Wednesday, March 31, 2004

"One small step for robots..." and the deadweight cost of taxation

This month's New York Review of Books features a great article by physicist Steven Weinberg that argues NASA is wasting its money on manned missions and should try something else. He points out that our manned space program, though justified on military and scientific grounds, does remarkably little science and produces no known military benefit at ludicrous expense. (Most of what astronauts do in space could be done by robots for a small fraction of the cost.) The "manned mission to mars" folks should take note. But towards the end of the article, Weinberg veers off into economics and makes a wrong turn, writing:

"My training is in physics, so I hesitate to make pronouncements about economics; but it seems obvious to me that for the government to spend a dollar on public goods affects total economic activity and employment in just about the same way as for government to cut taxes by a dollar that will then be spent on private goods. The chief difference is in the kind of goods produced by the economy—public or private." [emphasis added]

This is a common intuition, but it's not true. One thing being ignored is a factor known to economists as the "deadweight cost of taxation". The essence of the problem is that taxes cost the taxpayers significantly more than the value collected, and therefore even quite well-directed spending can produce a net loss for the economy as compared with simply lowering taxes.

Consider a simple sales transaction. I want to buy an apple and am willing to pay as much as $1.10 for it. (By "am willing to pay" I mean that I would be indifferent between having $1.10 and having an apple, but at any price less than that I'd feel like it was worth making the trade.) You have apples to sell and are willing to sell them for as little as $1. We meet in the marketplace and after some negotiation I agree to purchase the apple for $1.05. Since I was willing to pay a nickel more than that, I am richer by a nickel for having made the trade; the "consumer surplus" for this trade is 5 cents. Since you were willing to accept a nickel less than the agreed amount, you also are richer by a nickel for having made the trade; the "producer surplus" for this trade is 5 cents.

The economy as a whole is better off for this trade having occurred; resources (an apple) have moved to where they were more highly valued. The economy as a whole is "richer" by the ten cents worth of value that came from this trade, and the immediate benefit that flows from this is distributed as: 5 cents to you, 5 cents to me.

That was without any taxes.

Now let's add taxes to the picture. Let us imagine apple-sellers are required to pay a 5 percent sales tax. You and I still meet in the market, and you sell me the apple for $1.02 plus tax = $1.07 total. We still made the trade and the economy still benefited by ten cents from the transaction, but the breakdown on where the benefit went is: 3 cents to me, 2 cents to you, and 5 cents to the government. Assuming the government spends the money as well as you or I would, it's still consistent with Weinberg's intuition. Compared to the no-tax scenario I am 2 cents poorer and you are 3 cents poorer than we would have been, but the government is 5 cents richer so there's no value destroyed. So far, so good.

But suppose taxes are increased such that apple-sellers now have to pay a 15% sales tax. What happens?

With any tax over 10% there is no possible price for which the seller and buyer both benefit from this trade. Therefore the trade doesn't happen. I keep my money and you keep your apple because - given taxes - you can't afford to legally sell the apple at a price that I am willing to pay. So the economy fails to achieve the 10 cent efficiency improvement from the apple going to the person who most values it.

Compared to the situation where there were no taxes, I am 5 cents poorer, you are 5 cents poorer, and there is no offsetting benefit. Since the trade doesn't occur, we don't even pay taxes on it, so there's no gain in taxes to offset this ten cent loss. That is a deadweight loss. There are other losses as well. Costs from people hiring tax attorneys and restructuring their affairs to avoid taxes. Costs from trying and imprisoning tax offenders. Costs related to administering the tax system. Costs of opportunities foregone, businesses bankrupted, sales lost, inventions proven uneconomic due to an unfavorable tax climate rather than any inherent flaw. Even costs spent lobbying to benefit from tax changes (aka "rent seeking").

The higher taxes are, the worse are these costs. So yes, it really is better for the government to cut taxes in most cases. Because they'd have to spend $1.40 to do as much good with spending as cutting taxes by a dollar does.

["It has been estimated that the dead-weight costs of the federal government’s raising an additional dollar equal 39 cents."]

Sunday, March 28, 2004

Why bet more on a positive count?

A positive count means more high cards than low cards remain in the deck to be played. A higher than normal chance of high cards coming out means:

The dealer is more likely to bust. He has to hit until 17, and it's harder to hit that narrow range between 17 and 21 when high cards predominate.

