Saturday, November 30, 2002

Nice article from Cringely

Cringley documents how P2P filetrading will effect some media more than others. He thinks text will be the most effected, then music, and that movies will not change much. I agree -- consumers have already backward integrated into most text-based content--the web and my inbox are much more interesting to me than the largest library or book store. Mashups are the area of music where consumers have backward integrated into production (through becoming DJs), and it's hard to see what content owners can do about that. There isn't a good analog for film.

Wednesday, November 27, 2002


Generally, I'm skeptical about "converged" devices, but I think the mobile phone + camera combo is great. Not because I think such a gizmo would replace my Canon digital elph, but because sharing inanities with your friends is the killer telecommunications app. I'm looking forward to when Sprint offers cheap, integrated phone/cameras so I can replace my semi-functional Sanyo phone.

Happy Thanksgiving to US readers!

Tuesday, November 26, 2002

Wrapping up exec compensation

Many people have written in about executive compensation, insisting that CEOs are paid way too much and that CEO pay should be even more closely tied to performance. I point out that since CEOs (like almost all of us) are risk averse, tying their pay more closely to performance will result in even higher pay on average since they will have to be compensated for the extra risk.

For example, I would take a $100,000/year job over a job which would pay me $200,000 or $0 with equal probability. Force people to take on even more risk, they're going to hold out for a higher salary. So you can hold out for lower pay, or better incentives, but not both. As a shareholder, I vote for better incentives.

More on intergenerational transfer of wealth

I asked U Chicago labor economist Robert Topel about the recent (upward) revision to intergenerational transfer of wealth. He said that when Becker first thought about the problem he calculated it to be around 20%, which is actually surprisingly small. He argued that productivity clearly has inherited components to it, whether it be nature (intelligence genes) or nurture (family values include healthy work habits), and so you would expect earning potential to be as inheritable as other inheritable traits. The revised figure is closer to this. Moreover, you would expect this to be stable across income brackets and time because heritable traits are, well, heritable.

He also pointed out that intergenerational transfer of wealth does not answer the question people think it does, namely, "how mobile is the society?" The true test of that is how quickly does individual wealth react to some external shock. For example, does a prodigy with an IQ of 120 (I know, I know, IQ is a terrible measure of productivity, work with me here) born to very poor parents end up very poor? Or does a particularly dumb, lazy, obdurate, and feckless scion of the rich end up equally rich? I don't know the answer to that, but my anecdotal experience says "no." The US may be a lot of things, but it's certainly not a class based society.

I guess my point is that an immobile, class based society of privilege and an efficient, meritocratic society (where heritable traits influence productivity) can look exactly the same from an intergenerational transfer of wealth perspective.

Customer experience and DRM

I'm organizing the DRM panel at U Chicago GSB's upcoming High Tech Confernece. I inviited Don Norman to come and speak about user experience and DRM, but he declined saying he didn't know much about the subject. A pity. As Arnold Kling points out, Convenience is King, and customer pain adds to price as much as dollars and cents. One of the reasons that Napster never made unknown, garage bands famous was because there was no difference in price between a popular song and an obscure song, so the popular song always wins. I think that Arnold is right though, and there is so much legal involvement in these services that they cannot possibly serve actual customers.

Palladium, for all it's evils, promises to improve the experience of using DRM content. Since using non-DRM content on Microsoft systems is currently so horrible, I'm skeptical about how good a "good" DRM experience can me.

Sunday, November 24, 2002

Microsoft's great innovation

Turns out that Microsoft is bundling a new application into Office called OneNote. It's integrated software designed for taking notes. Funny -- I call this a text editor, and have been using it as such for years. Sadly, this is also a great innovation for Microsoft.

Friday, November 22, 2002

Breakfast with Epstein

I (and a bunch of other folks) had breakfast with famed U Chicago Law prof. Richard Epstein today, discussing corporate governance. Epstein was dismissive of the new corporate governance laws coming out of Congress, predicting that they will merely raise the cost of doing business, make it harder to build boards, and reduce the likelihood that a private company will choose to go public (and, by extension, that private companies will be funded at all). He argues that these laws increase the risk of being on a board (since investigating potential trouble carries legal liability) and the required commensurate compensation will have to come out of shareholders pockets. Shareholders paying for better corporate governance sounds OK by me, but given the immense responsibility of independent directors, I'm skeptical of how independent they can remain. Moreover, Epstein discussed how board members are chosen for prestige, not competence, which will exacerbate the problems caused by these new requirements.

