Monday, 11 November 2002 9:00 am
What is needed to return to the environment where demand for MIPS, storage, and bandwidth grows faster than cost decreases driven by Moore’s Law?  In order to assess this, we need to look at the industry structure, to see what drove demand in the first place, and whether those things are still intact.  The industry structure is different for the telecom industry than for the computer industry.  Let me start with the computer industry.

In the last twenty years the computer industry has evolved from vertically integrated companies attempting to do everything themselves, to a network of companies playing specialized roles.  Some companies just make chips, some just make software, some just make circuit boards, some just do final assembly, some just take orders over the internet and distribute to end customers, some just integrate solutions for businesses, some just attempt to commercialize a specific new technology or a specific new market, etc.  

Rapid technological innovation has been one of the results of this structure. A National Research Council study of where major computer industry innovations came from, found a non-linear path that often starts in government labs or universities, then moves back and forth between them and corporate labs.  Although the corporate labs played a key role in reducing research to practice, the parents of the labs often were not the companies who ended up bringing the most successful versions of the resulting products to market.  Often, that role was performed by startups funded by venture capital and created specifically for that purpose.

Corporate research labs, particularly Xerox Palo Alto Research (PARC) and Bell Labs, played very key roles in the transformation of many of the key modern computer technologies from research to products.  When both labs were established, their owners were monopolies who could afford to do research that advanced the state-of-the-art, but did not necessarily immediately benefit the parent’s bottom line.  Today, both the parents facing very tough competition, are in precarious financial situations and can no longer afford their corporate labs. Who will pick up the slack?  Although occasionally government labs have played the needed role (most notably Illinois Supercomputer Center in the case of the browser),  the National Research Council study showed that most fundamental technological innovation in our industry has needed the corporate research lab to become billion-dollar revenue generators.

The current lack of new “killer app” software that would drive the next wave of demand for MIPS, storage, and bandwidth may be a direct result of the demise of Bell Labs and PARC.   The computer industry needs a rescue/revitalization plan for both Bell Labs and PARC, and needs one soon.  Some sort of consortial approach is needed – but with a much better business model and governance system than MCC or Open Group Research had (both now history).

Sorry for posting the daily column a bit late this morning - we were under a tornado warning and had to shut down all of our computers.  I will be back to the 8:30 AM posting schedule tomorrow, God willing ...