Sunday, September 28, 2003

The turning point?

When I first read pre-insanity Krugman's "Unequal Exchange" essay (1998) it made no sense to me. But I shrugged and moved on, because in general Krugman was so worth listening to. But his obsession with the wealth inequality between the top 0.01% and 1% has now reached such fevered heights it's best to ignore him and go back to his older, lucid pieces.

On re-reading "Unequal Exchanging" it seems that this may be where Krugman went off the deep end. He uses an analogy to model two economies and state a preference for redistribution based on those models:
[C]onsider this simple parable : there are two societies. In one everyone makes a living at some occupation-say, fishing-in which the amount people earn over the course of the year is fairly closely determined by their skill and effort. Incomes will not be equal in this society-some people are better at fishing than others, some people are willing to work harder than others, but the range of incomes will not be that wide. And there will be a sense that those who catch a lot of fish have earned their success.

In the other society, the main source of income is gold prospecting. A few find rich mother lodes and become wealthy. Others find smaller deposits, and many find themselves working very hard for very little reward. The result will be a very unequal distribution of income. Some of this will reflect effort and skill: those who are especially alert to signs of gold, or willing to put in longer hours prospecting, will on average do better than those who are not . But there will be many skilled, industrious prospectors who do not get rich and a few who become immensely so .

Surely the great majority of Americans , no matter how conservative, instinctively feel that a nation that resembles the second imaginary society is a worse place than one that resembles the first. It is also no question that our nation today is much less like the benign society of fishermen-and much more like the harsh society of prospectors-than it was a generation ago.
Krugman argues that income redistribution in a gold mining economy is "better" than in a fishing economy because the outcomes of gold prospecting are more based on luck than skill. While I sympathize with the sentiment, I would expect better logic from a first year econ undergrad. If you want people to take on risky behavior, you need to offer them greater rewards (or, if there is greater variance in expected outcomes, the expected outcome must be higher to elicit the same effort). In other words, the more risky the behavior is (or the more luck based the outcome is), the higher the rewards needs to be to compensate for the greater risk. This argues for less redistribution because taxing wins in a high-risk environment reduces effort by a far more dramatic amount than taxing wins in a low-risk environment. A winner-take-all economy, harsh as it may be, needs lower redistribution for efficiency, not higher.

The other point is that if economies go from "fishing" to "gold mining", it is because of technological change, not because of any policy. Whether or not you can fish depends on if you're close to the sea. The two economic arenas I can think of that might work like gold mining are fashion industries (driven by hits) and technology industries with large demand side positive externalities (software like Microsoft Windows). But large swathes of the economy do not remotely resemble either of these. I'm not sure if Krugman would want the government to intervene to direct technological development (once I would have assumed "no way" but now, who knows?) but throwing sand in the gears of where technology is taking us whether we like it or not seems, well, positively conservative.


Here are the results of Marc Brazeau's minimum wage challenge (my entry was here). I certainly found it to be very challenging, which is why I will not be participating in any future "challenges". I will point to pre-insanity Krugman and point out that anyone who declares "I'm not as superstitious about theoretical supply and demand charts. I want to know what's really going on" is being superstitious about all sorts of theoretical assumptions, but is keeping implicit and inconsistent.

Friday, September 26, 2003

Blame the Penguin

The Register likes to berate Wall Street for Sun's abysmal stock performance but they may want to look at Sun's financial performance as well. Or maybe not.

The point is that McNealy offering $100 desktop systems points to where the pain is really coming from. Every part of the technology stack--hardware, OS, middleware, application--wants to make every other part of the stack as cheap as possible so they can sell more kit. The $100 desktop hacks away at the middleware/application end of things saying that they are simply overpriced. It also declares Sun to be, firmly, a hardware company.

McNealy is also right in saying that companies want to buy systems, not widgets, which is why being an integrated systems company is better than being a widget company. It is certainly true that some customers want entire systems, but they want systems integrated from the business processes down, not the technology platform up. Perhaps parts of the integration business are scammy, but end users stand a better chance of being considered by consultants next door than system engineers half a world away. And I don't give them even odds with the consultants.

If you really want to find the source of Sun's woes look no further than Tux the Penguin. It's not as good as Solaris, or even NT, but it's good enough, and limiting Sun and Microsoft's ability to price is almost as damaging as costing them actual customers. It is also dramatically reduces the hardware and software development costs for Sun competitors like IBM because they can now optimize to a single OS. But the Register likes Linux, which is why it blames Wall Street. *Sigh*

Thursday, September 25, 2003

Better Notepad

Will Cox pointed me to Pepper as a reasonable BBEdit replacement for Windows. Hey -- it came from NeXT -- how bad could it be?

Also, it turns out that a buddy of mine recently bought (or tried to buy) something from IBM for his small business. He implied that the whole transaction involved some comedy, and if it did, I'll write it up.

