August 2003 Archives

More Thoughts on Outsourcing

| 2 Comments | 2 TrackBacks

A lot of people are talking about outsourcing of IT to India.

Brad DeLong writes from the point of view of a political economist:

I think that the correct policy response is the one outlined by Robert Reich in his Work of Nations of a decade and a half ago: First, get our people out of industry segments where we are about to lose comparative advantage and where wages are about to take a big dive--this is the reason we Democrats like various forms of Trade Adjustment Assistance, for those who work in such industries are about to get shafted and have done nothing to deserve it [...]

How do we know if an industry segment is about to lose comparative advantage? Do we really want to give assistance to programmers who have been earning large salaries compared to their white-collar peers?

Not too long ago, Business Week featured an article on the same subject.

It means the career you studied for — and spent oodles of tuition money on — in college probably won't sustain you for your working life. Someone in India or China will be able to do it far more cheaply.

Most inflammatory of all, Sue Spielman (who wrote a pretty good book about the Struts framework) suggests that

When you are asked to train the foreign worker to do your job or to support your product line because in 6-8 months they will be doing your job (or the job of a co-worker), simply refuse. And if you must, quit the job.

In the comments threads it's clear that people are getting upset. It seems that economists (like Brad DeLong) and non-economists see the world in utterly different terms. The economist points out Ricardo's law of comparative advantage; the non-economist says, in effect, I don't care, it's hurting American IT workers, so it's bad. People get emotional about foreigners doing the work that highly paid Americans used to do for a fraction of the cost. Unemployment in the U.S. is low, but I suspect that it is worse to be unemployed in the U.S. for any length of time than it is in any European country.

I also suspect that Sue Spielman is right in many cases. Outsourcing to India might look like a quick fix for many an ailing corporation, but the problems she talks about do occur. They're not even unique to India — any kind of outsourcing is vulnerable to quality problems and hidden costs.

In part, it comes down to one's time frame. In the long run, countries benefit from specialization and, the way the world is going, there will be plenty of areas to specialize in, even within IT. In the short term, though, people know programmers who have had their jobs transferred abroad. Anecdotal evidence makes for a better argument than theory. As Brad DeLong implies, if the job market recovers in the U.S. nobody will care nearly as much about outsourcing.

UPDATE: the web log was mistakenly configured to not allow anyone to post comments. My apologies to anyone who tried to post but couldn't!

How to Convert Fahrenheit to Celsius Quickly in Your Head


When I first came over the United States, I was faced with new units for almost every measurement. One of the more common ones was temperature, which could be frustrating when I was talking to relatives and trying to tell them how hot (or cold) it was. I knew there was a conversion formula that involved 9/5 (or was it 5/9?), but I never managed to remember which way was which.

Passing the Northern Trust bank at Adams and Wacker most days, I get to see their time/temperature display which, like most of them, shows both Fahrenheit and Celsius. That inspired me to make the effort to learn how to make the conversion in my head.

The temperature in Celsius is defined to be 0° at the melting point of water and 100° at the boiling point. The equivalent range in Fahrenheit is from 32° to 212°. That means that a 100 degree interval in Celsius is equivalent to 180 [212 - 32] degrees Fahrenheit so 1.8 degrees Fahrenheit equals one degree Celsius. (1.8 is the same number as 9/5 but I find the decimal version easier to remember.)

If you're converting temperatures you need to normalize for the 32 degree offset. When converting to Fahrenheit, you need to add 32; when going the other way, you subtract the 32.

Fahrenheit to Celsius
c = (f - 32) / 1.8
Celsius to Fahrenheit
f = c * 1.8 + 32

To convert quickly in my head I just round the numbers, so

Fahrenheit to Celsius
subtract thirty, then halve
Celsius to Fahrenheit
double, then add thirty

The approximations are good enough to get an intuitive idea of how hot or cold it is. For normal temperatures that we humans can tolerate, they're usually within a few degrees. For more extreme temperatures, they're really not very accurate.

Some examples comparing the result of the approximate calculation to the exact conversion

Fighting Spam = Rolling Boulders Up Hills


It happened to us twice at work: boatloads of spam being relayed through our e-mail server.