Sunday, February 29, 2004

Oracle vs PeopleSoft

I think Oracle should buy PeopleSoft -- the enterprise application business is struggling with overcapacity, and there is a real customer preference for greater integration between vendor products. But it looks like the DOJ has put the kibbosh on it for anti-trust reasons. The article lists SAP as the only remaining competitor if the merger goes through, although older versions of the software should count as well.

I'm Back

I am back from Dubai. I visited Dubai Internet City again (it was being built the last time I was there) and it seemed to have a high occupancy rate, with all the majors present, and plenty of minors too. Remarkable. Dubai is building offices, homes, malls, and hotels at a rate that defies belief. They are planning to make the world's tallest building (the actual target height is a secret) which leads me to beleive that much of this investment is speculative waste. The rational brain does not pick "tallest" as the guiding principle for a building.

Monday, February 09, 2004

Contrarian view

Winterspeak reader JC points to this piece arguing that historic interest rates have actually been much lower than the 3.5% real we currently see in the US.

Saturday, February 07, 2004

How deficits matter

The NRO has a piece decrying Bush for running up a big deficit. I don't care about the deficit -- and neither should you.

Instead, you should be caring about long term interest rates. The 10-year bond rate is a touch over 4% nominal, which I guess makes it around 3.5% in real terms. This is VERY LOW by historical standards. This is also VERY LOW by international standards -- countries that are lower (like Japan) are actually suffering through deflation. When Paul Krugman was sane, he recommended that Japan print money to trigger inflation, and one way to print money is to run up deficits.

The harm from deficit spending comes from what it does to the interest rate -- by raising rates it crowds out private investment and this is bad for growth. But the problem is not the deficit itself -- that can be rolled over indefinitely.

Friday, February 06, 2004

Rational risks

It has been frustrating to listen to the shouting match on whether or not the intelligence around Iraq's WMD program was any good. I'm not calling it a debate because the issue is being framed in such an asinine way.

A complete and accurate CIA assessment of Saddam's WMD capabilities would be "we think there is a 75% chance that he was WMDs, and we are confident about this prediction to the 50% percentile". This means that, based on the evidence, the CIA would put a 75% probability on the WMD program, but they are only 50% confident about that 75% bet because hey -- this is hard stuff to do.

I don't know if 75% and 50% are the right numbers, but the above is a much more accurate description of the reality of working with intelligence information. Note that it does not answer the "should the US have invaded or not" question, but it does make a joke out of the "we've found no WMDs now so the intelligence was wrong and the war was an error" shout.

Suppose the CIA had said "there is a 99% chance Iraq has WMDs and we are 99% certain of this" -- would it have been the correct decision to invade then? Remember, there is still a chance that they are wrong and we would be in exactly the same position as we are in now. And suppose they had said "there is a 25% chance Iraq has WMDs and we are 25% certain of this" -- it's not clear that not invading is the right choice here either.

Even if the CIA could furnish good, accurate probabilities and confidence levels for their prediction -- people would probably still make stupid arguments and bad decisions. People are simply appalling at thinking about risk and probability, really mind numbingly bad at it, and giving people good data does not make them any better.

Some links:
David Brooks gets it completely wrong as he criticizes the CIA for poor intelligence methodologies. His main criticism is that
The problem with the CIA's brand of analysis, of course, is that human beings believe things and feel passions. They have visions of a future world and are often willing to sacrifice themselves, or at least to take amazing risks, to bring those visions about. Their gauge of reality can be radically different from that of a cool scholar at an antiseptic research campus.

I've heard since September 11 that CIA analysts now really do understand Islamic extremism. And in fact, Global Trends 2015 represents an effort to get intelligence analysts outside the CIA involved in the Agency's thinking. But bureaucracies just don't think well. Propelled by their "scientific" training and their desire to quantify, intelligence technocrats find themselves modeling further and further into abstraction, leaving behind the lessons of history, the clash of ideas, the passions of the soul—all those things that can't be rationalized but are the real drivers of human affairs.
Ah--that pesky desire to quantify! Brooks think that numbers get in the way of truth, but the sad fact is that counting is one of the few ways we can actually make our assumptions explicit, calculate real probabilities, and take good bets (ie. make good decisions) instead of bad ones. I'm not sure that Brooks would really accept that making a good decision is absolutely equivelent to making a good bet. Idiotic.