Wednesday, September 24, 2003

Death to email?

David Gelernter, a Yale CS professor, has written this screed against email which, among other things, argues that nothing should ever be deleted and that we need societal rules (rules of custom, not software) that do things like let people acknowledge email in haste, but respond in leisure. He also talks about things like reminders, purposeful neglect, etc.

In some areas I am in complete agreement. It is a bad idea to silo your email information from everything else in your harddrive. Everything should exist together. On my Mac, I deal with this by copying and pasting plaintext from an email into a text file and naming it appropriately. On my PC I screw around with the appalling Notepad. Hate it! My kingdom for a PC equivalent of BBEdit.

I also agree that keeping things in sequential order is a good idea. It's hard to remember keywords, but it's easy to recall seeing something a week or so ago. Timestamps are an integral part of my filing system.

I do not agree that you should never delete anything. Given the huge amounts of data moving through our lives, our attention is very scarce and clearing some of the crap is the least you can do to try and make sense of anything. Just because we can store everything does not mean we should store everything. Lotus Notes, which I'm using again now btw, has the single worst storage system ever: delete something and it stays in your face. Then clear trash and it is destroyed FOREVER. This is one of the most poorly implemented software features every, and it remains intact through 6 versions. No other delete feature on the planet works in as asinine a way. *Sigh*.

Gelernter goes to the other extreme and says you should never delete anything. Also not a good idea. Being able to put things in a big trashcan to peruse later helps you recover from mistakes and keep your desk clear.

Anyway, Gelernter has some product he claims will make everything better -- you can check out the screenshots here. While it is certainly true that the picture with the young lady (and fellow) in bathing costumes looks a darn sight better than text, it's not clear that a more sensibly arranged detail display (I pity the person whose program manager actually looks like the one Gelernter has on the left) won't let you machete your way through more data more quickly.

Gelernter is also a shameless self publicist and knows how to play to reporters. It's worth skimming this interview just to see the man in action.

My old boss Mark Hurst has written all about email and bit literacy. Worth rereading.

Sunday, September 21, 2003

Regulated incumbents are protected incumbents

Vonage, which offers voice over IP services, now has to pay a regulatory recovery fee of $1.50. If you go through your phone bill, you will see a very large number of (individually small) line charges, that add up to about 35% of the total cost of service (at least for me). This is what Vonage's new fee comes from.

These line charges are for things the government requires phone companies to provide, things like universal access. This means that even if you live in the middle of nowhere, the phone company must provide you a land line and not charge you more than anyone else. It means that a middle of nowhere community has to get a phone box, which must be repaired no matter how much it is vandalized, and must be maintained no matter how much it is unused. They must provide these services even if it would be cheaper to give all these people a cell phone and a million minutes for free because, well, the rules say so. Besides, that will teach those evil telco monopolists for being so evil and reign in the terrible excesses of hyper Anglo-Saxon capitalism while creating a feeling of solidarity and togetherness. Subsidizing all those calls Texan ranchers make to their buddies in DC is just gravy.

Now Vonage comes along and offers cheap telephony over the Internet, clearly threatening the incumbent telcos. It's unclear how they will argue that voice telephony over the internet is illegal, but in the mean time they can lumber Vonage with things like universal access charges, even though they do not enjoy any sort of natural monopoly that I can see (the original rational for telecom regulation). In fairness to telcos, Vonage, like many telecom entrants before them, want to cherry pick the good customers and leave unprofitable Texans to the incumbents, a practice known as "cherry picking". This means incumbents need to raise their rates, hurting us non-Vonage non-ranchers, which may cause us to switch to Vonage. In this scenario, telcos are saddled with loser customers and will exit the business, which does not sound so bad to me but will upset AT&T shareholders who will (correctly) argue that the government just confiscated all their money.

So the solution is to tax Vonage and prevent a shift from traditional telephony to internet telephony. If this results in higher phone rates and hurts the economy as a whole, well, ranchers are people too. Even if they live in Texas.

Thursday, September 18, 2003

When might minimum wage make low-productivity workers better off?

Minimum wage, like rent control and protectionism, is one of those things that people are convinced will help the less fortunate, but in reality it makes the poorest of us even poorer. No one will hire someone for more than the value of their output, so a minimum wage of $5/hour prices out anyone whose output is lower than $5/hour. Given that $5/hour isn't much, such a person is not very productive and it seems perverse and cruel to deny them the opportunity to build the skills they need to become more productive, and command a higher wage. Arnold and Jane (via Marc) have covered the usual case where higher minimum wages hurt workers, as well as why this is separate from executive pay, so I'll deal with the outlier: when might minimum wage help low-productivity workers?

