This week's NEJM contains a pair of articles confirming the relatively obvious fact that reducing medical intern's work schedules significantly decreased the number of errors that they made. Of particular interest is the controlled, prospective, crossover study by Landrigan et al, where the researchers explicitly measured the number of medical errors made by interns on a traditional schedule and an "intervention schedule and found that the traditional schedule produced:
So, clearly intern performance is a lot better under this schedule. Interestingly, the total number of errors doesn't seem to have differed between the two schedules. It may be the case that this is a result of other team members making more errors or it may just be an artifact--the study wasn't designed to determing the effect on aggregate error rates.
One of the common criticisms of reducing intern hours is that it reduces communication and that therefore other people will need to take up the slack. In order to counter the first criticism, the study designers tried to use an innovative records keeping system to ease intern-to-intern transition. It never got really used, but as intern errors still went down, it looks like that criticism wasn't as important as you might think. One possibility is that this informational context actually isn't as important as you might think.
However, the failure to find a decrease in the total number of errors is somewhat troubling--although, as previously mentioned, this isn't a slam dunk because of the study design. Clearly, we need a study to determine the overall effect on patient care of shorter work hours.
Colby Cosh nails the implications of actually having a working therapy that depends on fetal tissue:
Any of you pro-lifers out there feeling uneasy about your "It's not as though fetal tissue grafts are really medically promising" arguments yet? Just thought I'd ask--I know that, considering your crowd relies so heavily on "moral intuition", your memory doesn't seem to stretch back as far as the time when your predecessors were denouncing heart transplants. But maybe you'll get lucky, and Elisabeth Bryant will go blind again, right? (þ: Rescorla)
Despite the fact that IVF pretty much requires killing zillions of helpless zygotes, it's more or less managed to escape serious opposition for the simple reason that it works. Whatever one's moral positions, it's a pretty hard political argument to make when the other side can bring out a bunch of healthy babies to make their case.
I suspect that a lot of the traction that the anti stem cell/fetal tissue lobby has gotten is a result of the fact that we don't yet have any therapies that definitely work. Sure, that allows us to offer abstract hope to people with Alzheimer's, Parkinson's, paralysis and stubbed toes, but it doesn't provide the same emotional appeal as one solid case of someone who's been cured. If it turns out that there a large category of people who can be helped by this sort of technology, the chances of it being banned are pretty much nil. No doubt that's why there's so much enthusiasm behind a worldwide ban--if research isn't stopped everywhere someone will eventually find a treatment that works and then it's game over.
The New Scientist reports that fetal cell retinal transplants have shown some success in reversing blindness:
Three years ago Elisabeth Bryant believed she would be blind for the rest of her life. I couldnt see anything, she says. Now, although her vision is not perfect, she can see well enough to read, play computer games and check emails.
Bryant has retinitis pigmentosa, an eye disease that has blinded four generations of her family. What has saved the sight in one of her eyes is a transplant of a sheet of retinal cells. The vision in this eye has improved from 20:800 to 20:84 in the two-and-a-half years since the transplant a remarkable transformation.
There's only one small problem: the retinal cells come from aborted fetuses. This technique isn't exactly going to sit well with the anti-stem cell crowd. At least with ESCs the idea is that once you've isolated the ESC line and it works you can generate a more or less infinite supply of transplant material. In this case it looks like each aborted fetus can only be used for a small number of transplants.
In an attempt to address concerns about e-voting, the US Election Assistance Commission (EAC) has decided to store hashes of the official certified versions of voting software:
The US Election Assistance Commission (EAC) based in Washington DC, announced on Tuesday that it had persuaded the five largest electronic voting machine vendors to submit certified versions of their software to the National Software Reference Library (NSRL).
"Their acceptance of our request begins the process that assures the country that we will have a higher level of security and therefore confidence in e-voting than we have ever had before," said DeForest Soaries, EAC chairman.
At the NSRL, each program file submitted was converted via a mathematical function known as SHA-1 into a fixed-length string of digits, called a "hash". The hash is like a fingerprint for that piece of software - if the software changes, the hash changes.
The reporter for the New Scientist called Avi Rubin and David Dill who point out that this is totally bogus:
"Hashing will not catch malicious code that's already in the system, which is the biggest threat," he says. It will also fail to catch any bugs that are there by mistake, he adds.
David Dill, a computer scientist at Stanford University, California, and founder of the lobby group Verified Voting, says e-voting software undergoes state testing and federal testing by Independent Testing Authorities. But he points out that bugs have slipped through before.
As usual, when someone tells you about a security measure, you have to ask what threats it's designed to protect against. The theory seems to be that the CD you have, the one you received in a shrink-wrap box from the manufacturer, the one with the manufacturer's logo on it, is actually not an official software install. Presumably, some malicious voting machine company employee burned a fake CD or FedEx swapped them in transit.
The theory for how the NSRL protects against this threat seems to be that when voting officials get CDs in the mail they run some program1 over them and then compare the output to the reference hash. They have to do this with every single CD sent to them, of course, or the whole thing doesn't work.
There are two big problems with this idea. First, it doesn't protect you against any of the problems we know the current voting machines have, namely insecure design, buggy software, and incompetent testing. These are problems with the systems as shipped. It doesn't do you any good to know that you've got the real software because it's the real software that's the problem. The only thing that this protects you against is surreptitious software substitution between the manufacturer and the election officials, which is probably the most secure part of the distribution chain.