Tavis Smiley, who I often listened to on NPR in Chicago, interviewsChristopherr Hitchens and asks
But I wonder how I am to juxtapose your assessment of Bush and Blair's actions or reactions to the "bad intelligence" they received as prudent. How do I justify your labeling it as prudent with...the mothers and the fathers whose sons and daughters are never going to come home here stateside, and in the U.K., and for all the women and children and others in Iraq. How do I juxtapose those 2 things?
I wonder what Tavis would make of "there is a 75% chance that Iraq has WMDs and we are 50% confident of this"? Is that good intelligence or bad? What decision would you make on the back of that? Would it be a good decision or a bad one? How could you tell? But hey -- why do any hard thinking when you can just yell "Bush and Blair LIED!!!"

There is a nice piece in the Chronicle of Higher Education that discusses how certain aspects of Western Civ have always been hated -- in particular, it argues
Calculation -- the accounting of money, interests, scientific evidence, and so on -- is regarded as soulless. Authenticity lies in poetry, intuition, and blind faith. The Occidentalist view of the West is of a bourgeois society, addicted to creature comforts, animal lusts, self-interest, and security. It is by definition a society of cowards, who prize life above death. As a Taliban fighter once put it during the war in Afghanistan, the Americans would never win, because they love Pepsi-Cola, whereas the holy warriors love death. This was also the language of Spanish fascists during the civil war, and of Nazi ideologues, and Japanese kamikaze pilots.

The hero is one who acts without calculating his interests. He jumps into action without regard for his own safety, ever ready to sacrifice himself for the cause... The symbolic value of these attacks is at least as important as the damage inflicted.
The truth is that calculation -- working hard to get expected outcomes, probabilities, and confidence levels right -- reveals reality, and actions based on reality will, over the long run, win against actions based on illusion (or even actions based on a poorer understanding of reality). Making a bold action against impossible odds is heroic, but it is also dumb, and in the long run accepting bad bets will make you lose. Decisions = bets. Good decision = good odds. Bad decision = bad odds. If you have a better sense of the true odds, you will make better decisions. Figuring out the true odds requires math, not guts.

Heading East

Am going to Dubai and Karachi for 2 weeks. Posting may be light.

Coase On Demand

Coase is popping up on this site a lot these days, but when you are talking about the boundary of the firm he really did say it all. I think it's interesting to think about the effect of current IT on where a firm should begin and end.

Essentially, you want to do something internally only when the coordination cost of doing it outside is too high.

Web services, among other things, lowers the coordination cost of sourcing various IT functions from external vendors (although it also lowers the cost of doing them internally). Look to see more IT processes externally sourced and hosted.

IBM & Sprint

It is an open secret that most consulting projects fail -- if by fail you mean do not deliver the performance improvements or cost savings they initially promised. The reasons are legion: 1) the client did not implement, 2) the client implemented, but did it wrong, 3) one part of the organization stymied the other part of the organization from doing the right thing etc. Failure usually happens because of a failure in execution, not because the actual idea as bad.

In the IBM/Sprint deal, IBM has promised Sprint that it can improve its call center operations. The deal has been structured to increase the chance of successful implementation.

Firstly, IBM is taking over huge chunks of the call center operations. This means that IBM people are both planning and implementing--eliminating the excuse that the other guy failed to execute or executed wrong. IBM pitches its integrated services and delivery model as "one throat to choke", and it promises none of the fingerpointing you get when, say, Accenture implements a Siebel CRM system to execute a McKinsey marketing strategy, and the whole thing falls apart.

Secondly, because IBM is taking over so much of that business, it cannot really blame failure on other parts of the business being stubborn. The marketing guys cannot blame the IT guys because they are all from IBM. Of course, things like network performance impacts call centers, but you would need to be the entire company to be able to coordinate everything.

Thirdly, the deal gets incentives right. Generally, call centers are pretty low on the corporate totem pole, so they do not get much attention or resources. But this is a very large deal for IBM, and Sprint has contracted for very strict service level agreements that really hold IBM's feet to the fire. An unimportant line of business for Sprint has become an extremely important line of business for IBM, which means they have a very strong incentive to do a good job.

Coase wrote that a firm should decide to do something internally vs through a vendor depending on where the coordination costs were lowest. A major cost of doing work internally is supervision -- people work harder for a customer than they do for a boss, especially if the boss is not around much. I always felt that firms used external consultants instead of internal corporate strategy teams (who knew the company and industry much better) to avoid this issue--but it also meant nothing ever got done (not that internal strategy teams are known for their ability to execute). IBM's willingness and ability to take on a huge slice of business gives it both powerful levers and incentives to effect change for its clients. We'll see what happens.

Wednesday, February 04, 2004

Stuff

The blog's been quiet, but things have been busy. As you may or may not have heard, IBM has taken over a big chunk of Sprint's call center business. It's a 5 year long multi-billion dollar deal, and I happen to know a little something about it. I'll post more tomorrow.