The Hanover Collaboration

I was wondering what you guys think about Iraq’s current status. As much as I want to think otherwise, I think the current government has no chance of being successful—too many people want it to fail. 1) Their timeframe is simultaneously unrealistically fast and too slow to suck up the power vacuum. 2) they are not respected. 3) The police is not well integrated across religious/geographic lines nor is it well distinguished from the various militias. 4) There are too many religious leaders who would benefit from the strife, particularly foriegn. 5) The terrorists have seen how easy it would be to kick-start the civil war.

I think we need to start talking about a “controlled break-up.”

But then, civil war isn’t necessarily the worst thing that can happen. If you could convince all the militias to go out in the desert and duke it out by themselves, that might be the best outcome we can hope for. How about this stance, we just refferree the thing. Retreat to some carriers in the mediteranian and bomb the team with the worst ethnic cleansing. Isn’t that sort of what we did with Yugoslavia?

Hi there, I’m live blogging Juan Cole’s talk at Dartmouth. I’m here slightly late, but I’ve gotten a seat, they’re still doing introductions (pointless), and my CTS is sufficeintly medicated (for now.)

16:36 still intros. Still boring. U’m told it’s an important day for Muslims today.

16:39 We’re getting background now on JC. No word yet on how self absorbed he is.

16:40 JC will be talking about the new government that will be formed, “mmm, at a deliberate pace.”

1.2 billion Muslims in the world. Shiites are a minority branch, aroudn 10% of the world. There’s some kind of map on the overhead, but the coloring is unlabled. Also, no Israel (far as I can see).

He just claimed that Najaf and Karbola were the holiest sites for Shiite Muslims, but we know from previous blog posts of his that they’re trumped by Meca, Medina, and Jerusalem.

He hasn’t used the term ‘neo-con’ yet, but he’s talking about those people. I’m falling asleep. I think he just called Paul Bremer a Nazi, but I missed it. There was laughing.

16:06 Just woke up. Still talking about Bremer.

17:33 Woke up again.

Pecker and Bosner have been going on for some time about capital punishment and it has gotten me thinking.

I have long opposed capital punishment based on a vague (i.e. difficult to quantify) social argument. I am in the William Golding camp when it comes to my confidence in civilization. I think executions are giving in to some human nature that I would prefer stayed bottled up.

But suppose that execution (or torture-execution even) have a deterrent effect substantially greater than imprisonment. I am not the sort to argue that there can’t be comparison between innocent people killed, raped or tortured by the government and innocent people killed, raped or tortured by criminals. I’m not necessarily saying it’s a one to one trade-off but clearly such a trade-off exists and it isn’t far from one to one.

Since I conceede that much it is possible to construct a case where execution, or even torture might be the optimal choice. It just remains then how likely such a case is and how good we would be at distinguishing. To first order, deterrence goes like pubicity, as does loss of individual morality. Which means that deterrence and decivilization are going to be proportional. That’ll be helpful.

Back of the envelope: let’s say deterence is a generous 5% for murders—that’s 1000 people/year. Sounds like a lot but if we’re talking about quality of life of the rest of the population, that isn’t necessarily much at all. Let’s try to quantify the individual morality angle…

Let me suppose that by not executing people we could get 5% more like Sweden (as measured by murder rate). Some people would argue that the murder gap between states with and without the death penalty show that this is a gross under-estimate (I’m unconvinced—correlation is not cause). That would give us about 900 of those murder back plus save ~50 wrongful executions, and who knows, maybe there might be braoder social benefits as well. That 5% might be a good stepping stone for another 5% in a few more years. By contrast, I think the deterrence effect is going to max out.

It seems to me that the best long-term strattegy is hugs and puppies as opposed to public hangings. A disappointing result to be sure but I’m glad we thought this through (I started this post thinking you guys were going to end up convincing me to support the death penalty). Now I wonder what this analysis says about corporal punishment which I whole-heartedly support… I’ve got to get back to studying… Do any of you guys know how to compute phase-margin using a root-locus diagram?

