Perplexed looking for a guide

Red shift, Blue shift, ...

All companies process raw materials to produce more valuable products that can be sold for a profit. Until recently, these raw materials where usually physical resources: iron ore, cotton fiber, coal, etc. With the onset of computers, it become practical to create companies that process information, as a valuable raw material, to produce valuable products. Google has figured out how to convert (indices of) information into advertising targets and thus into profits. Salesforce.com and eBay manage and broker information to make it more valuable to their customers. Banks and arbitrage firms take the raw data of markets and identify times to make acquisitions and when to sell. The ultimate version of this is the intelligence agencies (and newspapers), who take scraps of noisy data, process, and clean them up, identify relationships and produce extremely valuable insights, when they get it right.

On the otherhand, almost all companies use information technology to manage customer lists, payrolls, etc. In this case, information technology is part of the overhead.

When information is a raw material and processing improves value, increasing the cost/performance of IT improves the total profit proportionally to the size of the customers (at least). If it is part of the overhead, the impact is much less.

As a consequence, companies that produce information based (or redshift) products are driven to increase their IT infrastructures as rapidly and as cheaply as possible (scale means more value, reduced cost means more profit). Companies whose product is does not depend on information as a raw material (blue shift products) are less much less sensitive to IT cost; it is just overhead and the cost (and benefit) isn't multiplied by their number of customers.

An interesting question that remains is what sort of computations are done to "refine" information raw material. Most that I am aware of are embarrassingly parallel operations. These are the easiest (and safest) to scale up. But are there more communication and computationally intensive "refining" steps as well?

Posted at 02:01PM May 14, 2007 by rezippel in Sun  |  Comments[0]