Friday, August 13, 2004

Schwartz unplugged

It seems that Sun head honcho Johnathon Schwartz has a weblog. Most of it is a tirade against IBM, which is understandable because IBM is eating Sun's lunch.

He has a post about Sun's definition of "open", which is, essentially, easily substitutable. But the ease of substitutability is only really a factor if it allows the vendor to raise prices. If a piece of open source software is priced at zero, it does not matter how easy it is to sub out of the stack and replace with something else. In addition, if all the rents from downstream lock-in are factored into the (reduced) upfront sale price, then lock-in, with subsequent prices raises, does not matter either.

When Schwartz claims that the real price of porting out, say, one piece of middleware and replacing it with another, is retesting it with other applications of the stack, he is correct. In the large IT implementation I am currently involved with, this type of "regression" testing is indeed tremendously expensive and time consuming. But even more expensive and time consuming are all the business processes and human capital investments that need to be reworked to get any value out of the technology at all.

The McKinsey Quarterly had a good article about how IT investment only lifts productivity if couples with "good management" (which I assume means decisions about business rules that are sane, clear, and precise, instead of the usual mix of complexity and imprecision). To wit:
Our research, undertaken in partnership with the London School of Economics, focused on the period from 1994 to 2002. It offers evidence that specific management practices foster higher productivity regardless of a company's location, size, sector, or historical performance.

...To put this into perspective, such [a management centric] improvement has an effect comparable to that of raising capital investment by 70 percent, going from 10 manufacturing plants to 17, or increasing the workforce by 25 percent. What's more, companies got the same benefit from improved management regardless of where they ranked on our scale. In other words, even well-managed companies get a big bang from these efforts.

As you would expect from such a large jump in productivity, the impact of better management on the financial performance of individual companies was also impressive. The same one-point improvement on our scale was correlated with a five-percentage-point increase in a company's return on capital employed.

Compared with those results, how do IT investments stack up? We found that additional computing power6 also translated into higher productivity—but the impact was modest. The top quartile of companies, as reckoned by the level of their IT deployment, had a total factor productivity just 4 percent higher, on average, than those in the bottom quartile—just one-sixth of the impact of a one-point improvement in management practices. Moreover, companies with more powerful IT didn't do better financially. That may seem odd, given the rise in productivity, but one likely explanation is that the cost of new IT investments balanced out the financial gain they generated. Again, these results held good regardless of a manufacturer's location, size, or industry.

Of course, managers shouldn't stop buying computers. Rather, the results show that companies can get the biggest benefit by combining IT investments with good management. For corporations scoring in the bottom quartile of management practices, the deployment of more powerful IT is associated with productivity improvements of just 2 percent. However, companies with increased computing power and improved management practices achieve 20 percent higher productivity. This result shows that better management practices can raise productivity a good deal by themselves and increase the impact of IT investments on productivity as well. Companies should first improve their management practices and then invest in IT.
The implication for calls to put IT into poorly managed, low productivity environments (such as schools) is clear.


Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home