This bring us to the second problem: it's a lousy way to ensure the integrity of the software on voting machines. So, election officials can be sure that they have the right CDs, but what's to stop them from installing fake software on the machines? In this scheme, nothing, but it turns out that there's an easy way to solve that problem: have all the software digitally signed by the manufacturer and have the voting machine refuse to run any software that isn't signed. This would solve both the threat from election officials installing fake software and the distribution problem allegedly solved by the NSRL, without any need for central coordination, with the additional advantage that it could be automatically enforced without requiring manual verification of the install media.
There's no magic technology required to make this work. You just build a custom BIOS that verifies the signature over the software and won't boot if the signature isn't there or doesn't verify. In fact, I used to work for a company that did exactly this with their routers. The big drawback with this scheme, of course, is that it requires modifying all the machines already in the field, but then they should have been designed this way in the first place. The fact that the manufacturers didn't do so should tell us that they don't understand security--as if we needed any more evidence of that.
1 I say "some program" because I don't know if Windows even has a built-in SHA-1 program. Most of the UNIX variants do at this point.
Ed Felten points to a disturbing story about problems with New Mexico's touchscreen voting machines:
[Kim Griffith] went to Valle Del Norte Community Center in Albuquerque, planning to vote for John Kerry. "I pushed his name, but a green check mark appeared before President Bush's name," she said.
Griffith erased the vote by touching the check mark at Bush's name. That's how a voter can alter a touch-screen ballot.
She again tried to vote for Kerry, but the screen again said she had voted for Bush. The third time, the screen agreed that her vote should go to Kerry.
She faced the same problem repeatedly as she filled out the rest of the ballot. On one item, "I had to vote five or six times," she said.
Michael Cadigan, president of the Albuquerque City Council, had a similar experience when he voted at City Hall.
"I cast my vote for president. I voted for Kerry and a check mark for Bush appeared," he said.
He reported the problem immediately and was shown how to alter the ballot.
Cadigan said he doesn't think he made a mistake the first time. "I was extremely careful to accurately touch the button for my choice for president," but the check mark appeared by the wrong name, he said.
...
In Sandoval County, three Rio Rancho residents said they had a similar problem, with opposite results. They said a touch-screen machine switched their presidential votes from Bush to Kerry.
Ed and his commenters propose some candidate explanations, ranging from calibration error to resting your hand on the screen while you try to vote.
I don't know what the problem is and to tell you the truth, I don't really care. The bottom line is that voting machines should give accurate results even if voters and voting officials make mistakes. This kind of problem would show up under any reasonable kind of field testing. All you'd need to do is get a bunch of semi-trained volunteers to set up the machines and a bunch of untrained voters to try to vote specific tickets and measure the error rates. There aren't any complicated software engineering or security issues here, just simple usability testing. If the voting machine manufacturers haven't done that much, it's hard to have much confidence in any other aspect of their engineering, especially things like security, which are harder to discover with this kind of testing.
Apparently Sprint will be offering generic Internet access via Bluetooth to the Treo 650. In fact, they insist they always meant to do that.
Cats are amazingly allergenic and yet approximately a third of cat-allergic people keep a cat. This makes hypoallergenic cats an attractive genetic engineering target and sure enough, someone is trying it. Allerca is currently taking deposits for a hypoallergenic cats to be delivered in 2007:
The ALLERCA research and development teams comprise some of the leading specialists in animal genetic engineering. The team's expertise places ALLERCA in a unique position to produce the world's first hypoallergenic cat.
A glycoprotein, Fel d 1, secreted by the sebaceous glands, is the major cat allergen. This allergen is found in the fur, pelt, saliva, serum, urine, mucous, salivary glands, and hair roots of the cat.
Using patented genetic technology, the ALLERCA team will focus on the particular gene that produces the Fel d 1 glycoprotein. Using a technique known as gene silencing, the process reduces the genes ability to produce the protein.
Sounds like a great idea--unless, of course, it turns out that Fel d 1 is also needed to make the cat's brain work or something.
Colby Cosh pointed me to this excellent Something Awful article about how to fake being a music snob, complete with reader comments. Useful tip:
Suppose someone says this: "Hey, have you heard of Flop?" Obviously, we're also going to suppose that you haven't heard of Flop, because you haven't. How would you react to this? Your first instinct might be to say "Yeah, I've definitely heard of Flop." This is bad idea for many reasons: first of all, it might be a trick. There might be no such band as Flop. Worse yet, your interrogator might have all manner of follow-up questions about Flop, and they're going to be increasingly difficult to weasel your way out of. It might also be tempting to say "I've heard of them, but I haven't really heard their records." This is a wishy-washy compromise, and it's an essentially meaningless answer. Basically, the only way to win at this game is to play it like Double Dare. You supposedly know everything about music. They ask you about Flop: dare. It's time to put them on the defensive. "Hmm, Flop... what label were they on?" Double dare. Chances are, they don't know. For all your opponent knows, you might know all about Flop, but he has failed to give you enough simple information, such as the name of their record label, for you to correctly identify them. Even if your opponent does know what label the band in question was on, you still can't lose. "Ah yes," you can say "I am almost certain that I have a compilation released by that label that has a few Flop songs on it." This at least buys some time; the worst that can happen is that you'll have to go home and look up Flop on the internet (this is known as the "Physical Challenge"). Your opponent is foiled, and you live another day.
Indeed.
Let me see if I have this right: we're missing 377 tons of high explosive (HMX and RDX). To get a sense of the magnitude of this screwup, consider that:
Outstanding!
UPDATE: Here's another way to look at it: at 1 lb/IED, that's 2000 explosions a day for the next year.
