Wednesday, 4 November 2015

Mighty Shannon

A few days ago, I was talking about John Cleese and his attempts to get to grips with the act of creation through reading the key sources and relating those findings to his own experience. I've also alluded to the sources of inspiration experienced by mathematicians such as Cedric Villani. On a personal note, in a life time of science, I've had three good ideas, one of which probably turned out to be wrong. I've had a few more Aha! moments, to be sure: it would have required enormous patience and dedication to carry on at the coal-face getting positive feedback only once a decade.  I've suggested that one way to get an idea is to have half an idea and then lob it across the table at some colleagues. Famously, when Sydney Brenner and Francis Crick shared an office in the MRC for 20 years, they had a rule that either could say aloud whatever came into his head.  If you've got no friends, then I've found that cycling to work has delivered, not only mighty thighs, but also a good number of insights and solutions to problems. The exercise increases the circulation to the brain (and everywhere else); to survive a decade of city cycling you have to be alert; that extra brain activity can unconsciously be channelled into problem-solving.  That's handier than PoincarĂ©'s solution to having ideas which asks you keep stepping onto buses preferably in Normandy.

Now here's a novelty: we live on - are the only inhabited house on - Shannon's Lane in the Blackstairs Mountains hills.  It's named for the family that used to live in our farmsteading. The most famous member of the Shannon clan is surely Claude Shannon the mathematician and engineer. He became the Father of Information Theory FIT by applying Boolean logic to circuit-boards while working in Bell Labs. We'd probably have got to universal mobile phone usage and rockets to alpha-Centauri without him but it would have taken longer.  During WWII, he met Alan Turing and had a conversation in which half ideas were lobbed back and forth across a canteen table. Part of me wishes I'd been a fly on the ketchup bottle taking notes, but the other half knows I'd have been unable to follow the Smart Guys as they romped through ciphers and circuits way above my head. It turns out that a re-found friend of mine is also interested in the trappings of creativity. His father wrote A History of Engineering and Science in the Bell System, so knew a thing or two about The Shannon. As well as being a whizz at information theory, code-breaking, robotics, juggling, unicycling and chess, Shannon had thoughts about where the ideas came from. His piece on Creative Thinking, written in March 1952, is pages 529-538 of Claude Elwood Shannon Miscellaneous Writings edited by N.J.A. Sloane and Aaron D. Wyner which is available as a mammoth PDF (half Gbyte!). I've filleted out the relevant pages as a separate, much smaller, PDF.

Here's the executive summary of what Shannon, a true creative genius, had to say on the matter of thinking creatively.  Back in our Geordie days in the 1980s, there was a radical Scottish theatre group called the 7:84 Company; named for the observation that 7% of the people owned 84% of the wealth.  Shannon believes this to be true of creative people: most of the useful ideas come from a tiny subset of the whole population.  The same power law applies to word use in any language and to publishing where Harry Potter sold as many copies as the whole rest of 20thC fiction.  He believed that three attributes needed to be present in a truly creative mind.
  • Experience, training, knowledge.  You have to know something about the field before you make earth-moving contributions.
  • Smarts, nous, IQ. Creativity is not for dullards.
  • Drive, curiosity, motivation. Creative people like to find things out, how things tick.
He then goes on to elaborate on attributes, some of them possibly trainable or learnable, that increase the hit-rate; perhaps particularly through early experience when we are little more limber in the brain.
  • Positive dissatisfaction.  A hard-to-shake certainty that we could do better . . . at everything. If we tweak this and try that, it might all work more efficiently.  It is the opposite of a lazy-arse, pragmatic "If it ain't broke, don't fix it".
  • A deep appreciation of a neat&nifty result, even if you haven't arrived at it yourself.
  • A terrier-like determination to finish things off, even if it means missing lunch or staying late. [I was never so alive as when I was in graduate school working all the waking hours].
Shannon then lists his toolbox, the tricks of his trade that had helped him find solutions to the most intractable of problems.
  • Simplification. Trim the fat off the problem and reduce it to its bare necessities. Identify the key issues and stop being distracted by details on the surface. Having sorted the core, you can re-clothe the whole with refinements and additions.
  • Seek similar.  This is where deep knowledge of the field pays off: you have in your head hundreds of previous problems and their solution. Andrew Wiles saw the similarity between two utterly different branches of mathematics to solve Fermat's Last Theorem.
  • Restate, review, look sideways.  Get a different perspective on the problem: go for a run, write it all backwards. This is where a [smart] greenhorn, with no experience (but no baggage), can make the key contribution.
  • Generalisation. Take your narrow, limited, specific solution and ask, with a creative person's restless curiosity, if it is applicable in other cases, under other conditions in different fields.  "Can I use this same clever idea represented here to solve a larger class of problems?"
  • Deconstruction.  If you think hard you may be able to show the essential structure and connectivity of the problem.  That will help break a lumbering monster into chewable slices which can be knocked off piecemeal.
  • Clunky is okay.  A first working model may be a kludge, held together by chewing gum and string, but, hey, it works.  See above for how positive dissatisfaction will streamline your monster into something more elegant and efficient.
  • Turn it over.  Sometimes the different perspective is to take the solution and work back to find the problem.  Our technological world is full of inventions that had no purpose until they existed as solutions: the glue on post-its, thalidomide as therapy for leprosy, teflon, Sugru.
Good luck, I hope you have good more ideas than me.

1 comment:

  1. That's good stuff from Shannon on creative thinking. Regarding the use of Shannon information in genetics my humble opinion is that information is the wrong metaphor for the whatever-it-is that associates sequences of DNA with features of genotypes. I am experimenting with the notion that 'meaning' is a better metaphor, if interpreted in terms of the semantic system that I have identified, that applies far beyond language into organismal cognition and perhaps beyond that into the intracellular domain. The definition of the constituents of this kind of meaning, in brief is: the dimensions of the space of interaction I propose that it also applies more basically to the functionality of natural artifacts such as DNA. Such meaning is the structure of the active content of genes viewing this content as 'DNA memory'. As well as this dimensional structure of DNA words, genes possess the equivalent of the 'encyclopedic meaning' of words that constitutes the bulk of the content of DNA meaning and memory. This is the know-how of phenotypic features. Given this structure, factoral (my term for this semantic system) meaning pervades the intracellular domain as it is involved in the innumerable interactions of molecules and organelles and other constituents. In my text I have previously hypothesised that the factoral system is the space of the interaction of organisms with their environments. My genetic hypothesis is a further extension of this notion and an attempt to demonstrate the extreme primality of this system. The system is in two parts. One part relates to the structural physical features of things and the other to the normativity of an event such as a cellular stimulus that requires a response.
    All of this implies that factoral meaning is integral to the genetic system.
    Your Shannon quote says "Information is the resolution of uncertainty". The role of my semantic system is the determination of the semantic flow in chemical form from stimulus to features of the genotype.