Monday, August 28, 2006

Theo Jansen on YouTube

I was first introduced to Theo Jansen at GEL, Mark Hurst's remarkable Good Experience Live conference in NYC. I never knew how to describe Jansen's beautiful kinetic sculptures, or "animals" as he sometime refers to them, but I'm thrilled he's finally on YouTube so people can go see for themselves.

Thursday, August 24, 2006

Internet duds

Snakes on a Plane got huge internet buzz, but flopped at the box office.

Why?

Take the priors -- how well do you think a tongue-in-cheek movie about a bunch of snakes going loose on a plane, with chaos ensuing, will do? It probably will not be a blockbuster, but may find an audience among young males. Certainly it does not sound like anything with mass appeal.

Now add internet buzz and suddenly the expectations for the movie rise. People think that how often they see something is correlated to how common that thing really is, which is why most people think disasters are fairly common when in fact they are quite rare (but highly publicized).

The movie comes out, and does about as well as a movie about snakes on a place could do, which is not well. People say the internet is a dud.

I think the crowd at Slashdot have the right take on this: the Internet buzz probably helped it do better than it otherwise would, but that baseline was low. I should probably invoke Bayes theorem somewhere in this post.

Pogue slams the Chocolate

My wife's flip Samsung mobile was disintegrating, and my candy bar Nokia had developed an "auto-repeat" function (where, if I pressed "3" the phone would dial "333", except when it didn't dial anything at all) so it was time for new cell phones.

I bought the latest candy bar Nokia Verizon offers (I've found that candy Nokia's have great durability and reception, with OK voice quality) while my wife got a free LG flip something.

The store was heavily promoting the new LG Chocolate, but in my experience phones that look cool don't work. Sure enough, David Pogue slams it in the NYTimes.
Whenever I review a product this badly designed, I just stare at the ceiling and try to imagine how it could possibly have gotten out the door. Haven’t successes like the iPod and the Treo taught the marketers anything about making things work simply and well? It’s stunning that nobody in a position of power at LG or Verizon actually tried this thing, tried pressing those infernal passive-aggressive buttons, and realized that the Chocolate is a usability disaster.
The cell phone industry seems to have forgotten that people buy phones to talk to their friends, and if a phone does not do that well, it's not worth having.

Wednesday, August 23, 2006

Teach kids to play cards

I very much like this post encouraging parents to teach their kids to gamble.
1. For younger children, it's a fun way to introduce them to basic maths.
2. Gambling is like science - a way of testing beliefs against evidence.
3. Gambling teaches kids about probabilities.
4. If we teach kids about probability, we prepare them to ask: how are estimates of probability formed?
Teaching gambling, then, is a good way to teach people how to think clearly. And this, surely, is the most important lesson a school can impart.
However, once you learn about gambling, you also figure out that the house always wins, and the smart move is not to gamble at all. I prefer games that mix chance, skill, and human interaction, like poker. Poke teaches you to put your money where your mouth is, know that doing the right thing does not always work out, and sometimes you can be saved by dumb luck.

Worst Malcolm Gladwell article ever? Update

Gladwell himself has a response to my recent post critisizing his article in the New Yorker about corporate pensions.

One of the points Gladwell made in the piece was that tying pensions to corporations was a bad idea because if the corporation went down, then those employees were in trouble. He recommended tying pensions to broader regions, such as states, or maybe even the whole country. I point out the obvious fact that the state pensions and national pensions we do have -- government employee pensions and social security -- are in the same straights as corporate pensions. Gladwell's response?
Social Security issue hasn't convinced me that the program is in all that much trouble--that is, with a number of not entirely painless (but not debilitating) adjustments now, we can avoid a lot of the trouble down the line. Put it this way: would you rather, as a retiree, put your faith in the federal government or General Motors?...

...she pointed out that its not pensions we should we worried about so much as retiree health care. But even there. General Motors is currently $40 billion behind in funding its retiree medical obligations. The federal govermment is behind as well. But I'd still rather take my chances with the feds than with GM.
So, to paraphrase:
1) I'd still trust the government before I'd trust a company, and
2) We should really be worried about health care, not pensions

Gladwell's first argument -- that it is betted to trust a government who will only renege slightly and shaft other people on your behalf rather than trust a company who will renege a lot and shaft you is true as far as it goes. It's also a weak position when there are perfectly good zero-shaft options out there (private accounts) which, if administered correctly, would have low operating costs and mandatory contributions. But I do not want to get into policy prescriptions here, I want to take issue with Gladwell's argument that broader paygo pension bases are somehow immune from the "promising what you cannot deliver" issues with corporate paygo pensions.