The Treo 650 has been announced and the Gizmodo guys are pretty annoyed that they've made it so you can't use it as a modem:
The big story already seems to be the news that Sprint will cripple its Bluetooth implementation to prevent the use of the Sprint Treo 650 as a modem for laptops and other devices. Marc Hedlund at O'Reilly already is making noise about how he'll be leaving Sprint because of it, and I'll just chime in and say that decision will prevent me from choosing Sprint as my 650 carrier, as well, although they were the front runner in my choice of new carriers. Hopefully, AT&T Wireless/Cingular won't break the DUN profile on their phones, because otherwise I'm going to have to execute every last one of these money-grubbing telco executives.
My prehistoric Sony Vaio doesn't have Bluetooth, so I have the luxury of being objective here. I can see why the Gizmodo guys are annoyed, but I also see Sprint's point of view. Sprint's package for the Treo includes unlimited Internet and I suspect that the use profile of someone with a cell modem is pretty different from someone using the Treo alone, so it's understandable that they wouldn't want people bypassing their data service. If I had one of these puppies and Bluetooth the temptation to nail up a 24×7 Internet connection to my laptop would be pretty strong.
Nick Weaver answers my puzzlement about Ford's earnings in the comments section:
There is a simple solution to the finacing mystery...
Discounted loans (eg, 0%), cost the car side X dollars, and the loan side GETS X dollars, so it assigns the cost of the discounted loan to the car side rather than the finance side.
Thus the finance side reports the profit, while the car side reports the loss, although they are really one company.
A good point, but kind of a funny way of doing things. Giving every customer a large rebate (whether it's a financing break or a cash payment) costs you money. Why not just make the asking price lower? Interestingly, American and Japanese manufacturers are very different
Ford has also not been matching G.M. in spending on rebates and no-interest financing deals. The company spent $4,048 for each vehicle on rebates and other incentives last month, according to Edmunds.com, a company that tracks automotive data. While that is well above the $911 spent by the average Japanese competitor, it was below the $4,593 spent by G.M.
Ford and GM have been offering very susbtantial incentives for quite some time now, so they certainly have had an opportunity to adjust their prices instead. Why haven't they?
One possibility (I'm theorizing here) is that they think that consumers use price as a segregation mechanism (i.e., I want to buy a car in the $20-30k range), so if you offer big incentives your cars get into people's selection set but then look radically cheaper when the customer goes to compare them head to head. Any other theories?
Ford has just announced their Q3 earnings, and once again they show an interesting pattern:
AUTOMOTIVE SECTOR
On a pre-tax basis, excluding special items of $64 million, worldwide automotive losses in the third quarter were $609 million, a deterioration of $61 million from the same period a year ago. The special items included a $23 million charge for restructuring at Jaguar, and $41 million in North America related to the revaluation of Ford's investment in Ballard Power Systems [NASDAQ: BLDP; TSX: BLD], a leading fuel cell manufacturer....
FORD MOTOR CREDIT COMPANY
Ford Motor Credit Company reported net income of $734 million in the third quarter of 2004, up $230 million from $504 million a year earlier. On a pre-tax basis from continuing operations, Ford Motor Credit earned $1.2 billion in the third quarter, compared with $808 million in the previous year. The increase in earnings primarily reflected improved credit loss performance and leasing results.HERTZ
Hertz reported a pre-tax profit of $249 million in the third quarter compared with a $186 million profit during the same period a year ago. The improvement primarily reflected higher rental car volume in the leisure and commercial segments, and higher profit from the disposal of used vehicles and equipment. The improvements were partially offset by lower pricing in car rental markets, due to the highly competitive environment, as well as higher costs.
In other words, Ford is losing money hand over fist on cars but making a lot of money on financing and car rentals.
So, here's the question: why? I'm not asking why Ford is losing money, that's a business question. Rather, how is that they're managing to make enough money from their financing business to offset the losses? Why don't people get their financing from cheaper outlets than Ford Moter Credit Company, leaving Ford in the lurch? It's obvious why the dealer won't offer them cheaper financing--they're tied to FMCC, but surely you could go to the bank.
Listened to part of NPR's Forum show on Electronic voting in time to hear Michael Alvarez from the MIT Voting Technology Project respond to a caller's concern about vulnerabilities in voting software with:
We're at a point now where these systems have been certified, both at the Federal and State level, and they do go through a fairly thorough regimen of testing and one would hope that before they would be rolled out to voters that those sorts of bugs or deliberate malicious code would have been spotted.
But the major take-home point of Kohno et al.'s Oakland 2004 paper on Diebold's voting system is that this isn't the case. On the contrary, Diebold's system had a number of long-standing security vulnerabilities that made voting fraud quite easy. Moreover, in some cases these vulnerabilities had been known about for quite some time and Diebold had done nothing about them. Moreover, most of the problems Kohno et al. found weren't hidden in any way. They were obvious design errors right out in the open that anyone with any knowledge of computer security would recognize. Alvarez absolutely should know this, and I don't see how he can say with a straight face that the in-place testing procedures provide us with any reasonable level of assurance--unless by "one would hope" he means "this would be nice, even though we know for a fact that it's not the case."
Like most everyone else, I've been blithely throwing around the 36,000 flu deaths every year statistic. Craig Hughes has done some nice research on the topic that suggests that this number has been significantly biased by the 1918 flu pandemic. Craig's data, drawn from the American Lung Association, draws a rather different picture, with normal yearly mortality being more around 1200 than 36000. Worth checking out.
Interval workouts have four tunable parameters:
The theory here is that you want to tune these four parameters so that you're pushing yourself hard throughout the workout, but your performance isn't quite starting to fade. A common rule of thumb is that when you finish the workout you should feel like you could do one more repetition but that two would be too hard. Given that general rule, you have some flexibility with the parameter values. Typically, you choose the length of the work interval and the total workout time and adjust the intensity level and rest time accordingly.