I'm not sure what the best way to tackle corporate governance problems are. Epstein left it at "enforce existing laws better".

More exec compensation

MH wrote in pointing out that simply subtracting CEO compensation ($6B) from market cap ($9.1T) is wrong -- one is a single year expense and the other is the cumulation of all future performance. Arnold Kling corrected my calculation by treating the $6B expense as a perpetuity, multiplying it by 25x (or 20x at a 5% discount rate) giving you a grand total of...$150B. More to be sure, but just 1.6% of total market cap.

Wednesday, November 20, 2002

Executive compensation

Although it has nothing to do with technology, I thought I'd post some numbers on executive compensation -- a hot topic these days. Executive compensation has risen between 1950 and now (although it was comparably high in the 30s when Ford was running Ford). According to Forbes, average CEO pay was $4M in 1996 and $7.5M in 2002 (it was 11% higher in 2001).

But to put this in perspective, total CEO pay in 1991 was $6B out of an organizational market capitalization of $9.1 trillian. That means that if all the CEOs decided to work for free forever, it would add just 0.006, or 60 points, to the Dow. To put it another way, if they were to work for free and all the money went to employees, the average employee salary would go up $54.

So, to say high executive pay is robbing shareholders or employees is wrong -- the transfer effects here are trivial.

Instead, let's look at how CEO pay effects their incentives. CEO's raise a classic agency problem, because a CEOs job is to manage the firm for the benefit of shareholders, not to indulge in ivory backscratchers. The thinking is that tying CEO compensation to firm performance reduces this agency problem and gets him to act in the shareholders' best interest. Moreover, since CEOs are risk averse and shareholderes are risk neutral, compensating them in options (which reward risk) aligns incentives even better. Most CEO compensation in the US (and to a lesser extent in the UK) comes from performance pay, of which options are a common component.

The best options isolate company performance from broader market performance. After all, why pay a guy if he does a lousy job but the market as a whole goes up, and why penalize him if he does everything right but the market tanks? If you look at how much CEO wealth depends on company performance vs. market performance, it turns out that things work out roughly correctly -- CEO wealth is more closely tied to how his firm does. But, compensation is more closely tied to broad measures of market performance (like the that S&P 500) than narrow measures (such as close competitors) -- which is the opposite to a good compensation scheme.

So, if you want to complain about CEO pay, it is not right to argue that shareholders or employees are getting a bad deal -- the effect on them is trivial. It is correct to complain that pliant boards don't index options to close competitors (which reveals a whole new agency problem between boards and shareholders). And it is right to say that stronger incentives also call for greater monitoring, as CEOs with lots of options have more reason to cook the books.

Tuesday, November 19, 2002

Pentium v. P4

A PC with a new 3 Ghz Pentium III is faster at video editing (and probably lots of other things) than a Mac with a 1.25 Ghz P4. While this may matter for all the video editors who live between QuickTime Pro and a clustered Lintel blade farm, most home consumers stopped caring how fast their computers were about 2 Ghz ago. This is because most important applications are now being written to the Internet Operating system, not Win32 API, nor MacOSX API, or anything else. And with Internet Operating system applications, what matters is not local processing ability, but bandwidth.

Thursday, November 14, 2002

Becker and social capital

MR writes in pointing to this article in the New York Times (reg. required) arguing that parental success is (much) more important to a child's success than previously thought. Combine this with the Krugman's recent "the rich are too rich" op-ed, and it seems that the fabled "American dream"is a hoax.

In particular, MR writes
What I don't understand is why their is no mention of social privilege as a major factor (see excerpt below). In my experience, children of the rich have much greater access to capital as well as a greater chance to interact with influential business people because of their "connections," ie. because their family is part of the social elite.

Children of poor people trying to get a bank loan are like "nobodys" trying to get into a fashionable Hollywood party.
While children of the privileged have many advantages over children of the poor, access to capital is not one of them. If anything, capital markets are *much* more democratic now than they ever have been in the past. Innovations like junk bonds, angel investing, warrants, and venture capital means that money cares less about privilege than it ever did. In home ownership, a more prosaic form of capital, low interest loans and zero-downpayment mortgages mean that more low income individuals have access to home ownership than ever before.

If you look at return to capital over the past 30 years, it's stayed pretty constant at around 8%. This is what one expects of capital markets are competitive -- investors seeking higher returns drive down the value of high-return investments. The returns to human capital however have skyrocketed, particularly to education, which is also reasonable as productivity improvements made by investment in capital ends up going to labor through higher wages (returns to capital are constant).