There's really only case in which this could be true. It's rare, but instructive, so here goes: If you have a monopsony, that is, a market where there is a single, monopoly buyer of labor, you will have an upward sloping supply curve where labor supplied is not very sensitive to price (wage). The monopsony employer might be selling into a perfectly competitive market, but it's the only employer (buyer of labor) in town.

In this case, the monopsony can maximize its profits by restricting the amount of labor it buys and lowering the price. This is the mirror image of the more common monopoly seller raising it's price, selling less, and maximizing profit. In a monopsonist market for labor, raising the price of labor through a minimum wage will increase employment and benefit the economy as a whole (the employer will be slightly worse off, but its loss will be smaller than the overall gain). In such a situation, a minimum wage will make both the employees, and society as a whole, better off. Note that if the market is not a monopsonist market for labor, a minimum wage will price low productivity labor out and hurt both them, and society as a whole.

The nice thing about this analysis is that it reduces an emotional "raising minimum wage will help/hurt labor" to an empirical test: how many similar employers are competing for labor in this market (or, how elastic is the supply of labor for a particular employer)? In a remote town with just one employer, labor supply will be pretty inelastic (there aren't many substitutes) so you could expect a minimum wage to be less harmful, maybe even beneficial, there. But in the areas we most closely associate with minimum wage labor: fast food, entry level retail, low level services, there are lots of similar jobs around so the labor supply should be pretty sensitive to price. It isn't hard for a fast food worked to flip burgers somewhere else.

So if anyone claims raising minimum wage will not hurt low productivity workers, ask them to demonstrate a monopsony market for labor. But don't expect to change any minds.

Wednesday, September 17, 2003

Esther Dyson Again

Esther Dyson, daughter of famed mathematician Freeman Dyson, spoke at Lotus this afternoon. I met Esther about a year ago and was not impressed, so today I was giving her a second chance. This time, again, I found her inarticulate and vapid, but at least she had the good sense to leave 75% of her time to Q&A. Q&A is good.

Here is a summation of what she said [my comments in square brackets]:
- ICANN was bad. ENUM may be better, if it works, which it may not.
- Network solutions is bad because they make money of accidental URLs. But they do good things with accidental URLs.
- Before a company can do anything, it should ask everyone's permission first. To make sure it's right. [Did she learn nothing from ICANN?]
- All governments are not equal, and the US government is better than most.
- Activists are crazy.
- Email will become "sender pays" to stop spam. [Clearly someone forgot to tell people who pay money to spam my regular paper mailbox.]
- The next big thing involves computers knowing what you want through context. [Microsoft has lots of this in XP and it stinks. This only works if you have people researching the context and then building narrower apps. But Esther puts her faith in software, not informed design. A mistake, I think.]
- The semantic web will be big. It's like a registry of concepts. [Clearly I don't know what's been going in in AI between 1998-2003, or Esther does not know what's been going on in AI since 1960-1998. Language is associative, not axiomatic. Computers are axiomatic. They will be bad at language. To Esther's credit, she did not push "natural language interfaces" as being a solution to anything, which it's not.]

Her latest issue of Release 1.0 is a) not written by her and b) talks about weblogs. *Yawn* Weblogs are so 2000.

More on micropayments

Stumbling Tongue has a good rejoinder to my comment on micropayments. Essentially, his argument is that we already use micropayments all the time and it's no big deal so long as it's convenient, ie. the inconvenience is less cognitive than Shirky likes to think. Very worth reading.

Tuesday, September 16, 2003

First week with IBM

After a week of training in Fairfax, VA, I'm now back at IBM offices in Cambridge MA, but this time with the services group, not Lotus software. It's really nice to get back to work.

I'm also reading "Who Says Elephants Can't Dance?" by Louis Gerstner, who goes over how he turned IBM from the verge of going bankrupt to the IT success it is today. It's filled with the usual stuff about passion and culture, which I don't buy so much, but also mentioned 1) how important the internet was to making IBM viable again and 2) how customers want technology that's close to their business processes and does not need to be integrated. I'd always thought of Apple as being the company brought back from the dead by the Internet, but I now see how it applies to IBM also.

IBM's big thing these days is "on-demand" computing, which seems to be a combination of outsourcing, ASP, webservices for integration and transaction, backed by new technology that makes running large data centers cheaper (grid computing, which federates data resources and autonomic computing, which automates the more routine administrative and maintenance functions). The basic deal is to take on a piece of a customer's data center, run it at lower unit cost, and split the savings.

Arnold Kling is skeptical about the value of grid computing, noting that there is no active market for old computers. (Although at this point there seems to be little active market for new computers too.) It's true that I'm now drinking the Kool-Aid, and I know nothing about what IBM's grid plans are, but I think its value to those who manage huge datacenters is pretty clear. And hopefully more datacenter management will be done by technology companies than by, well, everyone else.