Here are some general guidelines:
| Repetitions | 6 |
| Work Interval | 1 minute |
| Rest Interval | 2 minutes |
| Total Time | 18 minutes |
As your training progresses, you can start to dial up the total training time and work interval and dial down the rest interval. Here's a good goal workout for a serious athlete:
| Repetitions | 5 |
| Work Interval | 5 minutes |
| Rest Interval | 2 minutes |
| Total Time | 37 minutes |
This is a very hard workout, comparable to that used by serious triathletes. As with our more basic training plan, it's critical to work up slowly to give your body time to adapt. There's no hurry, and it's very easy to get seriously overtrained or injure yourself. I estimate it would take someone who had never previously done interval workouts 3-6 months to work up to this level of intensity.
Here's one of my successful workouts from 2001. The work interval is 1 mile on the track, with a 3 minute rest interval, spent walking around. I hit the effort level just right:
| Time | Heart rate |
| 5:55.6 | 166 |
| 5:43.4 | 173 |
| 5:43.5 | 176 |
| 5:43.9 | 177 |
| 5:42.3 | 184 |
I deliberately went out slow on the first rep to get warm. Even after the warmup, it helps to be careful on the first rep. The remaining four, I tried to hold a constant pace. The relatively stable times but slightly escalating heart rate are a good sign that I've judged the pace correctly. Note that I picked up the pace a little bit on the last rep and I was more tired so the heart rate went up a little bit.
One thing to be attuned to is the feeling that you're really cratering. If you see your performance really drop from rep to rep and you feel like garbage, it's probably time to bag the workout for the day and try again next week. Another indicator is if your heart rate is a lot harder than in previous similar workouts. Here's an unsuccessful workout. I was attempting the same workout as above a few weeks earlier but went out way too hard:
| Time | Heart rate | ||
| 5:47 | 176 | ||
| 5:51 | 174 | ||
| 6:08 | 173 |
| Sun | Wed | Fri |
| 60-75 minutes OD | 45-50 minutes EN | 50-60 minutes IN |
Generally, you want to put the put the longest rest after your long day, especially if you're running. This gives your legs time to recover. You can switch to this schedule more or less directly from your basic training plan, as long as you take it really easy on the Intervals the first few weeks. A slightly more conservative approach would be to add the Endurance workout first and then after a few weeks transition the third workout to Intervals.
If you get more serious and want to add another workout, you can add another Overdistance workout. You'll need to do two workouts in a row. It's best to do an Overdistance and Endurance workout back to back, like this:
| Sun | Tue | Wed | Fri |
| 60-75 minutes OD | 50 minutes OD | 45-50 minutes EN | 50-60 minutes IN |
This second training plan has the makings of a fairly serious workout schedule, suitable for local road races and the like. The next thing you would want to do in order to improve your training would be to increase the volume on your long day, but that's only worth doing if you're aiming for long distance events such as half marathons or centuries.
In the next post, we'll provide a detailed description of how to do Interval workouts.
Seventy-two years after prohibition, it's still illegal for bars in South Carolina to to sell liquor in normal sized bottles. Instead, you have to open a separate 1.7 oz bottle for each customer. No doubt this was originally intended as some sort of temperance measure but apparently now it's become some sort of price support program for the liquor distributors.
This year, however, SC voters get to vote on a constitutional amendment to allow the legislature to switch to "free-pour" (the system that everyone else in the country uses). It's particularly amusing to see the liquor distributors who oppose the measure explain that they're looking out for the consumer rather than just engaging in naked rent-seeking:
But opponents argue switching to free-pour drinks will allow unscrupulous bartenders to water down drinks, reduce tax revenues and could open the state's liquor laws to yearly changes at the whim of lawmakers.
"Just because it's different doesn't mean it's wrong," said Suzie Riga, vice president of Green's Liquors, a Columbia wholesaler that sells minibottles to restaurants and bars.
...
But Riga said bartenders and liquor manufactures have become creative. Long Island ice tea now comes in a minibottle, and few people can tell the difference in a margarita made with triple sec syrup rather than liquor.
Riga compares a minibottle to a bottle of aspirin. "You wouldn't use it if it wasn't sealed, would you?"
Minibottles are the same way. "When you see that bottle opened, it's clean, it's safe and you know exactly what you are getting," she said.
Wow! I never realized that ordering a drink was so dangerous. All those times I let the bartender draw me a draft beer I was taking my life in my hands.
A gallery of quick examples I examined to locate the offending tag (total time to find and extract them - circa 1 hour): http://lcamtuf.coredump.cx/mangleme/gallery/ * mozilla_die1.html Appears to be a memory corruption / overflow problem; TEXTAREA, INPUT, FRAMESET and IMG tags followed by NUL, then a number of extra characters, cause the browser to die while accessing NULL pointer, loop forever, or die while accessing invalid pointer. My guess would be that they calculate tag length using strlen-alike function, but then copy till separator or > - but this is just a guess. The behavior is tag and OS-specific, and is likely exploitable with some luck in those of the cases that do not lead to NULL ptr. I didn't investigate - Mozilla sources are 220 MB, mostly C++, takes forever to compile. * mozilla_die2.html Bogus pointer access triggered by a unusual combination of visual elements. * opera_die1.html Excessive COL SPAN in TBODY causes Opera to go down in flames, attempting to make a reference to uninitialized memory. Probably can be exploited in right conditions. * links_die1.html Table of an excessive size causes links to DoS the machine by consuming all memory until calloc fails, then write over what it managed to allocate. * lynx_die1.html Lynx loops forever trying to render broken HTML. Rest assured, this is merely a top of an iceberg; there are more crashes and other problems than one can asses and evaluate while retaining sanity.Outstanding!