Secondly, the old healthcare/pension switcheroo is a common one and is also correct. Uncovered liabilities in healthcare (both medicare, medicaid, and corporate healthcare) overshadow the unfunded liabilities in pensions. This does nothing to solve the problem in pensions, however, and it makes the pension problem harder to solve by raising taxes on those still working (since you've already raised taxes on them to cover the sick).

Lastly, I am not sure why people talk about pensions and health insurance in the breath -- the two are quite different. I don't see pensions as being any kind of insurance at all -- after all, what is less uncertain than old age? The risks you run around aging are 1) living longer than you thought you would, so you run out of savings and 2) the asset your savings are in goes down the tubes. The first risk can be dealt with using annuities (which are very unpopular) and the second through diversification (which corporate pensions, critically, do not have). (Thanks to Reemer for the trackback)

Monday, August 21, 2006

It's just not cricket

Pakistan is current on tour in England playing a series of test matches. In the latest match, the umpires decided that Pakistan was tampering with the ball, and awarded England 5 runs. Pakistan decided this was a blow to national honor and refused to leave the dressing room after the tea break. England was awarded the match by default. This win did not change the outcome of the test series, as England were already two games up.

5 runs is peanuts in a test game -- where scores are often in the 300s and half the matches end in a draw. But pride is powerful in Pakistan, and the rest of the series has been called into question.

Did Pakistan tamper with the ball? Quite possibly. Was the umpire being extra mean to call them on it? Certainly. Should they have forfeited the game by sulking in the dressing room after tea? Absolutely not.

Worst Malcolm Gladwell article ever?

I like most of what Malcolm Gladwell writes, but this latest piece in the New Yorker is awful -- he gets his economics just wrong.

In it, he talks about "the dependency ratio", which is the ratio of people who put money into a pay-go scheme compared to the number of people who take money out of a pay-go scheme. A "pay-go" scheme is one where current beneficiaries are paid out of the same pool of money that future beneficiaries pay into, and there are many public and private examples including social security, corporate pensions, etc. The alternative to a paygo system is a fully funded system, where an individual can withdraw what they contributed earlier, plus interest -- examples of this include private bank accounts and 401(k) plans.

In a paygo system, there is a gap the minute the program begins because early beneficiaries have not had time to contribute enough to the system to cover their withdrawals. This liability never really goes away, but it can be postponed way into the future so long as the system remains solvent from a cash flow perspective (each period, enough money goes in to cover expenses going out). This cash flow entity is determined by the number of people who take money out of the system vs those who put money into it -- Gladwell's "dependency ratio".

The workers of Toledo needed pensions. But, he said, the pension plan should be regional, spread across the many small auto-parts makers, electrical-appliance manufacturers, and plastics shops in the Toledo area. That way, if workers switched jobs they could take their pension credits with them, and if a company went bankrupt its workersÂ? retirement would be safe. Every company in the area, Gosser proposed, should pay ten cents an hour, per worker, into a centralized fund...

The labor movement believed that the safest and most efficient way to provide insurance against ill health or old age was to spread the costs and risks of benefits over the biggest and most diverse group possible. Walter Reuther, as Nelson Lichtenstein argues in his definitive biography, believed that risk ought to be broadly collectivized. Charlie Wilson, on the other hand, felt the way the business leaders of Toledo did: that collectivization was a threat to the free market and to the autonomy of business owners...

The United States, by contrast, has over the past fifty years followed the lead of Charlie Wilson and the bosses of Toledo and made individual companies responsible for the care of their retirees. It is this fact, as much as any other, that explains the current crisis. In 1950, Charlie Wilson was wrong, and Walter Reuther was right.
Is Gladwell on crack? It is true that old-time unionized companies made promises to their employees that they cannot keep and now someone is going to have to cover the difference -- but that is not a "dependency ratio" issue, it's a consequence of people making promises that they could not keep.

Gladwell goes on to argue that if, instead of corporate pensions, we had some sort of regional pensions system instead then everything would be OK is equally hallucinatory. We do have regional pensions -- most government workers in a state are covered by a pension in that state and last time I checked, those were deep in the red too. And how about the mother-of-all regional pensions, Social Security, which covers every man, woman, and child in the United States. Oh look -- it's deep in the red as well.