But the article argues that education only explains 60% of the data. This means they've done a regression, and the coefficient infront of the "education" variable was 60%. This type of analysis is pretty inaccurate, so I'd be wary of trusting it. I don't know why intergenerational correlation in income seems so high, but it's not because only the rich have access to capital. Capital has, without a doubt, become more democratic than it ever was before.

Siebel v the world

I spoke with a couple of Siebel guys the other days about who their big competitor was. They are mostly worried about ERP vendors bundling CRM into their product, so see SAP, Oracle, and PeopleSoft as their big competitors. I argued it was Microsoft, since they will make easy applications that will be first class citizens on .NET and squeeze Siebel out, starting at the midmarket and moving upwards. While Siebel felt that .NET would become the enterprise standard, they did not think Microsoft was a threat because it only went after little businesses.

Outlook is used by everyone from individuals at home to people in big companies, even though it does just a fraction of what Lotus Notes does. Even people who value the extra stuff Notes can do are not willing to pay much for it, because they take Outlook as their reference product. Even if Siebel only targets big companies, there will be potential buyers on the margin who will compare Siebel's expensive application to Microsoft's cheap application and feel that, even if Siebel does more, is it really worth that extra price? A low end competitor reduces the high end incumbent's ability to price, even if the high end product is better.

Microsoft enters Enterprise IM

The Register has a piece on Microsoft entering the enterprise IM space next year. Oddly enough, the article mentions AOL and Yahoo! as potential competitors but does not note that Lotus is currently the dominant vendor in the market with SameTime. That says something about Lotus's visibility these days. I used the system when I was there and (to my surprise) really liked it. It turns out that the killer feature is "presence", the warm fuzzy feeling you get when the little green dot lets you know your friend is online. Who knew?

Wednesday, November 13, 2002

Chimera rocks

I've been using Opera 5 in Classic on my iBook because no OS X.x browser came close to its speed and responsiveness. But, being a classic app, it used to get bogged down under after a few days of use. Chimera 0.6 is fantastic though, it even renders blogger pro correctly. Nice.

Tuesday, November 12, 2002

Movielink is finally up

So the Hollywood consortium has finally put up an online movie site. It's terms: $2.99 - $4.99 for 24 hours worth of unlimited access within a 30 day period. That seems comparable to renting, but it does not include hassle costs -- the 17 minute estimated download time is way too optimistic. And I have no idea what the viewing experience is like. I don't like watching movies off my computer and I don't think that's going to change. I expect the quality to be poor.

Movielink is engaged in a law suit for collusion against independent VoD services. There is no economic benefit to the studios owning content and pipes, but it makes it easier for them to keep an eye on each other's pricing schemes. I don't know...I think VoD was solved by TiVo and Netflix.

Friday, November 08, 2002

Hollings out

After the midterm elections, Hollywood Handpuppet Fritz Hollings (Dem South Carolina) is out of the Senate Commerce Committee. Den Beste thinks this will make the world safe for PCs again, at least for two years. I hope he's right.

Wednesday, November 06, 2002

Murphy on Microsoft

Chicago economist Kevin Murphy commented briefly on CKK Microsoft ruling this morning. His take was that the Microsoft case was a hard one to decide, and remedies in particular were very difficult to work out to maximize social good. He also mentioned how the politicization of the Attorney General's office, how it's transformed into a "learn to be Governer" role, means that Attorney Generals are more interested in staying in the limelight that enforcing laws and defending the public interest, and this became an issue in DOJ v MSFT.

People who applaud the nine dissenting States who are considering appealing the CKK decision should bear in mind why they are doing it. Certainly, their alternative settlement proposal was ludicrously not in the public interest. This isn't to say that I think the current settlement is any good--I think it's terrible--but I don't view the politicization of the Attorney General's office as a good thing either.

"" update

A reader sent in an link unravelling the mystery humour site a friend pointed me to earlier.


I just deleted a number of trial posts to this weblog I made using Archipelago a client-side weblog editor. Right now, the only thing I use Internet Explorer for is writing to, and I would like to no longer need to do that.

What was interesting was how my initial stupification with Archipelago matched by initial experience with Radio. Getting everything set up was confusing, and I had no idea what any of the menu commands did. I don't mean to pick on either program's authors-- creating good software is hard, especially in a new design area where there aren't many norms.