The part of Granny D's response that really bugs me is the following:
That abstraction is most likely to happen where the people do not have the means to represent their own interests. So, as a democracy becomes less so-- say, as a result of the imposition of special interest campaign donors, a corporate-dominated news media, and so forth-- the jobs and needs of the people are more likely to be considered as abstractions and as replaceable moving parts, rather than as real lives.
Special interests? Special interests?!?!?! The whole point of protectionism is to favor the welfare of a special interest--a few American manufacturing workers--over that of the rest of the world.
Here's Granny D's response to the criticisms of her protectionist post:
Granny D here. I love this blog world--you make a general statement and then some people write a book for you about it. Thank you all for your comments on protectionism. I am totally persuaded and will now stop pruning my garden, leaving behind my old fashioned notion that editing and flowering are necessary partners.
What does continue to bother me, however, is the unsaid notion that labor is one of the several components of manufacturing, when, in fact, it is us. Economists (and their hunchbacked evil blogger assistants) tend to make such deadly abstractions that they lose sight of this, as if the Economy were a demigod or at least a being unto itself, whose health we must serve by sacrificing our own.
That abstraction is most likely to happen where the people do not have the means to represent their own interests. So, as a democracy becomes less so-- say, as a result of the imposition of special interest campaign donors, a corporate-dominated news media, and so forth-- the jobs and needs of the people are more likely to be considered as abstractions and as replaceable moving parts, rather than as real lives. The conversation in many of the blogger replies reflects this abstraction in favor of the demigod Market or the demigod Economy.
The rationalization that the Good will Eventually be served in a widespread manner is rather a crock, as Eventually, like the Leisure Society, never really comes. Let's have a democracy that really expresses our need for good jobs and health care and all the rest, and see if, in fact, we don't craft a better society than the horror show now being crafted by the abstracted free market. Wouldn't real democracy be a better kind of free market--with We the People really free? It is a given that nations with the most poverty have the least democracy, as people do not chose to be poor and exploited if they have a say in it. Right now, we do not have a say in outsourcing, for example, not because a natural market force is working in the world, but because our interests are not being represented, and corporate interests are. It is fairly simple, despite all the charts and books, and as our wise Yogi said, you can see a lot just by looking.
Or, as once was said to me by a Canadian Communist (really) "I don't believe in economics. I believe in people".
Unfortunately for my interlocutor, and for Granny D, economics is about people. It's about looking at their preferences and behaviors in a rigorous manner. And in this particular case, what you can see "just by looking" turns out to be seriously misleading. When jobs move offshore three constituencies are affected:
Protectionism helps the first group at the expense of the second two. Now, it may be the case that one prefers the wellbeing of that group of people--especially if you're one of them--but framing the issue as if it's "the economy" vs. "people" is just plain wrong. Anybody who talks that way has forfeited their right to be taken seriously on economic issues.
If you're interested, you can check out my slides from DIMACS here.
Thanks to Nagendra Modadugu for converting the PPT to PDF for me.
The observation that trade is a form of technology isn't original to me. I got it from Steven Landsburg, who got it from David Friedman. Here's Landsburg's version of the argument, from The Armchair Economist:
A thing of beauty is a joy forever, and nothing is more beautiful than a succinct and flawless argument. A few lines of reasoning can change the way we see the world.
I found one of the most beautiful arguments I know while I was browsing through a textbook written by my friend David Friedman. While the argument may not be original, David's version is so clear, so consice, and so incontrovertible, and so delightfully surprising that I have been unable to resist sharing it with students, relatives, and cocktail party acquaintances at every opportunity. The argument concerns international trade, but its appeal is less in its subject matter than in its irresistable force.
David's observation is that there are two technolgoies for producing automobiles in America. One is to manufacture them in Detroit and the other is to grow them in Iowa. Everybody knows about the first technology; let me tell you about the second. First you plant seeds, which are the raw materials from which automobiles are constructed. You wait a few months until wheat appears. Then you harvest the wheat, load it onto ships and sail the ships eastward into the Pacific Ocean. After a few months, the ships reappear with Toyotas on them.
International trade is nothing but a form of technology. The fact that there is a place called Japan, with people and factories, is quite irrelevant to Americans' well-being. To analyze trade policies, we might as well assume that Japan is a giant machine with mysterious inner workings that convert wheat into cars.
Any policy designed to favor the first American technology over the second is a policy designed to favor American auto producers in Detroit over American auto producers in Iowa. A tax or a ban on "imported" automobiles is a tax or a ban on Iowa-grown automobiles. If you protect Detroit carmakers from competition, then you must damage Iowa farmets because Iowa farmers are the competition.
The task of producing a given fleet of cars can be allocated between Detroit and Iwa in a variety of ways. A competitive price system selects that allocation that minimizes the total production cost. It would be unnecessarily expensive to manufacture all cars in Detroit, unnecessarily expensive to grow all cars in Iowa, and unnecessarily expensive to use the two production processes in anything other than the natural ratio that emerges as a result of competition.
That means that protection for Detroit does more than just transfer income from farmers to autoworkers. It also raises the total cost of providing Americans with a given number of automobiles. The efficiency loss comes with no offsetting gain; it impoverishes the nation as a whole.
There is much talk about improving the efficiency of American car manufacturing. When you hve two ways to make a car, the road to efficiency is to use both in optimal proportions. The last thing you should wnat to do is artificially hobble one of your production technologies. It is sheer superstition to think that an Iowa-grown Camry is any less "American" than a Detroit-built Taurus. Policies rooted in superstition do not frequently bear efficient fruit.
