convenient to actually be on it, so it's at least arguably useful to keep suspected terrorists off the plane.

But this doesn't apply to trains, which are (a) not particularly well secured and (b) easily accessible when they are in transit, which means you need to secure hundreds of miles of track. This means that if you want to attack a train, you don't need to be on it, you just need to get access to the track and damage it at the right time. (See here for a list of train accidents). It's like these guys have never seen Bridge on the River Kwai.

Even if that weren't true, it's important to remember that while planes are already a limited-access type thing, trains often are not. Amtrak may do some kind of passenger identification there are lots of commuter trains (e.g., Caltrain) where that not only aren't passengers identified, you can get on the train without a ticket. Instead, the conductors just come by periodically and audit. Converting to a system where you actually checked ID for each passenger before they got on (and remember that something like 50-100 people might get on in 2 minutes on an open platform) seems like it would be prohibitively expensive. These trains don't carry as many people as some Amtrak trains, but they certainly have enough passengers that if you could kill a significant fraction of them it would be bad. And this doesn't even get into the question of subways. In general, train security is set at a level designed to deter fare evasion, not to protect the train itself.

Even if you think that airplane security is set at an appropriate level (which IMHO it probably isn't) this seems like a security measure which comes at a huge amount of cost and very little benefit.


May 7, 2011

Like anyone else who listens to public radio, I've often felt that donating money would be a lot more attractive if you could just make the pledge breaks stop. Of course, since radio is a broadcast medium and you're only donating a small fraction of what they want to raise, that's not really possible. Technically speaking, I suppose they could set up some alternate, encrypted broadcast, but that sounds like more trouble than it's worth since conventional radios can't do any of that stuff and while satellite radios can do encryption not that many people have them and of course those who do aren't exactly the typical public radio demographic. It's of course obvious that Internet streaming could be used to provide this service, but that's not really that great a substitute either, especially for those of us who tend to listen to radios in our cars. However, KQED, at least, seems to have decided it's worth a try; this time around they are offering a pledge free streaming option, at least sort of.

I say "sort of" because it's not like anyone who donates gets access. Rather they're positioning it as one of their "gifts":

Public comments about our on-air fundraising drives have not been ignored! KQED has listened and is proud to have developed technology to respond with an alternative. The new Pledge-Free Stream is the first attempt by any public radio station to offer listeners the satisfaction of giving without pledge break interruptions. We believe it is critical for our organization to recognize how we can best serve you -- our members and listeners. Through this launch of the Pledge-Free Stream, we will be evaluating listener interest and feedback to inform us on the viability of this product in the future.

A few observations about this plan. First, it seems to reflect a rather different view of the role of the pledge breaks in KQED's programming than I would have expected—and certainly that I have. As suggested above, I always figured that the point of the pledge break was to hassle you to donate some money and that once that purpose had been fulfilled they would naturally stop with the hassling, except that with radio it's kind of an all-or-nothing proposition. That always seemed implicit to me in the exhortations from the announcers that once they hit their fund-raising goal they would go back to regular programming. However, that's clearly not how KQED sees things, since they're actually requiring you to separately pay for pledge-break freeness:

Because the Pledge-Free Stream is a separate gift item, you must select it when making your donation. For example, if you'd like to donate $75 and receive the KQED Wave T-shirt, you would still need to select the Pledge-Free Stream and give an additional $45, for a total of $120.

And since they do pledge 3x a year and this "gift" only applies for this pledge drive, you're looking at paying $135/year not to listen to fund raising.

So, why isn't KQED just providing this service to anyone who pledges?

At this time, we are not offering the Pledge-Free Stream as a free gift. Current members and recent donors will also need to give an additional donation of $45 to receive this service as a separate thank-you gift. Since this is the first time KQED or any public radio station has offered a pledge-free stream, it is important for us to accurately measure public interest. By your response and the number of people who donate for this gift, we will be able to evaluate how to offer the Pledge-Free Stream in the future. Your feedback will help us improve our service to you, as well as understand the value of this gift. There is also a substantial cost that KQED must cover to produce this secondary stream, from equipment to doubling the number of announcers. The funds raised through this service will help offset those costs.

As far as I can tell, the cost rationale is basically bogus. There's a fixed cost to producing the extra content, but as soon as they offer it to anyone then they've already incurred that cost. [Also, how high can it really be? They programming being preempted is largely national programming they pay NPR or PRI for plus the interstitial announcements they need to run.] As for the costs of hosting this, if we assume that they're running a 32kbps stream and every listener uses the system 24x7 they would be able to host the service on Cloudfront for something like $.05/listener-day or about $1.00 for the entire pledge period. Really, it's far less since nobody listens 24x7

The opening part of this paragraph seems to me to suggest a more likely likely rationale, namely that they're trialing a subscription service and want to see what the market will bear. There's actually one more factor to consider that they don't raise at all: I've been talking as if forcing people who have already donated to listen to pledge breaks was all loss, but arguably it's not. If nothing else, it forces those people to listen to a lot of free advertising for the station's sponsors (and especially to listen to the announcers pimping the various "gifts"). Allowing people to opt out to some extent diminishes the value those sponsors are receiving for funding the pledge drive, and perhaps diminishes their willingness to donate.