Both Archipelago and Radio cross the divide between client and server. This is similar to the Lotus software I worked with at IBM over the summer. Simple menu commands like "save" and "new" are confusing in this new context. Is "save"ing something on my desktop the same as "post"ing something on Blogger? Or is the same as "publish"ing something on Blogger? Or is it the same as "save"ing something in Word, with a seperate "publish" coming later to get it all on the site. And given the agony of synchronization (a daily struggle at Lotus) how can you seperate/integrate data between client/server so the user doesn't go home and freak out because all his email has vanished (easy to do on IMAP if you haven't set things up correctly).

People moan about web interfaces all the time, but web apps are often *much* easier to use than their desktop counterparts. It's easy to forget how bad things were before, just as it's easy to forget how good Microsoft software is compared to what it has replaced (check out SmartSuite if you don't beleive me).

I haven't thought this through, but here are my rough and ready rules-of-thumb for these client/server type apps:
1) What people want is essentially the web functionality, but with the fast client-side responsiveness and free of the necessity of having to be online.
2) This does not mean you need to look like a webpage. Focus on the *simple* functionality of the webpage and deliver that in the best way possible.
3) Don't worry about seamless synchronization. If snychronization makes the easy, core stuff harder, drop it.
4) Focus on delivering the web functionality, but faster and easier.

The poster children I see for this is 1) POP email, 2) Watson, and 3) Napster.

With POP email you write your message, connect, and send the message. There is no complication with editting your email once it's sent. Your email lives on your client or on the server -- getting some setting wrong does not mean everything vanishes the minute you disconnect from the Internet. Now I know all the drawbacks of POP, but it delivers a simple experience when bridging the gap between client and server.

Watson simply brings key web application functionality (eBay, Amazon) to a desktop app. Yes, you need to be online to use it, but that's OK. It takes a website and makes it simpler.

Napster did the networked drive this really well. Pick a folder on your hard disk and call it the network drive. Drag something to that, and it's shared. Delete it, and it's gone. If you copy it to another local drive, it's now in two places. If the two go out of synch, they go out of synch. This model of network drives (with lots of gunk) is essentially what Ray Ozzie is trying to do with Groove now that he's done with he synch-happy Lotus Notes. Other network sharing systems require you to go to some website and upload stuff. This is like forcing the user to *save things twice*. Predictably, the user refuses to do this, keeps stuff up to date on his local drive, and then emails it to everyone once it's "kind of" done. The whole notion of "place" does not work because networked places are harder to get to than local places if you need to access a remote server through a web interface.

My ideal desktop weblog editor would be like POP email. I compose the post offline, see how it looks "online" offline, edit, and then post whenever I connect. If I want to edit it, I fire up I.E. and do that through the Blogger UI.

Saturday, November 02, 2002

Random stuff

I've been having some frustrating economics-related discussions with some of my friends, both from U Chicago and from before. I take this frustration on my part as a bad sign, it means that my own thinking isn't clear enough to explain things in sufficiently simple language. Gary Becker puts it well -- economics is both really easy and really hard at the same time. When things get cheaper, people buy more, and when they get more expensive, people buy less sums up about 50% of it. I stumbled upon this great Becker interview again and thought I would link to it.

On an unrelated note, an old school chum has these comments on this site. Read his comments first (funny).

Robert Rubins speaks at Chicago GSB

Recently, Treasury Secretary Paul O'Neil came to speak at U Chicago. He was a nice enough fellow, but kinda dumb. He began by talking for 40 minutes on improving safety at Alcoa, for reasons which remain opaque to me. This being Chicago, the Q&A session focused on securities regulation and tax policy. About 10 minutes into it, Paul encouraged us to ask questions about his trip to Africa with Bono. We laughed, and then refocused on Glass-Steagal.

Yesterday, Robert Rubin, Treasury Secretary under Clinton, came to speak at the school. The contrasts between these two men were staggering. Rubin is a very very smart and thoughtful man. O'Neil just seemed way out of his league. The 30 minute Q&A afterwards spanned everything from expansionary fiscal and monetary policy, moral hazard in emerging market financing, total factor productivity growth and economic expansion, and market consequences of Reg-FD. Rubin handled them all intelligently without breaking a sweat. Unfortunately, he is no longer in charge. I am afraid.

Microsoft trial

As expected, the government caved into Microsoft in its anti-trust settlement and the company got off scott free. The loopholes around "business models" and DRM APIs mean Microsoft does not have to disclose anything to anyone -- just like now. Read about it on /..