Together with Ingram's parable about Mister X, this is the clearest explanation of why Free Trade is good that I know of.
Larry Lessig has temporarily given control of his blog to Granny D. So, I tune in to find the following:
I know, why am I worried about protection at my age? I really don't want to be labeled a protectionist, but I think there is a happy medium between raw free marketeering and highwall protectionism. My father worked as a laborer in a furniture warehouse in Laconia, NH. He was able to own a house and raise five kids pretty decently. You can't do that anymore, and the reason is that the economy is no longer self-contained in the way that a good system or a good machine can be. Without some containment, it's rather like trying to farm without scarecrows, on the theory that the hungry birds are part of the free market of the farm, or letting the irrigation go wherever it likes, without channels to keep it from seeking the lowest point of the field. Healthy systems have their boundaries.
Can't. Help. Myself.
What an outstanding example of unclear on the concept. It used to be that you could make pretty OK money operating a Linotype machine. You can't any more because desktop publishing has completely obliterated the market and the job can be done almost completely mechanically. And when it is done manually, it's done with a computer, not by spitting out hot lead. And it looks a lot better, too, by the way. Compare a modern typeset document to something made before 1950 to see what I mean.
"But wait" I hear you say. "That's not the same. The furniture formerly made by Granny D's father isn't being made by machine. It's being made by people in China." But a moment's reflection should convince you that the difference is immaterial: today's American furniture makers are just as surely out of a job.1 The point, of course, is that the openness or closedness of the economy has absolutely nothing to do with it. The only issue is whether we as a country should decide to manufacture products (or provide services or whatever) in a cheaper way, even if that turns out to make certain ways of earning a living more difficult or impossible. Trade is just another technology for making stuff cheaply.2.
The comparison to hungry birds is particularly offensive because the people who are now making furniture in China or Pakistan aren't stealing from us--they're trying to make a living like anyone else and in most cases are in far worse financial circumstances than all but the very poorest Americans. Maybe it's bad to improve their lives at the expense of the Americans who get priced out of the market (not that I think so) but their wellbeing isn't immaterial either.
Incidentally, I wonder what self-contained "good machine" Granny D. is thinking of. Most of the machines I'm familiar with require a regular input of power, lubricants, and other supplies.
1 Actually, even this isn't clear. The IKEA furniture in my house looks to have been made by a highly automated manufacturing process.
2 A point made ably by Ingram's classic parable.
UPDATE: minor writing cleanups...
Interesting article in Science about rotavirus vaccines. Rotavirus-caused diarrhea is a major killer in the third world, causing roughly a half a million deaths a year. In 1998, Wyeth introduced the first rotavirus vaccine Rotashield. However, in summer of 1999, it was discovered that Rotashield increased the rate of a potentially fatal condition called bowel intussusception and Rotashield was withdrawn from the US market.
Rotavirus deaths are relatively uncommon in the US (about 1 death per 100,000 children) and so the intussusception risk (about one case per 2000-40,000, noone is really sure) is a bad tradeoff. However, in the third world, where the risk is more like 1 death in 200, the cost/benefit ratio looks a lot better. Unfortunately, after the vaccine was withdrawn in the US, the third world wasn't interested either:
The decision sparked an outcry among international health experts, who felt deprived of a potent weapon. Albert Z. Kapikian, one of the developers of RotaShield at the U.S. National Institute of Allergy and Infectious Diseases (NIAID), argued for a permissive recommendation that would enable U.S. physicians to use the vaccine at their discretion. That would have sent a powerful message to developing nations, he says, and perhaps spurred its adoption there. "But it fell on deaf ears," says Kapikian. When the World Health Organization (WHO) held a pivotal meeting in 2000 to assess whether and how developing countries might introduce RotaShield, health ministers gave it thumbs-down. "They said they didn't want their population to be seen as second-class citizens. If it was not good enough for U.S. kids, it was not good enough for their infants either," recalls Kapikian.
An understandable sentiment, I guess, but not really that rational. Whether a medicine has a good cost/benefit ratio in your country doesn't really depend on whether it has a good cost/benefit ratio in the US.
Rolling Stone reports that Wal-Mart is exercising it's famous buying power to drive down the price of CDs (+ /.):
Less-expensive CDs are something consumers have been demanding for years. But here's the hitch: Wal-Mart is tired of losing money on cheap CDs. It wants to keep selling them for less than $10 -- $9.72, to be exact -- but it wants the record industry to lower the prices at which it purchases them. Last winter, Wal-Mart asked the industry to supply it with choice albums -- from new releases from alternative rockers the Killers to perennial classics such as Beatles 1 -- at favorable prices. According to music-industry sources, Wal-Mart executives hinted that they could reduce Wal-Mart's CD stock and replace it with more lucrative DVDs and video games.
"This wasn't framed as a gentle negotiation," says one label rep. "It's a line in the sand -- you don't do this, then the threat is this." (Wal-Mart denies these claims.) As a result, all of the major labels agreed to supply some popular albums to Wal-Mart's $9.72 program. "We're in such a competitive world, and you can't reach consumers if you're not in Wal-Mart," admits another label executive.
Tensions are not as high now as they were last winter, but making sure Wal-Mart is happy remains one of the music industry's major priorities. That's because if Wal-Mart cut back on music, industry sales would suffer severely -- though Wal-Mart's shareholders would barely bat an eye. While Wal-Mart represents nearly twenty percent of major-label music sales, music represents only about two percent of Wal-Mart's total sales. "If they got out of selling music, it would mean nothing to them," says another label executive. "This keeps me awake at night."