You're more likely to make your double-downs. When you double down on a 10 or 11, you're hoping to see a high card come out to make your hand 20 or 21.

More blackjacks. Even though half the blackjacks will go to the dealer, the "blackjack bonus" means you win 1.5 times your bet when you get blackjack, whereas you only lose 1x your bet when the dealer gets blackjack.

For all these reasons, you're more likely to win when the count is very positive, so you should bet more then. You will probably still lose the majority of the hands you play, but betting more on the winning hands helps you come out ahead.

Thursday, March 25, 2004

Good advice: Don't talk to cops.

How Do You Count Cards?

Here's a basic plus-minus count.

Keep a number in your head. That number is called the running count, and it starts at zero whenever the deck is shuffled.

As cards are played, every time you see a card with a value of 2, 3, 4, 5 or 6, add one to the count.

Every time you see a card with a value of 10, J, Q, K, or A, subtract one from the count.

When the count gets larger ("a positive count"), the odds have swung a bit in your favor.

When the count gets smaller ("a negative count", a count that's less than zero), the odds have swung a bit in the casino's favor.

Take the running count and divide it by the number of decks remaining to be played to get the true count. This is what you base your play and betting decisions on. Here's a sample betting strategy: "If the true count (TC) is less than 2, bet one unit. If TC is 2 or greater, bet (TC-1) units."

You will also vary your play based on the count. For instance, if the count is positive you will stand on a 16 versus a dealer 10 showing (or on a 12 versus 4 showing); if it's negative, you'll hit those two hands. If the count is +3 or above you'll take insurance, whereas basic strategy says you normally wouldn't take insurance. There are many index-based strategy variations you will want to learn; 16v10 and insurance are only the biggest two.

The bottom line here is that you're not having to keep track of very much as the cards go by. The index numbers, basic strategy, and your betting strategy are things you memorize well in advance of sitting down to play. As you play, you only really need to keep track of a single number - the Running Count. It goes up, it goes down, and sometimes you convert it to a true count and make decisions based on it.

This is why playing to a 6-deck shoe isn't hugely different than playing single deck. The same skills are used. The strategies employed are different, of course, but the essence of the game is still the same.

Monday, March 22, 2004

Double-deck S17 DAS Basic Strategy - now in color!

(there seems to be something wrong with the way my style sheet is displaying tables. Let's try displaying the same info another way, see how it looks...)

This chart was generated by BJC Chart Maker.

2 Deck : S17 Stay : DOA : DAS : no Surrender: Hole Card Normal




Blackjack strategy 2D DAS S17

For a double-deck game where the dealer stands on a soft 17(S17) and Double After Split(DAS) is allowed, you should:

Hit untilvs

Sunday, March 21, 2004

Don't you have to be a Rain Man to count cards?

Short answer: no.

You're not trying to remember every card during play. You do have to memorize a basic strategy table and some variation index numbers, but you can do that long before sitting down at the table. The mathematical skills used to count cards during a game aren't at all difficult. They involve addition, subtraction, and a little bit of division or multiplication - the sort of math most people learn in grade school.

Playing blackjack well enough to have a small edge over the house is actually pretty easy. A "Rain Man" could learn it, but so can you.

Friday, March 19, 2004

Don't they use six decks now so nobody can count?

Short answer: no.

There are many casinos in Nevada that still offer single deck and/or double deck games. If that is your game of choice, you can play it.

More importantly, the skills required to count cards don't depend on how many decks are used. Adding decks does slightly reduce the player's edge over the house, but it doesn't make counting significantly harder. There are some other factors that make blackjack more difficult to win at today than it was 20 or 30 years ago, but number of decks in play is not the crucial element.

The movie Rain Man contributed to the spread of this myth with a scene in which a pit boss incorrectly states "nobody can count down a 6-deck shoe".

Simplified Basic Strategy

I don't have time to learn full basic strategy. Is there something simpler that will minimize my losses?

Here's a set of rules so short and simple you can write it on a cocktail napkin and take it to the table with you.
  • Always split aces and eights.
  • Double down on 10 or 11 unless the dealer shows an ace.
  • Hit until you have 12 if the dealer is showing 2-6.
  • Hit until you have 17 if the dealer is showing a 7 or better.
  • Never take insurance.