Ever notice how some web sites, often banks or credit cards, have crazy strict rules for their passwords, like 2 numbers and 3 capitals, but no punctuation, except again underscores.

That drives me nuts, because my good passwords always have punctuation. Also, if you allow punctuation, then mandating 2 numbers actually makes the passwords worse rather than better.

But what really drives me nuts, is when my awesome password is followed by a “Security Question”, which has all of the power of the original password, except that it is extra guess-able / hackable. Who designs these security systems? What’s the point of making everyone come up with a new “secure” password, if you going to ask for a shitty password too (in case they forget). Might as well just use the shitty password.

When confronted with the “security question”, I suggest you type in a long string of random letters and numbers, and then forget that it even exists.

The democrats have an opening. Now what they need is a message.
Do you have that message? Go here with it. They might give you money for it. Heads up: 175 words plus 35 char title.

I sent them this one because it’s probably the only one of mine in here that could actually be implemented. Plumer not only agrees but hits his nails more squarely linking “actual” research

I’m very disappointed at how little discussion has taken place on whether or how to rebuild New Orleans. If I’m ever in a car accident, I’m sure as hell going to see if it’s cheaper to buy a new car than it is to fix my wrecked car.

And if my car is just going to wreck again the next time a hurricane comes through, then I’d have to work that into the equation too.

Vox Baby! and Becker-Posner both spend some time on the math, if anyone is interested.

I’m going to skip to my original thought: Why don’t we just punt on the whole dikes and flooding problem in New Orleans? Leave the place flooded. Write off the basements of the skyscrapers, and all the low lying houses (which probably only had poor people living in them anyway). Refurbish what is still above water. Build some kind of public transportation. Call it done. Like Venice with a football stadium.

I had a nice discussion with Anthony (who, I hope, is a future Collaborator) about this last night.

The conventional wisdom these days is that oil refineries are a bottleneck in our gasoline supply. The United States doesn’t have enough refineries to make all of the gasoline we want to use.

Question 1: Why doesn’t someone make more refineries?

Answer 1: It seems that the barrier to entry, in this case, is strict environmental regulation. Refineries work by boiling the oil into gasses, and distilling them fractionally. A lot of those gasses go floating away (and smell bad).

My Question: Why would you want a refinery in the US anyway? I think I’d much rather set up my refinery in Mexico or North Africa or anywhere that 1) I could pay my employees $.25 an hour and 2) there weren’t air quality regulations.

Why Don’t We Import More Gasoline? My guesses are:

1) For some reason gasoline is hard to ship. Maybe it explodes too easily, or something. Counter evidence: We DO import some gasoline.

2) Gasoline standards are heavily localized and boutiqued, from state to state, and sometimes county to county. If you want to sell gasoline across the US you need thousands of different variations of gasoline. Obviously, if you’re trying to fill an enormous boat with gasoline to sell this is going to present some logistical challenges.

Apologies for the inconsistent style. Caffeine intake, I’d guess.

I have a poster in my room that sums up one of my frequent thoughts.

Pharmacists, their job is to do what exactly?
1) Fill prescriptions. Count n-pills from the hopper and put them in the bottle. This has been in the robot domain for 50 years—totally solved.

2) Check that the person doen’t have medication interactions. They already use the computer for this. The pharmacist just looks to see if the screen says “warning, possible interaction”.

3) Ask you if you want anything explained. If you’ve ever asked them anything you probably noticed that they just read to you from the card that was going to be in the bag with the pills.

For this job we need certified experts?

Worse still, we get certified experts with morals That’s another nice thing about robots. They don’t worry about their souls. For that matter they don’t worry about those of their clientele or those of their clientele’s potential blastocysts.

The irony of boycotting CVS and Walgreens (in favor of Eckerd and Brooks) on “religious grounds” is just too irresistable.

