Common Lisp expert at SRI International for the Aura project. The system consists of 750,000 lines of Common Lisp. It represents knowledge and reasons about it. It took a long time for the system to start up and it used much memory. Together, these issues reduced its usefulness as a platform for development.
By a careful study of exactly what was happening during system startup, I modified the initialization process, resulting in the startup time speeding up from 15-20 minutes to a few seconds. This was a game-changer for the developers, as their thoughts were no longer seriously interrupted by the inevitable occasional system restarts.
I addressed the memory issues by collecting and presenting data and statistics about how much memory was used for each data type. With these statistics at hand, the original designer was able to improve significantly the memory usage of the entire system.
As is often the case, the system's functionality had been given priority over its documentation. Together with a colleague, I developed a system for automatic documentation of the entire system. The resulting documentation system was very complete and helped newcomers to get introduced to the system much more quickly.
Ported a large graphical interactive expert system developed on Lisp machines and Sun workstations to VAX workstations with the XWindow system. The system represents knowledge and reasoning about geological reservoir modeling. Purpose was to make the system much faster and to make it robust enough for production work overseas. By careful analysis of the graphical parts of the system, I improved its performance by a factor of 5. Other improvements made the system stable enough for successful deployment overseas. The original developers adopted my version as platform for further enhancements.
Accepted the position of President of the Association of Lisp Users (ALU) holding the private vision to organize International Lisp Conferences (ILCs) at well-known universities, such as Stanford, Cambridge (UK) and MIT. Lisp was invented at MIT and it was suitable to organize an ILC there at the 50th anniversary of the Lisp family of languages. As other board members lived closer to these institutions, my purpose as president became to provide leadership, which meant supporting and providing the means for others to succeed in organizing these conferences. The ILC conferences have been and remain very successful venues for the community of Lispers to meet and discuss the things close to their hearts. During my presidency, ILCs were organized in 2005 at Stanford, CA; in 2007 in Cambridge, UK; in 2009 at MIT, Cambridge, Massachusetts, USA; in 2010, at Sparks, Nevada, and in 2012 in Kyoto, Japan.
At the University of Amsterdam gave a course on Structure and Interpretation of Computer Programs, after the book by Abelson and Sussman. Clearly, the software discussed in the lectures was in Scheme. The lectures always had a large audience during the entire semester. Although the examinations had the lowest averages of the faculty of Mathematics and Computer Science, those with high grades quickly got their degrees and, if applicable, were successful PhD candidates. Statistical analysis of the examination data revealed that the examinations were very reliable. By the same analysis, students who repeatedly failed could be told which principles to study better, by which advice they succeeded with good grades.
"Solutions as business"
"Solutions as business"
Changed the initialization process, speeding up the starting-up-process from 15 minutes to a few seconds.
Developed a system for automatic documentation
Created a widely adopted platform for a GIE system
|President ALU -
Organized ILC's all over the world