camdez.com

Rule #1: There are no rules.

The Importance of Tiny Things

| Comments

This weekend, as I was waiting at my doctor’s office for the results of my rapid strep test, I downloaded and started reading Kurzweil’s new book, How to Create a Mind. The first chapter relates the story of geologist Charles Lyell (1797–1875), his influential Principles of Geology, and the impact it had on Darwin’s work on natural selection.

Particularly interesting to me is the idea that (apparently) one of the greatest contributions of Lyell—and one of the hardest to swallow—is the understanding that geologic change, such as the creation of canyons, results from the aggregation of tiny changes over long periods of time. This seems so incredibly obvious now, but Kurzweil writes that “this proposal was originally met with ridicule, but within two decades […] achieved mainstream acceptance.” Darwin’s On the Origin of Species was met with similar doubts (and occasionally still is).

It makes me wonder if there’s a larger theme of a collective realization of the importance of tiny things in aggregate. In some ways I think the realization might still be on-going, and ever will be since the human mind doesn’t seem especially adept at this kind of reasoning. We tend to round numbers to quantities easier to handle (e.g. the odds of getting in a car crash seem so low so we effectively round them to zero, and 3:13pm is a lot like a quarter past three), which works fairly well until we’re dealing with a system where these numbers compound.

There are scattered traces of this understanding going all the way back to Aristotle and Lao Tse, but I suspect a widespread, shared understanding within the scientific community was absent, and certainly a formal framework was. Even the term “emergence” was coined in 1875, the year Lyell died. Chaos theory originates in the 1880s with Poincaré. These things are certainly coincident, if not correlated.

It makes me wonder if the title of Stephan Wolfram’s A New Kind of Science, a weighty tome discussing cellular automata, might not be quite so presumptuous as it seems at first glance. We now know that the simplest of things like a trivial cellular automaton can implement a Turing machine, the whole of modern computation can be reduced to what can be implemented by a Turing machine, and cellular automata can be found in nature. Obviously computation can happen on larger levels (after all I’m typing this on a computer), but at this point it’s hard not to think that computation is happening all the time, all over the universe, at even the tiniest levels. We need a framework to understand these things if we’re to progress in our understanding of the universe.

But as this formal framework matures, we’d all do well to start to accepting the import of smaller things, even if we don’t have a rigid understanding of it. After studying artificial intelligence and natural language processing in graduate school I decided that I didn’t want to work in that field immediately because the state of the science was / is, to me, frustrating—most modern AI research is done using complex systems like neural nets which, once trained, defy understanding. The golden age of artificial intelligence, where we build machines with an organized, clean internal understanding of the external world is seemingly gone (here’s a fantastic explanatory interview with Chomsky on the subject). Whatever you say about neural nets, though, the one thing you can’t deny is that they work (at least to a point). We do know how to use them. But I, for one, don’t particularly want to father children I can’t understand at all. The framework for understanding needs to catch up to the framework for creation. Maybe we’ll always be able to cause phenomena more complex than we can understand, but it feels like the bar for understanding is frustratingly low.

I’m wildly off-topic, but it’s interesting to me to look at my previous post and realize that all of the points I outlined were about friction, the impact of tiny forces. Not knowing where to write (categorization), not knowing who to write to (audience), and not knowing if I should write publicly (privacy) are all impacting the much larger issue of actually writing. It’s so much easier to see on “paper”, when a bit more consideration is given and a bit less gut instinct applied. But little things do matter, and a broken pencil or a drafty house has probably forestalled a Hemingway or a Handel.

As it turns out I do have strep—for which I can thank an extremely large number of low-probability transmissions of a minuscule bacterium rapidly evolving through numerous unlikely mutations of a tiny computational molecule called DNA.

Comments