So, if company paygo plans underfund, and State paygo plans underfund, and National paygo plans underfund, you would think that paygo plans, simply underfund as part of their very nature. Unless you are Malcolm Gladwell, of course, where you would instead recommend State and National paygo plans.

Gloriously, he ends on something as mad as a box of mittens:
What happened to Bethlehem, of course, is what happened throughout American industry in the postwar period. Technology led to great advances in productivity, so that when the bulge of workers hired in the middle of the century retired and began drawing pensions, there was no one replacing them in the workforce. General Motors today makes more cars and trucks than it did in the early nineteen-sixties, but it does so with about a third of the employees. In 1962, G.M. had four hundred and sixty-four thousand U.S. employees and was paying benefits to forty thousand retirees and their spouses, for a dependency ratio of one pensioner to 11.6 employees. Last year, it had a hundred and forty-one thousand workers and paid benefits to four hundred and fifty-three thousand retirees, for a dependency ratio of 3.2 to 1.
The argument here is that increases in productivity have worsened the pension crises my making the dependency ratio even worse than it would have been otherwise. Malcolm, my friend, increases in productivity is all that's been keeping these companies in business and it's all that's going to have them (and us) be rich enough to pay that liability gap that the paygo systems created when they were first put in place.

As a matter of fact, I think that increased productivity, which is what makes us all richer, is the only reason to have a paygo system in the first place. When Social Security was first implemented, its immediate beneficiaries were old people, who were poor as a rule, and coming out of the Great Depression in the 1930s. All of us today are far wealthier than those initial beneficiaries -- thanks to improvements in productivity -- and if we need to pay higher taxes so those folks back then got some help, so be it.

Incentives for leaders

I recommend this interview of Bruce Bueno de Mesquita where he looks, through a public choice lens, on the incentives various forms of political rulers have given how the government is structured (themes he details in his book: The Political Economy of Power).

He identifies two groups within a population -- the "selectorate" (those who get to select the ruler) and the "ruling coalition" -- a subset of the selectorate who keep the ruler in power once they are in.

The broader a group of people who make up these two groups, the more the ruler needs to pay off his supporters in public goods rather than private goods, since he needs to keep a lot of people happy and it is easier to do that with good roads and reliable electricity than lucrative government contracts. When the ruling coalition is a small (as a % of the selectorate, and as a % of the population) then the leader can maintain power more efficiently through the provision of private goods needn't worry about public goods so much.

The interview is too long, and the interviewer too twee/snarky, but I think the essential insight is correct -- that leaders do not exist in a vacuum, they need support to stay in power, their primary interest is to stay in power, and they will be *forced* by their environment to confiscate goodies from everyone to pay off their supporters. If they did not redistribute wealth, in this way, their supporters would replace them with someone who did.

This model has some interesting consequences:

1) Don't give any foreign aid.

Foreign aid acts like natural resources -- namely, it is a source of wealth that the rulers get that does not need to be taken from the people. This wealth, therefore, can be directly transferred to cronies who, in turn, keep the ruler in power. The ruler will sell policy concessions to get this money, and the policy concessions will likely rile the population, who will hate the donor country for propping up their leader, while wanting to move there themselves to enjoy the public goods that that country provides but their country does not.

2) Increasing the selectorate, and the ruling coalition is critical to making rulers focus on public goods over private goods.

The easiest way to increase the selectorate is to allow people to vote. Essentially, buying people off with public goods is much better than buying them off with private goods, so anything that makes public goods provision more important to a ruler is beneficial.

3) If you can't vote, and don't have guns your opinion does not matter.

Note that if you are not part of the selectorate, and you are not part of the ruling coalition, your opinion will have no impact on policy. Most of the population in dictatorships are in this situation, which is sad, but true. When dealing with such countries, it is a waste of time to consider this disenfranchised majority -- if you want to influence the leader you should focus only on the groups that 1) selected him, and 2) keep him in power.

Wednesday, August 16, 2006

The power of incentives

Driving in places like Karachi is much more chaotic than driving anywhere in the US (even Boston). Yet, despite the craziness on the road, people do not seem to die much. Moreover, when safety measures (such as seatbelts) are introduced, you do not see any decrease in mortality. Drivers choose to consume "safety" by driving faster, keeping the risk they expose themselves to the same but getting to their destination faster. Pedestrians come out poorly in this calculation, but whatever. The most important element in safe driving is to what degree the driver is paying attention.
Reversing decades of conventional wisdom on traffic engineering, Hamilton-Baillie argues that the key to improving both safety and vehicular capacity is to remove traffic lights and other controls, such as stop signs and the white and yellow lines dividing streets into lanes. Without any clear right-of-way, he says, motorists are forced to slow down to safer speeds, make eye contact with pedestrians, cyclists and other drivers, and decide among themselves when it is safe to proceed.
If you were actually serious about dramatically reducing deaths from car accidents, you would ban seatbelts from the driver's seat and affix a large metal spike to the steering column, pointing towards the heart. With that installed, I don't even think we'd need speed limits.