It's no substitute for learning full basic strategy, but it's a good start.

Basic strategy

How do you play blackjack as an advantage game?

First, memorize the basic strategy for your preferred game. For every hand you might have versus every upcard the dealer might be showing, there is a single mathematically correct play. You should either hit, or stand, or split, or double down. There are many options, but only one is correct. This is not a matter of opinion or following a hunch. It doesn't matter whether you think the dealer is "lucky" or is "running a streak". It doesn't matter whether the other players are "mixing up the order of the cards" by coming in and out or playing badly or switching how many hands they play. All that matters is the probabilities. There is a right strategy for your game, and the first order of business is to learn that strategy.

Is playing perfect basic strategy good enough to have an edge?

No, but it reduces the house edge to a very small percentage. If you play perfect basic strategy AND learn to count cards and vary your play and betting according to the count, THEN you will have an edge. But basic strategy is the foundation on which a winning blackjack strategy is built. Don't try to learn counting cards until AFTER you have learned basic strategy. Find yourself a book, a training program, or a website that has the info you need. Print up some flashcards. Then study basic and use it. Practice dealing out the hands and writing out the rules until knowing basic strategy and following it is second nature for you.

Wednesday, March 17, 2004

Who Am I?

That's the name of a Jackie Chan movie I saw on cable this weekend (Side note: don't you hate it when they show ads during movie credits, taking over the audio track and squishing the picture off to the side so you can barely see it? I wouldn't complain for most movies, but with a Jackie Chan movie the credits are often the best part of the film!)

So who am I, that has chosen to write this blog?

I am a blackjack player. A libertarian anarcho-capitalist. A software/QA engineer. A musician. That guy who wrote the Taebo FAQ, various libertarian tracts and some random bits of well-indexed unicycling advice.

Let's take that first item for now, and leave the rest for some other time. I play blackjack at a semi-professional level. I could probably make a living at it if necessary, but currently it's a hobby project, a side interest. It's a way to see nice shows, eat at nice restaurants, stay in nice hotel rooms and maybe earn a little extra spending money now and then. I have read dozens of blackjack books; I subscribe to a blackjack newsletter, and I have discussions on a private blackjack web forum. Some years back, I was involved with a blackjack team like this one. When I'm playing, I sometimes have stories to tell that might be of interest to a rarified audience. I could post trip reports, book reviews, strategy ideas, that sort of thing. Some of this would be work-in-progress material, a way to bounce new ideas off of my audience and strengthen them before I go risk actual money, time, or status.

Hence the name Blogjack.

(Sadly, somebody else has registered But happily for me, they don't seem to be doing anything with it yet. Should I snag the .org or .net?)


Blogging was light last weekend because I was in Seattle taking my CSTE exam. What's that, you ask?
"Acquiring the designation of Certified Software Tester (CSTE) indicates a professional level of competence in the principles and practices of quality control in the IT profession."

It's a test I took so my employer can tell customers we have "certified" QA people on the job. So how was the test? Gruelling. I do have a minor nit to pick. The test is supposed to measure the sort of knowledge you need to do the job, and mostly that's what it does, but it also has a few test questions that cover random bits of useless and/or dubious QA lore. There were a few questions like this:

Dr. Gregory Smith's "Five steps to continuous improvement" do not include which of the following steps:
(a) Do
(b) Act
(c) Measure
(d) Refine
(e) None of the above

If your initial response to that question is "Dr WHO?", does that mean you're not qualified to do QA?

Another one: Requirements errors are more expensive to rectify later than are production errors. But how MUCH more expensive are they? 10x? 50x? 100x? Well, it probably depends on who did the measuring and what they were building and how they were building it. There are some answers in the literature, but not any I'd put a high degree of confidence in. Another question along those lines is: How many more bugs are produced in one stage than another? The relevant-to-doing-your-job answer to both questions is: important bugs can creep in at all levels, and it's pretty darned important to make sure you're building the right house at the right address on the right foundation before you get around to installing the plumbing and painting the walls. But if it turns out you built the wrong house, do we consider that ONE bug or a thousand? It's kind of arbitrary, but "it's kind of arbitrary" doesn't make a good multiple-choice test answer, and doesn't help you pick from a set of highly precise suggested answers such as 10, 50, 100.

