By stern on Feb 07, 2010
One of our running conversations was how to tackle a problem that hadn't been seen before. What choices do you make, or constraints do you put in place, if you have to think about scale, speed, or complexity that isn't in the literature? This came up in our meeting with Hot Potato, who worry about real time and real life events, and in talking with a major sports league that provides video on demand but isn't sure how to quantify the "excitement" quotient of that video from day to day.
Borrowing a page from college days, I wrapped up our internship session with a reading list:
George Polya's "How to Solve It". The classic, rooted in mathematics and algorithms, to build up an arsenal of hard problem cracking approaches.
Simon Singh's "Fermat's Enigma". Aside from a lot of the mental action taking place at Princeton, the background on how Wiles derived his proof of Fermat's Last Theorem is great. I use this to highlight how a seemingly minor topic covered in one area becomes a major factor in another -- I had finished a podcast on elliptic curve cryptography when I read the book, and the overlap in mathematical bases was eye-opening.
The July 1997 Wired issue on scenarios, describing how a pandemic might be solved by a graphics designer and gene hacker working together. Since that issue first showed up on newstands, we've faced SARS, avian flu and H1N1 flu outbreaks. Our response mechanisms haven't gotten much better. On the other hand, both of the students had been working in an "integrated science" curriculum, where mathematics was more directly incorporated into the appropriate scientific fields. We just need to add computer science in there as well. I made the remark that one of my good friends got into computer science because she was a psychology PhD student who needed to analyze data; today the data analysis experts at social networking sites are creating work for the psychologists.
Michael Lewis' "Liar's Poker" followed by Lawrence McDonald's "Colossal Failure of Common Sense." Two views of Wall Street, from the mid-80s birth of the fixed income derivatives business to the second, third and fourth order effects of its growth that led to the demise of Bear Stearns and Lehman Brothers. What happens when mathematics isn't integrated fully into the financial engineering sciences.
Bruce Tognazzini's "Tog on Design". I've always found Tog's approach to design and user interface refreshing, and I think ideas like selective disclosure would improve much of today's popular (but badly used) software.
Cory Doctorow's short stories "Anda's Game" and "When SysAdmins Ruled The Earth." Networked business models, networked organization and networked government. Both written before the Facebook boom, and therefore more important in light of it.
Finally, each of the people we met with had some advice or guidance on life in the real world: (1) Think big and unconstrained, beccause that's what's happening to compute and storage environemnts. (2) Cross-scientific disciplines matter. No single science is isolated. (3) Watch out for "end arounds" caused by cost or time disruption (4) Stuff happens. When it does, it affects brands, reliability, user experience and customer attraction. Be ready for it. (5) It's always harder than it looks to pull the pieces together: Moore's Law hasn't applied to integration costs.
Spending time with university students is always refreshing, both to find out what they think is interesting and to see what hasn't yet registered in their curriculum. And it shows them a literal world of eating options beyond the undergraduate cafeteria and campus pizza place.