Randall Munroe's xkcd is a necessary visit for anyone doing science, math, or engineering. Artistically simple, he often sums things up with an almost poetic simplicity. Some have suggested topic ideas - he's taken two of mine and created beautifully distilled views of the underlying concepts that make me envious of his talent. And he does it about three times a week. Every now and again he creates something that resonates across a community. It happened again last week:
Anyone who reads a lot of scientific papers recognizes this. Dozens of variations emerged overnight. Some were very specialized - at least three relevant to astrophysics alone and I only know that because I know the community. Here's one Abeba Birhane posted.1
Serious truth! A good deal of the community is this clueless.
It made me think of friction and objects.
One of the most remarkable features of physics is its simplicity. While it may be non-intuitive and sometimes requires a bit of tricky math, you can remove complexity and discover underlying bedrock. Once you understand the simple basics you can restore the complicating bits a piece at a time and describe a vast set of physical processes. The fact that you can deconstruct and reconstruct so effortlessly is unique among the sciences. It's why a physicist will tell you physics is the simplest science.
Beyond physics complexity and emergent properties become important. It turns out that asking "what is life?" is one of the most difficult questions. Working backwards from biology to biochemistry to chemistry to physics breaks down along the way.
But back to simplicity
Students learning about motion and collisions start with a statement like "assume a frictionless surface." Concepts become clear and it's easy to describe interactions where everything is understood. Two blocks colliding elastically for example. Later on you can restore friction. Mechanical engineers, working in the real world, are forced to deal with it. In some of their systems - say a chain and gear bicycle transmission - it's possible to have efficiencies that exceed 99%. Frictionless is the ideal here. On the other hand you don't want frictionless tires or brakes - you'd never start and, if somehow put in motion, you couldn't stop. The bicycle and road system is complicated enough that friction is good in places. Fortunately good engineers know how to deal with such things.
The concept of "frictionless" has extended to interactions between computational objects and people. In fact people are often treated as objects that contain a long series of properties. Constructions like these lack the meaningful emergence that social creatures have. Serendipity and insight are often the result of social "frictions" in our interactions with other people. We can have simple low friction interactions. Sometimes they're great as we're not looking for much, but they're generally shallow. A worry is large scale machine learning can create a faux sense that interactions with machines are deeper than they really are. And many of these interactions are with other parties who make decisions about and for us.
A few who build large scale social platforms have deep backgrounds in the social sciences and the humanities and think deeply about underlying social issues, but much of it is done without much clue by coders who build things quickly and then iterate to make them better - often more frictionless.
Separately I had two wonderful reports from people at a very creative company. Fully vaccinated people are back at the office with some precaution and maskless outside. One reported she's had more insight and exciting ideas in the last week than in the last six months - largely through random talks with coworkers where they could "read" each other. I hope all of you find the light at the end of the tunnel soon.
__________
1 She's a cognitive scientist from Ethiopia who specializes in critical race theory - a rather important intersection these days.
Comments