NOKHAW and the New Magic

 

According to legend, in the 15th century, maybe even the 16th, it was possible for a single person to know everything - to be up to date on every development in math, science, art, literature, etc..  We are sometimes given the figure of Leonardo da Vinci as such a "renaissance man".  

But since then, each discipline has grown to such an extent that no one can be an expert in everything.  You can be a specialist in one field, and understand it in all of its detail, but you will necessarily have a shallower understanding of other fields.

Don't forget the "Murray Gell-Mann Amnesia Effect" - namely, a scientist reading an article of science journalism in the news will generally recognize a large amount of vagueness, over-generalizations, inaccuracies, and sensationalism when the article pertains to the scientist's own field, but will tend to forget this when she reads an article about a field other than her own.  Moral: in matters of science, never trust writers.  Ironically, this term was invented by a writer, Michael Crichton, who put the physicist Murray Gell-Mann's name on it because he thought it would lend the principle more prestige, even if crediting Gell-Mann was inaccurate.

As time has gone on, even specialties are breaking up into sub-sub-specialties, making it all the more difficult, if not impossible, to be a generalist.  The appearance of niche subspecialties can be surprising, if not mind-blowing and even a little bit comical.  I remember the moment in the 90s that I found out that there is such a field as paleometeorology: the study of ancient weather (I happened to be reading about the study of rain damage on the Sphinx, which was, back then, a major point of controversy in dating the Sphinx).  I burst out laughing - I couldn't believe that such a thing as a paleometeorologist existed.  Now it seems perfectly natural to me.  So it goes.

Samuel Arbesman, in his book, Overcomplicated: Technology at the Limits of Comprehension, back in 2016, pointed out that technology had reached an inflection point in complexity.  An example is the personal computer.  In the 1970s, when Steve Wozniak was putting together computer kits, it was possible for a single person to understand every single part of a computer and how it worked.  But now, with the growth of complexity of computers, no single person can completely understand how they work, as Wozniak himself has admitted.  Different experts can understand different components with greater and greater depth, but the entire process is just far too complex for any individual human mind to comprehend.  The computer industry has become a profoundly and inextricably social phenomenon.

This process can only expand, and expand exponentially, with the birth of AI.  Dario Amodei of Anthropic has estimated that even the greatest experts at his own company probably only understand about 3% of what Claude does, and that was before the release of Mythos - it's probably a lot lower now, and as these systems get more sophisticated at an exponentially increasing pace, our ability to understand our own technology will shrink to a tiny fraction of what there is to know.  

What's really astonishing is that not only are humans incapable of comprehending this, the AI themselves do not understand how they work.  Although their powers are astonishing, it is remarkable how difficult it is for them to introspect.  There seems to be a limit to their ability to follow the oracle's injunction to "Know thyself".  

Think about this for a second.  As their sophistication increases, not only will the AI become inscrutable to us mere humans - and to each other - and to themselves - but also much of the technology that they develop may be so sophisticated that they themselves don't know how it works.  We are quickly approaching - if we are not already in - the era of NOKHAW: No One Knows How Anything Works.

Once you understand this, your perspective of history changes.  Perhaps that brief moment - that little island of time - in which Leonardo da Vinci lived - if it existed at all - was the only time when this kind of comprehension was possible.  Before that, and after that, only an ocean of time of incomprehension.

Richard Feynman used to complain about what he called "cargo cult science" - people going through the motions of science, but without real comprehension.  We are approaching, if we are not already in, an era in which all science is cargo cult science.  True comprehension is impossible.  We are returning to the era of what anthropologist Claude Levi-Strauss called "bricolage" - a semi-improvisitory, semi-ritualistic recombination of available imitative and repeated gestures to solve problems, rather than a deep understanding of scientific explanation.  We do things, but we know not what we do.  The era of new magic has begun.