So there are a few questions on which I had to make wild guesses as to what answer they wanted, and I know I got some of them wrong. But I think they allow enough of a margin of error that I still passed. If I didn't, well, next time I'll pass for certain.

UPDATE: I passed!

New google feature - local searches lets you do distance-based searches. For instance, enter your address with the search term "clam chowder" to find discussion of nearby fish restaurants. Very cool.

Tuesday, March 16, 2004

Here's a cool optical illusion.

Wednesday, March 10, 2004


I've been experimenting with the anti-sleepiness drug Provigil for the last few months, long enough to have reached some conclusions. [Caveat: I'm a Diet Coke fanatic. I drink it all day long; I have at least a 6-can-a-day habit. I feel groggy on days when I drink less than that. So there's always a pretty high background level of caffeine in my system.]

The great news about Provigil is: it really, really works for the purpose it was designed for. If I happen to stay up very late some night and I need to be awake and alert the next morning, taking a Provigil in the morning gives me that ability. I can be wide awake and productive at work on just a few hours of sleep, when I need to.

However - contrary to the hype - it does not make sleep optional or all-nighters particularly easy and productive. At least it doesn't seem to do so for caffeine-addled me. I tend to find I am unusually tired in the evening on days when I had to take Provigil in the morning. It feels like it resets my sleep clock. Like changing time zones, it adjusts my metabolism so "wide awake" comes earlier than it otherwise would, but "tired and ready to sleep" is likewise brought in a few hours ahead of schedule. And it doesn't entirely remove the "sleep deficit" effect although it does reduce it a bit; my body wants to sleep a bit more to make up for sleeping less earlier.

Also, the Provigil doesn't help me get out of bed. Once I manage to drag myself up, take a shower, down a pill, I'll feel awake. But the dragging-myself-out-of-bed part is still pretty hard, and the less sleep I've had the harder it is.

I take a single 200 mg pill in the morning on days when I find I've had less than 6 hours of sleep and really need to be alert at work. On days that don't meet those criteria I don't take the pill, and I don't miss it. I never feel jittery or have difficulty sleeping due to it, I merely feel awake at the times I need to.

I use the drug a couple times a week. My initial prescription of 30 pills is getting low; I plan to get it refilled. It is very nice to have this tool at my disposal. Like aspirin, imodium, and throat lozenges, it's just something I want to have handy in the medicine cabinet or by the bedside. Not to take all the time, but to use when in need.

Provigil: It does a body good. :-)

How to balance the federal budget

This looks like a good starting point, but if I had my druthers I'd take a meat axe to Defense and Homeland Security too. There's no reason we need to have troops stationed in over a hundred countries; we should cut foreign aid and bring "our boys" back home from everywhere they aren't actively needed to repel a clear and present danger to the nation, which is to say, just about everywhere.

Thanks to MarginalRevolution for the link.

Tuesday, March 09, 2004

I've got happy feet!

My tap dancing shoes arrived in the mail today. I must be near a personal record for current unfinished projects. Desperately waiting on my attention are:

  • Fixing GetFit! for PalmOS, especially the expiration code, and posting the updated version everywhere so I don't have to keep emailing purchasers by hand with a "here's how you work around the problem" note...
  • Finishing my pet drivers license store so people can actually order stuff from it...
  • Making a decent GarageBand recording of my latest song, the one that actually shows promise, while certain opportunities are still available for it...
  • studying for the CSTE exam that I'll be taking in less than a week...
  • many, many other projects...

And what did I work on tonight? Learning to tap-dance. And fiddling with my new blog. Bah.

Why do colleges expect money from alums?

Got a call from a telemarketer for the Berkeley College of Engineering. As usual, they'd like to "get feedback from the alums to benefit from their wisdom and experience" and, oh, while they're at it, give us an opportunity to donate. While attempting to gracefully extricate myself from the conversation, it occurred to me that I might have an unusual attitude towards my alma mater.

Which is: we had a straightforward business relationship. I paid them my money in exchange for them providing me with educational services.

When I buy a book, does the bookstore call me up later and ask for donations to subsidize future book sales? When I go out for a meal, does the restaurant call me up later and ask for donations to subsidize future meals? So why is a college different? What makes them expect that I will feel obligated to give them extra money above and beyond the amount I paid them voluntarily in tuition and the amount I continue to pay them involuntarily in taxes?