This is going to be my first boycott ever. CVS is actually a store where I buy stuff. They had 3L of grape soda for $.25 the other day. I’ve recently noticed the market share they have around here. CVS wins two to one in the region and 10 to 1 on my most recent commute. Despite that, I am confident that I can inflict more hardship on them than on myself.

In the meantime we should remove the artificial barriers to entry for robots who want to be pharmacists.

Drugs this time.
Slate has a good one.
NYT has a funny one.

If there are sets of things that I am not allowed to buy all at once, I would like to be notified. Sometimes I buy a lot of stuff all at once. For instance, if I think that the price of cold medicine will not be lower (relative to my income) for n-years, I buy roughly a (1 – e^-gamma*n^)*n year supply, where gamma is a discount factor whose details are beyond our current scope. The only reason I haven’t gotten in trouble is that an n-year supply of cold medicine happens to be zero. What about when I go to Walmart looking for parts for my next Rube-Goldberg machine, there are a lot of ways I could put it together. I’m sure some of those ways make meth at some stage.

Second, if a store is recording a transaction history that can identify me by any means (which in my mind would include a timestamp and security footage), I think they should be obligated to say so outright.

Third, if some guy comes up to me and is like, “Hey, do you have any tinfoil?” I’d be like, “Sure, here you go.” Fine right? It’s got to be. If some guy comes up to me and is like, “Hey, do you have any tinfoil, I need to finish up a cook.” I’d be like, “Sure, here you go. What’s a cook?” Blam, tackled, submission hold, arrested.

Luckily, I’m white so they’d let me off on bail, I’d get a sweet lawyer and beat that those charges like Rocky, seeing as how I am a good (white) kid.

As the Slate article points out, if we just made it legal, there would be no point in having poluting meth labs and people could get cold remedies again. As a bonus the meth users wouldn’t get so messed up on residual solvents.

The case is even stronger for MDMA. It has all the same issues as meth but there I think that safe and potent pharmacuticals could be developed using R&D based on MDMA. You could say the same for meth, but those drugs have already been developed and so meth isn’t particularly interesting anymore.

It isn’t like Phyzer, (or Merck as the case may be) couldn’t get the license you need to study the stuff. It is that they don’t want to spend the money to develop a drug that works too well and gets scheduled level 1.

Man was I excited when RSS showed a Thimerosal article on Slate. It was just the other day when I was harping on that one.

Better still, the subtitle had “read the studies…” Man was I excited.

Then what do I get? A lousy editorial with no worthwhile links. If you follow the links out three degrees of separation I found the abstract of a survey article of articles that would be worth reading if I didn’t have to pay money for them (I would have had to pay money for the text of the survey as well).

That and this freaking Constant Gardener banner that keeps getting in my way. You’re on thin ice, Slate.

So let’s give it a shot on our own.

Case closed, no need for more study.

I was just thinking about how pacifist I am. I was working through it the obvious way. Was this war worth it? This war? How about this mistake, should someone have sent in the troops? Of course, if you sit out all the time it makes you more prone to future attack, like Switzerland. But I’m ignoring any affect like that.

I’m deciding that not a lot of wars were worth it. From a lives standpoint, it is usually pretty easy. If you count injuries, it tends to get even easier. But then there are some harder things, like “freedom”. It’s not like I value lives that much. It is just that I think “freedom” is pretty over-rated. I could imagine circumstance that might motivate me to rebel against a really opressive government. But they are pretty absurdly opressive. Pol Pot sure but maybe not Saddam.

Let’s say that I were an Iraqi and I could by sacrificing, say, 50,000 good guy lives instantly transform my country into a democratic utopia on par with USA. I would be hesitant. Even if I used the most extreme projections, I have to admit that the BGs weren’t going to kill anywhere near that number. And how much additional freedom is that really? There might be a lot of things that I can do that I couldn’t before, but are they worth i? I’m pretty sure I would want to wait Saddam out. But then I might try to emmigrate.