Terrorism on the margin

Bruce Schneier has a good post on why stepped up security at airports -- particularly in light of the thwarted London terror bombing -- is ineffective:
It's easy to defend against what the terrorists planned last time, but it's shortsighted. If we spend billions fielding liquid-analysis machines in airports and the terrorists use solid explosives, we've wasted our money. If they target shopping malls, we've wasted our money. Focusing on tactics simply forces the terrorists to make a minor modification in their plans. There are too many targets -- stadiums, schools, theaters, churches, the long line of densely packed people before airport security -- and too many ways to kill people.
His point is that a would-be terrorist has such a plethora of targets to pick from, that hardening any particular one (airports, planes) has little impact overall because there are so many substitutes. On the margin, there will always be easy alternatives to target.

Tuesday, August 15, 2006

Getting the basics right

Moving to California has been an interesting experience with regards to buying stuff and changing services. I am in the market for all kinds of things that I usually don't care about -- things like banks, cars, utility companies, etc. Some of these services do well, some poorly.

For things that go well, I cannot recommend Carmax enough. They take all the hassle and worry out of buying a used car. It's a great experience overall. The only downside is once you've actually bought the car they take an age to clean it, gas it, and get the keys in your hands. You want to drive away with your new purchase immediately!

Car insurance is another issue. MA continues to defend its title as "most insane car insurance regulations in the US" by not letting you cancel your insurance until the MA DMV has your plates. This means you need to mail your plates back to some PO Box in Natick, and wait for a government employee there to punch into a computer "yes, they have returned the plates". Until then you need to continue to pay the usurious insurance rates they charge. This is assuming nothing gets lost in the mail.

There is also a well respected weekly periodical which I shall not name which does a lousy job of taking your money and mailing their magazine to you in exchange. I had let the subscription lapse since I was moving, and I get a phone call from them offering me a great deal if I sign-up now. So I do so and pay over the phone with my credit card, giving them an email address as a confirmation. I hear nothing back from these people, get no email, and worry that I've been phone scammed. I call them up and they have no record of the order, nor do they have any idea who might have placed the call. That said they cannot say for sure that the call was bogus because sometimes orders takes weeks or months to show up, and it seems there are free lance elements selling subscriptions for this periodical all over the globe. Eventually someone from the periodical recommends that I simply call my credit card company and cancel, since it's the only way to be sure. A most unsatisfactory conclusion since I really did want to give them my money in return for their product. I asked them, if I signed up for a new subscription there, could they offer me the same deal, and when would I start to get the magazine. They told me they could not offer the same deal, and if I still chose to sign up, I would get the first issue in the mail in 8-12 weeks. Which in my book is 2-3 months.

Ye gads.

After dramatic escalation, involving calling people I knew who worked at the company, it's all sorted out (I think) but you would think that taking money in exchange for product would be something the company was concerned about and be very very good at. It's remarkable how companies can lose sight of nailing the basics of their business. And it's a delight to interact with folks who get it right.

Tuesday, August 08, 2006

Google, MTV, and advertising

Given that YouTube has shown people are perfectly happy hosting content themselves, it will only be a matter of time before online video becomes ad supported the same way online text has. I am interested to see what Google's latest video alliances will bring.

New Apple mail application

His Steveness revealed Leopard yesterday at the MacWorld Developers Conference. I am sure it was just fine, and was intrigued by some of the features in the new Mail application.

One new feature is "Stationary" which lets normal people crud up their email with ugly graphics. Joy.

The other interesting feature is todos. I have no clue how well or poorly they are implemented here, but the iCal implementation is terrible, as was the last Outlook implementation I tried. Now-up-to-date has good todo functionality, but the best continues to be Gootodo, designed by Mark Hurst -- by boss and buddy.

Wednesday, August 02, 2006

Top Gear online

After being available only via torrents, the BBC's Top Gear has put its entire season online.

The question is: why? By hosting the media themselves, the BBC have to pay for bandwidth, site maintenance, etc. Why not just let others do that for them via torrents?