Alan Kay - Programming and Scaling

(Republishing the comment linked by Said Achmiz below.)

What Is This Talk About

This talk is about several things.

Software ‘Engineering’ - For the amount of functionality we get out of systems, we seem to have to write a lot of code. Alan Kay tries to give the viewer a sense of the vast scale of how much code we’re writing, and then asks the open question of whether that’s how much code we really need to write to get that much functionality. There’s good evidence it isn’t, including both research Kay has done at the Viewpoints Research Institute, along with a biological metaphor showing how the machines we build don’t have the necessary capability to let us easily modify them without understanding the details of their every inner component.

Epistemology - This is the talk where Kay gives his famous ‘perspective is worth 80 IQ points’ quip. Kay discusses how the sheer genius of someone like Leonardo merely allowed him to dream about vehicles he couldn’t build without a working engine. He then goes into a short analysis of how many new useful resources exist in software when it comes to outlook, versus the scant progress that has been made in utilizing them. He compares the situation with computer science to that of music, where only a tiny fraction of a fraction of all possible musical theory is actually used to produce popular music.

Complexity versus Complication - Kay theorizes that there is a complexity curve and a complication curve, and that the curve for complication (i.e, inefficient modeling) grows much faster than the curve for complexity. To the point where the right abstraction is the difference between humans being able and unable to manage a system.

Tinkering versus Mathematics versus Science versus Engineering - Kay provides a taxonomy of human artifact production, which distinguishes between several categories of endeavor that are not usually well defined. For example, he points out that mathematics had to be invented, and before it people used giant books of known tested values and binary search to build large structures.

T-Shirt Science - Kay provides a notion that science is about taking a complex idea and boiling it down to its most basic components, so that the ideas which can produce a bridge fit on a t-shirt. He provides the first Lisp interpreter as an example in this class, and hypothesizes that more work in this vein is necessary for Computer Science to progress as a field.

Demonstrations - Finally Kay demonstrates a few software products developed by his team that try to utilize the principles outlined in the rest of the talk.