If you can move to a different country, maybe that would make more sense. How about we let people in shitty countries immigrate more easily? That would be nice. Where was that touching plaque, you know, with the ”... yearning to breath free.” bit? The statue of liberty? Does it still say that?

Of course when someone just up and attacks you, you have to fight back. In those cases, there isn’t a lot to think about. But I am talking about real invasion type of things. Like Israel ‘67 or Europe WW2. I am not talking about sept 11 or (Oklahoma city for that matter). I am not even talking about Pearl Harbor. As unpaletable as it may have seemed, “Oh Japan, you really got us there. I think we are going to sit the rest of this out.” was an option.

Whoops, I wasn’t even planning on talking about Iraq. Forget I mentioned it. What I wanted to talk about was the civil war. Having never been South of W. Virginia for more than two weeks at a stretch, I never really questioned the civil war. The GGs won, the nation stayed together and we’re all better off.

But let’s say it got played differently. The South secedes the North is like, “Call me if you get lonely.” and the South is like, “We’ll still be friends.” I have a hard time imagining enough ill effects to make it worth the war, 3.5 million slaves notwithstanding. How can I say that?

600,000 dead 400,000 wounded and the freed slaves’ lives didn’t improve very substantially. They certainly did improve. And freedom (not in quotes) is definitely pretty nice. But if you asked a slave how many people would he sacrifice for his freedom (even if you didn’t tell him about sharecropping, segregation and all that) I think he would come up with a lower number. Maybe 200,000/133,000. And, if you ask me, slavery was on it’s way out anyway. I think one generation of political effort would have accomplished just as much.

So let’s say there is this dangerous thing that people want to do anyway. Why you may ask? Clearly because it’s really fun. So we have something concrete to talk about, let’s call it rock climbing.

So a lot of people are rock climbing and some are killing or hurting themselves. So the governent in loco parentis decides it should stem the tide of our nation’s youth being smashed on the rocks. It finds itself facing the decision of how to go about it.

On the one hand, they could keep it legal and regulate. They could certify the gear, require licenses, tests, saftey courses. On the gear side, the certification is great. Climbers would only buy certified gear which would force all the manufacturers up to a safe standard. On the licensing side, it might be less effective because the more restrictive the requirements got, the more people would circumvent it. But it would probably have some positive effect.

On the other hand, you could make it illegal. But as I said, it is very fun. Some people are going to do it anyway. You are going to eliminate the people who are unwilling to take the legal risk. You can increase that number somewhat by making the penalties unimaginably harsh. But keep in mind that these are the same people willing to risk their very lives rock climbing.

But what about the gear in the illegal case? It will become mindboglingly expensive and simultaneously very shoddy. A gear dealer is held to very little standard since there is no recourse for someone he rips off. Well none exept violence. The expense could conceivably drive people to stealing, though I think this aspect is often over-emphasized. (Has anyone shown a decent correlation between theft and cigarette tax? If this affect really existed, I would expect it show up there.) And since these large transactions are unrecorded, the money could be more easily used covertly, for say, financing terrorism, to use a buzzword.

Let me digress on the cost for a second. Regulation might cost a fair bit but keeping it legal gives you grounds for brutal taxation which in the cigarette case might be working okay. Enforcement on the other hand requires paying lots of people to physically prevent people from rock climbing. Maybe they can try to recoup some of that with fines. But we aren’t talking fines, we are talking long mandatory jail sentences.

So what are the outcomes? Illegal: less people doing it with increased risk. Legal: more people doing it more safely. I guess we are going to need some more quantitative measures. We also need to look at the secondary effects. I think everyone here has a decent estimate of the relevant parameters for rock climbing. Everyone get “regulate”? Now how about alcohol? nicotine? prostitution (this one has some really primary “secondary” effects)? ecstacy? cocaine? heroin?

You can check your answers against mine in the comments.

I’ve been wondering when the media will get a hold of this one. Maybe now is the time.

