caltech: (talk on) the power of the scientific e-thesis

Many thanks to Eric Van de Velde and colleagues for inviting me to Caltech to give a talk on the scientific e-thesis. Besides being excited about going to Caltech, I am delighted that they wished to record the presentation. (I gather this was technically successful). This is of enormous value to me as it is the only true immediate record of what I have said. When I give a presentation I organize my material as a large number (perhaps 10,000) HTML pages with associated images and possibly movies. To this I add various demos on my own PC – things like OSCAR, Bioclipse, etc. (I didn’t get as far as showing Bioclipse this time). The pages are roughly arranged in a 2-level tree category/page but I cannot remember what all of them are.
The actual presentation depends on the makeup of the audience (I try to ask a few people or for a show of hands) to get an idea of the topics that will be most interesting. I have a chunk of topics I am definitely going to cover but otherwise it is often what comes to mind as I go through the talk. (There is always far too much that I want to say). So there is no clear navigation path. I rehearse the major components to make sure they are still working. Unfortunately every browser “upgrade” destroys more demos. (At some stage I expect my SVG to stop working). I have to use both IE (for the SVG animations) and Firefox (for DBPedia). I had hoped we had left “best viewed in browser X” in the 1990’s but no such luck.
So there is no simplistic linear record in the slides. It isn’t easy (or often meaningful) to leave “a copy of the presentation” as it consists of thousands of slides which make little sense without my commentary and a record of what was shown in whta order. That’s why it is so valuable to have regular recordings using video and audio as Caltech have provided. So many thanks.
FWIW I try to add something new each talk – each one is a moving average of past and future. For this talk I very happily discovered a Caltech thesis which used Jack Dunitz ‘s ideas and work extensively. From my introduction:

Jack Dunitz postdoc’ed at Caltech 1948-1951 with Linus Pauling and Verner Schomaker. He pioneered the use of very accurate crystal strucures in giving an insight into chemistry. Here is an example of a Caltech thesis (1992) which uses his ideas. See Chapter 9.
In 1973-4 I spent a sabbatical with Jack in Zurich showing how data from the literature could be combined to show new phenomena.

PMR: So I found a rather nice timeline. In the early 1970’s Jack and co-workers would investigate chemistry by looking at very accurate crystal structures. These could take a month to complete. The thesis refers to their work on non-planar amide bonds where they would design a molecule, persuade a colleague to make it, and do the structure. It could take years to investigate the idea. When I visited in 1974 Jack and Hans-Beat-Buergi had shown that it was possible to use structures already in the literature. I used this idea to look at the distortions of tetrahedral molecules (simulating the SN1 reaction). There were over 100 data points and each required me to locate a (physical) paper in a journal with geometry about the molecule. Often I would find a paper that looked promising only to discover that there were no data or it was too inaccurate. So it took severl months to find enough papers with good enough data. I got to know the system of document retrieval rather well! Now, 30 years on, we have all the information Openly gathered in crystaleye. We can pull out tetrahedral molecules in seconds.
That still relies on the conventional publication process. If we can do the same thing for theses we shall have unlocked an enormous source of data.
I look forward to the recording and will alert you when it appears. I warned the audience that details were unpredictable and the machine would get tired towards the end. So in the spirit of Openness you will see it all as it happened!