Of course, the down side is that Wal-Mart also exerts considerable influence over the content of albums, often demanding "clean" versions of albums that otherwise wouldn't be family friendly. So, the good news is that Ice Cube's next album will be cheap. The bad news is that every other word will have to be deleted.
A standard complaint about documents published on the Web is that they're not permanent. Just because a URL is valid now doesn't mean that it will be valid next year. The document might have changed or been taken down, or I might have simply changed companies and moved my documents from Server A to Server B. Worse yet, I might have lost interest in the whole project and just stopped serving the pages.
DOIs provide a partial solution to this problem. A DOI is a short permanent identifier that can be dereferenced to point to the actual physical location (i.e., a URL) of the document. That neatly solves the problem of what happens when I move my pages from Server A to Server B, but doesn't help people at all if I stop serving them entirely.
I envision an archival web service system that would solve this problem. The way that this would work would be that you deliver the data to the server and pay a single one-time fee and the server operator agrees to publish the data in perpetuity at some permanent location identified by URI or whatever. This would let other people reference that document with confidence that future readers would be able to access it. Obviously, there are some practical problems: what happens if the document is really popular and requires a lot of bandwidth? What about when the Web becomes obsolete? But these seem like they are solvable with appropriate contract terms.
Does a service like this seem useful? Does it already exist?
Hovav Shacham just pointed me to a useful tool, SeatGuru.com. SeatGuru has complete seating maps for pretty much every airplane flown by US Carriers, complete with notations for good and bad seats. Pretty convenient for seat selection...
Larry Lessig provides a nice summary of HR 4077, which is part of Orrin Hatch's grand war on copyright infringement. However, I notice that he missed one important clause:
SEC. 201. DESIGNATION OF NATIONAL TREE.(a) DESIGNATION- Chapter 3 of title 36, United States Code, is amended by adding at the end the following:`Sec. 305. National tree`The tree genus Quercus, commonly known as the oak tree, is the national tree.'.(b) CONFORMING AMENDMENTS- Such title is amended--
(1) in the table of contents for part A of subtitle I, by striking `, and March' and inserting `March, and Tree';(2) in the chapter heading for chapter 3, by striking `, AND MARCH' and inserting `MARCH, AND TREE'; and
(3) in the table of sections for chapter 3, by adding at the end the following`
'305. National tree.'.
That tells you pretty much all you need to know about how laws are made in this country.
When the Internet was young, one of the big concerns was what was called the "Last Mile". The Internet core was plenty fast but getting the data to the customer's house was a nightmare. The problem, you see, was that the copper that went into your house was considered unsuitable for high-speed data delivery. At enormous expense, business could order dedicated lines (initially 56k, which now appears laughably slow, and then T-1s, T-3s, etc.) but the poor schmucks at home were stuck using dialup modems, which weren't much of an improvement over two tin cans and a string.1
Everyone knew something had to be done many solutions were proposed to this Last Mile Problem. At the end of the day, we ended up with a mixed bag of imperfect solutions that let you use existing infrastructure with minimal upgrading:
You'll notice that the one thing that these solutions all have in common is that they don't require running new wire (at least most of the time), or worse yet, fiber, which everyone agreed was going to be too colossally expensive to sign up for. Apparently, however, it just took some incentive, since that's exactly what the telcos are now doing. Indeed, according to this NYT article they're absorbing about $1000/home worth of costs to do it. Anyone know if we can get that deal here in Palo Alto?
1 I remember how excited we were when we got our first prototype 28.8k modems. There wasn't a standard yet and they couldn't talk to anything else but at least you could nail up a "fast" connection between point A and point B.
So, you've had a concussion and unsurprisingly, your brain is swelling. That's bad, right? So, a common medical response is to treat you with steroids to bring the swelling down. It turns out this isn't a good approach (+ Chris Rangel):
The study involved 10,000 patients from 239 hospitals in 49 countries and was co-ordinated by scientists from the London School of Hygiene and Tropical Medicine and the Universities of Manchester, Edinburgh, Birmingham, and Oxford.
It found that within two weeks of treatment in hospital, 21 per cent of patients given corticosteroids were dead - compared with 18 per cent of patients given a placebo, the researchers report in the Lancet.
The findings astonished doctors, many of whom assumed that the drugs made no difference to a patient's chance of survival.
The trial was abandoned half way through when the dangers of corticosteroids emerged.
Prof Roberts, of the London School of Hygiene and Tropical Medicine, said the difference was an extra three deaths in 100 patients.
"There is no evidence of benefit. At best it is useless, and at worst it is harmful," he said. "In theory the drugs should have worked. But theory as a basis for treating patients is dangerous and we really need clinical trial evidence."
Score one for Evidence-Based Medicine.
As you may have heard, the governor has just signed a bill banning hte sale of foie gras made from force-fed birds starting in 2012.
Does this mean the end of foie gras? Not according to a statement released by the governor's office, which clarifies that the bill bans a production practice, not the product itself. The law "provides seven-and-a-half years for agricultural husbandry practices to evolve and perfect a humane way for a duck to consume grain to increase the size of its liver through natural processes," the statement said.
Guillermo Gonzalez, owner of Sonoma Foie Gras, the state's only foie gras producer, said he is happy the bill passed with these conditions. "We asked the governor to sign the bill so that we have seven years to demonstrate that foie gras production is safe and proper. We will go on with our business, with the continued support of scientists and the agriculture community," Gonzalez said in a statement.
