CONNERLABS

I’m somewhat of a fan of Nassim Nicholas Taleb’s book, “The Black Swan”. To me at least, the book is about the human tendency to tell ourselves stories about reality, and then substitute the stories for what is really there. This idea should be familiar to any student of Zen. Taleb calls it the narrative fallacy, and explores its messy implications in business and finance.

It struck me today that this is how electronic design proceeds, too. We start by telling ourselves a story about how the proposed circuit will work. The electrons will go down here, some of them will go this way, this part will oscillate at 123 megahertz, and so on. We either make it up in our heads, or orchestrate it on a circuit simulator, but in either case we are dealing with a Platonic approximation to the real circuit.

It follows from this that the lab bench is Taleb’s “Platonic fold”, where our narratives collide with the messy reality of what the prototype circuit actually does. This is the origin of the pearl of wisdom attributed to (the sadly late) Bob Pease or someone similar: that a circuit always works, it just doesn’t always do what you expect. Anyone who has done any practical work with electronics knows the brain-wringing feeling of struggling with a circuit built on wrong assumptions in this way. It simply refuses to do what you want, for no reason that you can see, because your reasoning is based on the same faulty assumptions. The best you can hope is that you have the “Aha moment” and come away with your narratives more firmly grounded in reality.

It also follows that by going into production with a circuit that doesn’t work the way you think, you invite it to start doing things that you didn’t expect when it gets out in the field. This can generate Black Swans in exactly the same way as Taleb’s example of running a hedge fund based on invalid mathematical models.

Usually the results are negative and your company simply goes bust, but once in a while you can benefit, as in Bob Pease’s tale of the Philbrick P2 op-amp. This was a groundbreaking product that contained about $5 worth of components, but delivered enough value-added to the customer that it could be sold for the price of a small car. The P2 made the company, even though (according to Pease) nobody in the company actually understood quite how it worked. But in spite of this they managed to produce it consistently and have it work reliably.

What’s more, if this is true then the world of electronics must have its “Fat Tony” characters, rather than being purely the province of “Dr. Johns” as one might expect. (For those unfamiliar with the book, you might like to mentally substitute Thomas Edison for Fat Tony and Nikola Tesla for Dr. John.) They are probably the same people that George Philbrick called lightning empiricists, after the fashion (though before the time, this being the 1950s) of Taleb’s skeptical empiricists. Indeed, Fat Tony would probably have wholeheartedly approved of the above mentioned P2, if he didn’t actually design it.

Anyway, that’s me on the narrative fallacy in electronic design. Next time I’ll write about the normal distribution and power laws. Taleb has his “Great Intellectual Fraud”, and communications theorists have their AWGN – “Additive White Gaussian Noise”. Until then, what are the odds of Bob Pease and Jim Williams dying the same week? I make it about 1 in 9 million, but sadly it happened, as Pease crashed his 1969 VW Beetle on the way home from Williams’ memorial service. Both were legends of analog electronics, and the impact is hard to overstate: it’s as if Jane Goodall and David Attenborough got trampled by the same elephant.

In a twist that Taleb may have found bitterly amusing, Pease had just self-published a book on safe driving, which didn’t sell.