Sat, 26 Nov 2005

Expert Analysis of Risk

You see this all the time. An expert stands up and says "Through my expertise, I see a problem that nobody else sees." If you listen a little more closely, you find out that the reason the expert concludes that nobody else sees the problem is that they're not paying him money to solve it. That may seem excessively cynical. I don't think so. Being experts, they overestimate the importance of their field of study (no blame: this is the human condition).

The general public lives in a sea of risk. You know what they say: "Life is short and then you die." For some people life is shorter than others, if only because humans are fragile. People perceive some risks irrationally, particularly when you get into very small risks of very bad things. I think that that is simply because people cannot make the proper mental trade-off. Which risk is worse: the risk of dying in a coal-related accident or the risk of dying in a nuclear accident? Mathematically, coal is a bigger killer, and yet people are opposed to replacing coal with nuclear power.

Nonetheless, people who mis-estimate risks under their control are likelier to die. In this way do trees serve to eliminate the imprudent from the pool of automobile drivers. It's reasonable to assume that people are correctly evaluating the risks in their life. So when an expert says "I know better than you", they're technically correct in their field of expertise, but their recommendations do not automatically make for good policy.

Posted [12:33] [Filed in: economics] [permalink] [Google for the title] [digg this]

It's a feature, not a bug!

In the world of computer programmers, we have a phrase: "It's a feature, not a bug!" We use that phrase when somebody doesn't understand the subtleties of how something should work. They apply a naive analysis to it and conclude that it was a mistake, and needs to be fixed.

Many people do not understand the structure of the US government. Sadly, they are as likely to be Americans as not (the blame for which I lay at the feet of Social Studies as taught in government schools.) Because they fail to understand the subtleties, they call for the government to be fixed. Usually this entails centralization of power.

This morning on our local NPR customer I heard a news report about the dangers of chemical plants. The main thesis of the story was that mere citizens don't understand the risks of chemical plants, because if they knew what the experts knew, they would call for federal laws regulating these plants. There are two problems with this idea. I'll tackle the problem with expert analysis of risk in another post. The other problem is the call for federalization. The NPR reporter whined at the end of her report "Without a federal law, when one state restricts chemical plants, that only transfers the problem to other states."

The structure of the US government is designed to handle error. Part of being human is making mistakes. Part of being a god is being omnipotent. So did The Christ know he was making a mistake as he was doing it? If that never happened, then he was either not human or not a god. But I digress. Complete knowledge is not available to us. What we know, we know because we have experimented.

The US government is one vast, continuous experiment, or so it's supposed to be. Unfortunately, we have greatly reduced the amount of experimentation in space, and turned it into experimentation in time. That's just plain stupid. Everybody knows that "many hands make light work." That just says that work goes faster if you have lots of people doing it. The same effect works in government.

The original structure was designed to be an parallel experiment in space. The federal government was strictly limited in the laws it could pass. All other laws were to be passed by states. Of course, not all states would pass the same laws. Thus, some states would make mistakes that others would not make. That's how science works: you have a control and you have a test. You keep one thing constant and you change the other.

We have destroyed all this experimentation by allowing federalization. We no longer restrict the federal government, and in doing so we have given up science. We no longer have a control. Everyone is a test subject, so we never really know what are the effects of laws. Without having US citizens who are not subject to those laws, we can't tell if they had good results or bad. Also, instead of running multiple experiments, we can only run one experiment across the entire country. If that experiment fails, as some people have said the Telecommunications Act of 1996 has failed, all of that time has been lost. With a more distributed set of laws, other states could have been trying something different.

The other problem with federalization can be seen by flying over the US. The many regions of the US are radically different. We have mountains and streams and lakes and deserts and plains and cities and forests. How can anyone think that one law could fit everywhere? Take, for example, telecommunications. The way you address "tele"communications depends on how far is your "tele". Telephone service in an apartment building is vastly different than telephone service out west where it's not unusual to have miles between customers. Beehive Telephone serves rural Utah and Nevada. They own an airplane to fly between their central offices. I can't imagine any eastern telephone company needing an airplane.

I'm not opposed to the use of governmental power. Many problems are easier to solve by forcing everyone to solve a problem the same way (e.g. water and sewer systems). I'm opposed to the use of governmental power in inappropriate situations. But how do we, as fallible humans, to discover which solutions are inappropriate without experimentation? If you agree with me that federalization is a philosophical mistake, please contact your state representatives and tell them to take back the power that is rightfully theirs.

UPDATE 11/16: Roy asks "How exactly are they supposed to do that?" Roy, you're trying to solve problem #2 before you solve problem #1. Problem #1 is to get the state legislators to realize that the federales have stolen their power. Each individual citizen is relatively powerless. In order to magnify their power, they need to convince the powerful to do their bidding. Since power seeks more power, the most effective path is to get the slightly less powerful to attack the more powerful. Right now, the most powerful single entity on the planet sits on Capitol Hill. Collectively, the state legislatures approach them in power, but first they must be convinced to exercise their power. Exactly how they do that is problem #2. First things first.

UPDATE 11/16: Scott contributes two examples:

Flush toilets - Al Gore (and many others) thought it was great idea to limit flush toilets to 1.6 gallons per flush. The unintended consequence is that many people flush *twice*!. However, while the dry western states might very well have thought such a law was a good idea and passed it on their own, does someplace like New Orleans, literally drowning in water even when not flooded, really need to suffer through such a restriction? I was in New Orleans back in 1991, and I saw city employees clean the streets with firehoses!

911 service - I live in Israel and got a Packet8 VOIP service earlier this year. One reason I chose this service was the cost, which didn't include the overhead of 911 service. Packet8 was going to eventually offer 911 as an option. But no, that wasn't good enough for the Feds. They completely overreact to a few people who obviously didn't read the not-so-fine print that their VOIP service 911wasn't the same as standard 911, and instead of merely requiring more visible notice or disclaimer, they required all VOIP services to provide 911, whether the user wanted it or not. I live in Israel. I want an American phone line for various reasons. I don't want or need 911 service and I don't want to pay for it. I had a choice before. Now I don't.

Posted [12:32] [Filed in: economics] [permalink] [Google for the title] [digg this]

ipkg versus dpkg

On the maemo-developers list, people have been arguing about the advantages of ipkg versus dpkg. Hopefully I can show rather than explain why we (the folks prefer ipkg.

I have Debian 2.2 on a server and Familiar 6.2 on a handheld. Since we're talking about using either dpkg or ipkg on a jffs2 filesystem, all space consumption is given as a compressed tarball. Here is the space consumed by the package manager's overhead for installed packages and for available packages against ipkg and dpkg.

packages size bytes/package
ipkg total 378K
ipkg available 939 317K 388
ipkg installed 147 61K 415
dpkg total 10620K
dpkg available 15272 8660K 567
dpkg installed 301 1860K 6180

Here's how I got these numbers:

ipkg total is:
tar cfz - /usr/lib/ipkg | wc -c
ipkg available is:
tar cfz - /usr/lib/ipkg/lists | wc -c
ipkg packages installed is:
ipkg status | grep ^Package: | wc -l
ipkg available packages is:
ipkg list | wc -l
dpkg total is:
tar cfz - /var/lib/dpkg | wc -c
dpkg available is:
tar cfz - /var/lib/dpkg/available* | wc -c
dpkg packages installed is:
grep '^Status: install ok installed' /var/lib/dpkg/status | wc -l
dpkg available packages is:
grep '^Status:' /var/lib/dpkg/status | wc -l

Posted [11:24] [Filed in: 770] [permalink] [Google for the title] [digg this]