Héctor Garcia-Molina, 1953–2019

We all lost a great person

Cropped from Mexican NotiCyTI obit

Héctor Garcia-Molina died just before Thanksgiving. He was a computer scientist and Professor in both the Departments of Computer Science and Electrical Engineering at Stanford University.

Today Ken and I write in sadness about his passing and to express some personal appreciations.

We at GLL have talked about him before. See this for his story about the fun of using an IBM mainframe for teaching. Or see this for a story about Héctor and meetings.

I, Dick, had the pleasure to have worked with him, while we were both faculty at Princeton in the 1980’s and beyond.

His Clock

Héctor was the chair of the Stanford Computer Science Department from January 2001 to December 2004. Stanford then rotated the chair so all took their turn. I know that he “hated” being an administrator in general. But of course being a team player he took his turn.

One way to see his real feelings about being a chair was to look at the clock his students constructed for him. The clock was a digital timer that counted down the seconds that remained in his term as the chair. It started at roughly 126227808. I am sure he did fine, but the clock was a statement.

His Worst Paper

While Héctor was at Princeton we worked together on a project—the MMM project. It led to one of his least cited papers—a 1984 paper with me and Jacobo Valdes. OK, it has 48 citations according to IEEE. The idea of the project was to use memory rather than processors to speed up computations. He was a joy to work with: he was careful, and thoughtful, and just fun to work with on any project.

We did write a second paper with Richard Cullingford instead of Valdes, which Héctor presented at a meeting on knowledge-based management systems. The above Google Books link goes to the end with a discussion in which a simple issue was raised—we paraphrase:

Won’t it take forever just to write all zeros to the memory?

Héctor had a scientist’s answer: the project was still not at the prototype stage so he didn’t know. He said the project should be viewed as a scientific experiment to find out. Maybe he also had an inkling that what was coming was a decade of breakthroughs in CPU design and parallel/pipeline processing after all.

His Predictions

Héctor was unparalleled at seeing the future. I always thought that one of his abilities, one that set him apart, was his ability to predict directions of research. This allowed him, and his students, to write early papers in research areas before they became hot. This is one of the talents that made him so amazing.

I recall way before the world wide web was created he had students working on adding links to documents. I recall a talk by one of his students at Princeton that discussed what we now call URLs. One question that was raised during the talk was: How were the links going to be created? There was a lively discussion about this. Could they be created automatically? If not why would people take the time to create the links? Indeed.

Héctor saw that links would be created. That people would take the effort to create them. I must admit that he was right, and he saw the future better than most. I wish I had a fraction of his ability to see directions like he did.

Guiding Students

Héctor told me that when he first got to Stanford the fund managers and investors roamed the halls. They would ask anyone they could if they had an idea for a company or a startup. It was a constant issue that Héctor had to deal with. They were continually trying to steal away students.

He told me he felt like he was the head of an abbey and was always having to protect his charges within the walls.

When the impetus came from within it was different. Of course, Héctor was the advisor of Sergey Brin at the time he and Larry Page conceived Google. Brin and Page found that their search engine prototypes were so good the dataflow was constantly straining Stanford’s machines. They needed to scrounge for more disks and processors to mount their servers. Héctor already oversaw the Stanford Digital Libraries Project and he arranged for funds to purchase spare parts for the data servers.

It is interesting that in this 2001 interview in the SIGMOD Record, Héctor did not have a high opinion of the industrial side of his area:

Again, I don’t think industry really does very much research. They come up with an idea and they try to sell it. If it was a good idea, maybe they will make money. Even if it was a bad idea, if they have good marketing people, they might still make money and we never know … I don’t think they have an advantage over [academics] in testing the ideas and evaluating them and performing measurements and really understanding what are the right techniques.

In the same interview, he had sage advice for students after completing their PhDs and during the tenure process, mainly on the side of not trying to play the system but focus on doing what you love.

Stanford Home Team

Héctor had been a graduate student at Stanford. So his return there as a professor was a kind of homecoming. He was a home-team player in many senses. One of them is shown by this photo:

Stanford University obituary source

He was a registered Stanford sports photographer. He also taught a course at Stanford on photography. We don’t know if he had special insights on detecting “deepfake” photos and videos.

Open Problems

Our condolences go to all his family. Héctor you will be missed. You are missed.

from Hacker News https://ift.tt/2rqTxWF