Sunday, September 07, 2003

Shirky on Micropayments

Clay Shirky has another good piece on why micropayments will not work for online content. His basic argument (which he has made before) is that even if monetary transaction costs are low, micropayments impose significant mental transaction costs which make them not worth the bother. And even if micropayments dissuade only a few potential readers, since most artists can realistically hope for fame over fortune, they will be better off just offering their work for free.

Largely I agree with Shirky, although I wish he had also discussed ringtone sales in Europe and the Apple iTunes music store, both of which have had success with a micropayment scheme. This is why I support the RIAA's attempt to implement DRM schemes -- only when "big 5" content becomes hard to get will small artists be able to complete on price. This is also why I oppose mandatory licensing schemes, such as the broadband tax proposed in Canada, which prevent this competition from appearing.

I start work at IBM tomorrow and will be in Virginia all week for orientation. Blogging may be light.

Friday, September 05, 2003

US Macroeconomy

Hmmmm... thick southern accents on NPR -- it's gotta be a show supporting protectionism! And sure enough we have the harried housewife trying to Buy American, the crusty factory owner arguing that manufacturing is the lifeblood of the country, and the eager young academic tying trade to inequality to the environment to labour rights to international human rights. Not an economist in sight. *Sigh*

Two good pieces on Slashdot

So, Universal Music has just reduced album prices to $13, and some guy is trying to sell a song he bought on iTunes on eBay.

I have no idea how much of an effect Universal's move will have because I can't remember ever paying much more than $13 for a CD (I always had a coupon, or there was some sale going on). But if a move in suggested retail price from $18 to $13 means actual retail prices go from $13 t $8, then it might become competitive with free, but less convenient, online services. But from their press release, this move in price only applies to historic records, so the new hits will be as expensive as ever. But I do not know whether people download new hit singles most or older "archive" material, and I also do not know which is more profitable to the labels, I just know that they don't sell older material much. And what is most significant about the price cut to me is how it breaks the monolithic uniform pricing we see for music, which I suspect came from tacit collusion within the industry. I'm sure the remaining big 4 are watching Universal's experiment carefully.

The guy selling the song on eBay is stirring up unpleasant legal issues that we (I) would much rather just elide. I am not sure what Apple's deal with the labels included, but I bet they promised that people would not take songs off the iTunes store and put them up on P2P networks because of how the iTunes experience was structured, not because of DRM. Offering the music on eBay seems to put that deal into question and makes a case for DRM which, as Apple understands, would kill the experience iTunes offers.

Wednesday, September 03, 2003

Are computer games too hard?

Xbox director Laura Fryer says that console games have a smaller audience than film or TV because they are too hard. I think they have a smaller audience because the market for passive leisure experience is larger than that for active leisure experience, and that console games are a poor substitute for other forms of active leisure (such as going out with friends).

But I am very sympathetic to the argument that games have become too hard. I used to think that I was pretty average at video games, but I now find the controls too difficult and am significantly below average. I liked "Doom" but never got the hang of aiming with the mouse and moving with keys and so could never play anything else. I also find the current console controls, with 2 joysticks, a pad, and over 8 buttons to push, very hard to operate. Which is why, if I ever get a game system, it will be a humble Gameboy with the first game being Wario Ware, a game so simple and fun that the gaming industry calls it weird.

Game designer Greg Costikyan, who I once had privilege of meeting in NYC, points out that, although some games have been "captured" by fanatic gamers and made impenetrable to normal people, the most popular games are still the simplest: Minefield, Solitaire, Snood, etc. I had a similar conversation with Impression Studio's lead developer Zeb Cook recently who agrees with Greg about keeping games simple and accessible, but does not feel that games necessarily need to be experimental or innovative, they must be fun.

The reason this is interesting to me is that the open source community counts its "scratch your own itch" production technique as a great strength, and it is when developers are like users, but this falls to pieces when they are not. I'd wager that this is the more common scenario and is where the software business will make its money going forward (operating systems don't fall into this category, btw).

It's also amusing to consider the type of feedback hard core customers give, certainly in games, when asked by developers. A friend of mine went to a DARPA sponsored wargaming convention in New York hoping to model terror attacks, but found the participants more focused on coming up with rules to cover "suppression fire". Another friend of mine, a hardcore gamer, helped playtest a role playing game involving wizards but complained that the spellcasting rules and component requirements were not "realistic" enough. It seems that when people say "realistic", they don't mean closeness to the actual, real world that we all live in.

Tuesday, September 02, 2003

UI and project management

Winterspeak began, in part, with me trying to figure how to make software more useful to end-customers, and one of the areas I explored was project scoping and upfront user testing. David Anderson was one of the people I chatted with while working on this, and he has a blog up supporting his new book on agile management. Nice.