Now blogging at diego's weblog. See you over there!

from playstation to supercomputer

A group at the NCSA has built a supercomputer using PlayStation 2s and Linux's clustering technology for 50,000. Cool.

Categories: technology
Posted by diego on May 26, 2003 at 11:08 PM

news from the world

A compilation of various news articles from around the world on the "crisis of democracy", from Salon.

Categories: geopolitics
Posted by diego on May 26, 2003 at 11:01 PM

in praise of cooperation

An excellent speech by former US President Bill Clinton.

Categories: geopolitics
Posted by diego on May 26, 2003 at 1:12 AM

the Economist on Ethernet

The Economist this week has a good article on the history of Ethernet, and what made it survive past other "contenders" that didn't make it, such as Token Ring, or even ATM:

The first reason is simplicity. Ethernet never presupposed what sort of medium the data would travel over, be it coaxial cable or radio waves (hence the term “ether” to describe some undefined path). That made it flexible, able to incorporate improvements without challenging its fundamental design.

Second, it rapidly became an open standard at a time when most data-networking protocols were proprietary. That openness has made for a better business model. It enabled a horde of engineers from around the world to improve the technology as they competed to build inter-operable products. That competition lowered the price. What is more, the open standard meant that engineers in different organisations had to agree with each other on revised specifications, in order to avoid being cut out of the game. This ensured that the technology never became too complex or over-designed. [...] That, coupled with the economies of scale that come from being the entrenched technology, meant that Ethernet was faster, less expensive and less complicated to deploy than rival systems.

Third, Ethernet is based on decentralisation. It lets smart “end-devices”, such as PCs, do the work of plucking the data out of the ether, rather than relying on a central unit to control the way those data are routed. In this way, Ethernet evolved in tandem with improvements in computing power—a factor that was largely overlooked by both critics and proponents when Ethernet was being pooh-poohed in the 1980s and early 1990s.

Beyond the technology, there is even a lesson for companies investing in research, albeit one learned through tears rather than triumph. Xerox failed to commercialise Ethernet, as it similarly missed exploiting other inventions created at PARC, such as the mouse and the graphical user interface. To develop Ethernet fully, Dr Metcalfe had to leave PARC and found 3Com, now a big telecommunications-component firm. The lesson may have sunk in. In January 2002 PARC was carved out as an independent subsidiary of Xerox. That allows it to explore partnerships, spin-offs and licensing agreements without having to get its parent's permission.

So true. I have the feeling the lessons of Ethernet will become more relevant as time passes and we move more and more into a fully decentralized world.

Categories: technology
Posted by diego on May 26, 2003 at 12:17 AM

moving mountains for fun and profit

Ole reviews How would you move Mount Fuji? a book on interview techniques that focuses in particular (although not exclusively) on Microsoft and their famed interview questions. Great review.

Categories: technology
Posted by diego on May 26, 2003 at 12:09 AM

Copyright © Diego Doval 2002-2011.