Has anyone else heard of this chemical? Way back when they were making kiddie vaccines, they needed some preservatives. Somebody had the bright idea of making one with mercury in it. Yes that’s mercury, as heavy metal as in mad-as-a-hatter. Mercury is not good for you and people have known it for two hundred years.

Now, don’t get me wrong, you can only do so much testing and then you have to release your vaccine. You can never be sure. But at some point, you also need to admit your mistakes and take steps to rectify them. This one should have got caught right away. The article makes a good case that the reason it didn’t was corruption. The selling point for me is when the researching body said “case closed, no need for more study”. Scientist just don’t say that ever.

Cases can be made for secrecy on the grounds that we would like to avoid a vaccine scare. Other cases could be made for simple stupidity. Whatever the relative weightings are, this stuff got used for far far longer than it should have and people have a right to be angry about it.

I think we probably should offer some legal protection to companies (like pharma) as long as their methodology is rigorous and we can trust them. But it looks like in this case, there probably ought to be some lawsuits. Who was it again who blocked such suits? Frist? What was the justification again? Terrorism? ... what?... Did I just make that up or was that in the real world? I have had trouble telling ever since my last tetanus booster. Or maybe it was the flourinated water.

Let me start with a robot model. Let S be the space of all states, A be all actions, O(t) be all the information available at time t about the state S(t). The goal of the robot is to generate a policy Pi in (P(S),t):A. That notation means “a time-varying mapping of probability distributions over states to actions.” Our policy PI should try to maximize some sort of value function F (S*,A*,t):R, that is, a (time-varying) function that maps sequences of states and actions to the reals. There are a few more things that we need to know before we can begin. We need to know a model for how the world acts and a model for how we percieve the world. So we need (time varying) probability distributions P(S’|S*,O*) and P(O*|S*).

There are a number of independences that we can try to exploit but those wont be relevant to the discussion that will follow. What is important is the framework. We seek a policy. We have the value function that we are trying to maximize and we have probabalistic models for how the world operates and how we receive information.

If we have that information, it is possible to come up with an optimal policy. Of course tractability is a concern but lets say we have a very very powerful computer and that we aren’t strictly concerned with optimality, just with a “good” solution.

The robot, if left on its own, might initially have a bad estimate of any of the three components and use that to make a bad policy. This bad policy might or might not generate data to correct the model. If it doesn’t, the robot behavior will probably not converge to a good solution. But if we are equiped to give the robot high quality initial estimates of those three components, the robot will pursue better policies that will lead to better data. That stands a very good chance of convergence to a “good” solution.

The interesting point is that we don’t have to furnish an initial policy, only information with which to generate one. This might come as a surprising result to some.

When I have kids, I am going to make it clear that my rules are based merely on my more acurate model estimation. As my kids’ model improves, they will be allowed to make more of their own rules. If everybody behaves rationally and honestly this should work just fine.

The Supreme Court (finally) handed down it’s opinion on the case of Ashcroft vs. Raich this morning. The case involved a California woman who purchased locally grown marijuana to fill a prescription to treat her chronic . This is totally legal under California law. The Feds busted her none-the-less. At issue is whether the Constitution’s “Commerce Clause”, which empowers Congress alone to regulate interstate trade, can be applied to the entirely intrastate sale of MJ.

6-3 for the government, written by Justice Stevens. When drugs make it to the Court, don’t bet on the drugs.

Justice Scalia concurs, ”... Congress may regulate even those intrastate activities that do not themselves substantially affect interstate commerce.”

[NB for Originalists: The Text actually says Congress shall have the power “To regulate Commerce with foreign Nations, and among the several States, and with the Indian Tribes.”]

Federalism, the belief that states should have substantial power to make their own rules, is a value traditionally claimed by conservatives. I’ve said it before, and I’ll say it again: there are only fair-weather Federalists. I’d like to see some conservatives coming out against this decision. Especially conservatives who support drug decriminalization.

I’m looking to you, Bill O’Reilly, and my I suggest the phrase “activist judges”.