Here at EG, we are pleased to report that we have a preliminary solution to the knotty problem of getting the geese to gorge themselves without force-feeding them. True, there are a few minor problems: the geese can't light the hash pipe on their own, we need to make it offshore to avoid the DEA, and the foie gras tends to taste a lot like Doritos... However, we're confident we can have a working solution by 2012.
I'm stating to enjoy Slate's Surfergirl column. Check out this line from Wednesday:
An amateur diagnosis: Kerry is more fun on the therapist's couch than you might expect, but he should consider attending a workshop on overcoming self-sabotaging behaviors. Asked if one of his daughters was more like him than the other, he responded, and I swear to God I'm transcribing word-for-word: "Yes. No. Well that's ... gosh, I'd like to say yes, but I guess ... yes, the answer is yes." Which daughter, Mrs. Dr. Phil inquired? "Well, that's why I hesitated. Because in some ways my daughter Alexandra is more like me, but in other ways my daughter Vanessa is more like me." Senator? It's not an appropriations bill with riders. Just pick a name! Memo to Joe Lockhart: For the remaining three weeks of the campaign, do not let your candidate appear to vacillate on any subject, no matter how trivial. When he pulls into the fast-food drive-through on a campaign stop, have him bellow "COMBO CLASSIC, HOLD THE CHEESE!" before the intercom even comes on, and keep on repeating it at top volume until the last french fry has vanished down his gullet.
Sounds like good advice to me.
Well, that didn't take long. The Times is reporting on the claim that prices for flu shots are starting to go up:
Asked at a news conference about charges of $65 for a flu shot that usually costs much less, Dr. Bruce Gellin, the official in charge of the government's vaccine program, said, "We, too, have heard these stories about the price gouging."
While no specific examples of gouging were cited at the news conference, Dr. David Lang, a pediatric infectious disease specialist at Children's Hospital of Orange County in California, said in a telephone interview that a vendor had called a pharmacist at his hospital offering to sell the flu vaccine made by Aventis for 10 times the usual amount.
The pharmacist, who asked not to be identified, said in a separate interview that the vendor had asked $700 for a vial of 10 doses that usually costs $67.
...
Giving flu shots is an overwhelmingly private enterprise, "and we know that there are some people who may be in it just for the money," said Dr. Dixie Snyder, an official at the Centers for Disease Control and Prevention. "But we hope that most people are of good will who want to do the right thing."
At the news conference yesterday, Dr. Gellin said of the price gouging: "Obviously, that is a great concern to all of us and these are the stories that we need to track down because it is immoral."
While criticizing the ethics of the practice, Dr. Gellin did not say which government agency would investigate and what the penalties would be if distributors or physicians were indeed price gouging.
My take: it would be nice to see some explanation from Dr. Gellin and colleagues as to why raising prices is bad. The simple claim that it's immoral doesn't really do the job.
In response to my post about buying places in the flu vaccine line Nick Weaver mentions the issue of externalities:
Aggregate cost of a hospitalization is what, $20k? Lets assume a fatality is only $100k (It's mostyl old people, they're gonna die sooner anyway, rather than the $1M figure you see in a lot of places). Remember, thanks to insurace, medicare, etc, many of these costs are born by the overall society, not just the corpse.
US Population is ~30M over the age of 65. Thus lets assume the "immune system vulnerable" group is ~50M for which to take the deaths/hospitalizations figure, thats .4% hospitalized, .06% dead.
So thats $80 expected cost for hospitalization, $60 for the deaths.
If you REALLY want a market solution, it should be market rate only for the non-triage, or those on the triage list should automatically get a $140 discount on the market rate, because thats the rough aggregate cost of a person from the vulnerable population who refuses the flu shot for monetary reasons (selling his spot).
It's certainly true that there are externalities here, primarily because you don't necessarily pay for your own health care, either because you have insurance or because the government covers it. It's certainly arguable that you shouldn't be able to buy someone's place in the flu vaccine line without somehow paing for these externalities.
However, when it's phrased that way, it raises an interesting question. In ordinary years there is plenty of vaccine and yet many people choose not to get vaccinated. This, of course, imposes externalities on the rest of us because those people increase their risk of getting sick and/or dying. Should we fine people who don't get vaccinated?
So, we're running short on flu vaccine and the moral consensus in the comments section seems to be that we should be reserving it for young children, old people, etc. which is indeed what health care providers are doing. So, try this thought experiment: Say I go to the flu shot clinic and pay some oldster to let me have his place in line (and not come back!) in return for $50. Is that OK or not?
The problem with saying "no", is that then you're on record as not being in favor of a Pareto-dominant exchange.
The problem with saying "no" (assuming you were in favor of triage in the first place) is that the logical extension of "yes" is when the word gets out that old people who weren't planning to get flu shots can make good money selling their right to flu shots and we start to approach a market system again.
In the wake of the Chiron flu vaccine debacle, health care providers have already started triaging flu shots. This kind of rationing of medical care isn't something you see a lot of in this country. It may be necessary, but I can't say I much like it.
That said, I have to ask: why not just raise the price and let the market sort it out?
This hasn't been a good week in medicine:
The Vioxx/heart disease link has been a concern for a while, but it wasn't clear if it was actually a side effect of Vioxx or a result of Vioxx not preventing heart disease as well as other NSAIDs. Now it appears to be pretty clear. That said, the additional risk was pretty small--7.5 additional heart attack, stroke, or blood clots per thousand over three years.
Ted Barlow writes
Recently, Christopher Hitchens wrote a typically deeply-principled piece in which he accused "most... Democratic activists" of rooting for bad news in Iraq and Afghanistan. I would be deeply ashamed anyone supposedly on my side cheering for death and injury to Americans and civilians. Unfortunately, Mr. Hitchens doesn't help me