Notes on Alan Kay's "Power of Simplicity" (2015)

1st April 2020 - 12:57:15 - 6483 characters

From Alan Kay's 2015 talk on the Power of Simplicity, he considers tactics to avoid becoming stuck in a convoluted web of complexity.

We begin by considering how we treat ideas. Are they akin to matter, unable to occupy the same space? Or are they akin to light, additive and allowing for arbitrary overlap? Are the ideas themselves things or are they processes?

Kay notes that most corporations, and universities as they follow the lead of corporations, generally prevent progress as they solidify existing practices that don't allow for disruptive innovation. The legacy codebases of millions of lines of code ossify the approach and flexibility of any corporation, dooming future work to follow their destructively complex quagmire of a process.

Kay exemplified this by considering the history of planetary motion, where even Kepler was initially constrained to circles by the Copernican model before he crafted Kepler's laws of planetary motion with ellipses instead of circles. Many of the previous systems had become ever more complicated trying to adjust the underlying incorrect idea to reality with ever more problematic fixes.

You get simplicity by finding a slightly more sophisticated building block to build your theories out of. It's when you go for a simple building block that anybody can understand through common sense, that's when you screw yourself right and left, as it just might not be able to ramify through the degrees of freedom and scaling that you'll have to go through. It's this inability to fix the building blocks that is one of the largest problems that computing has today.

Traditional companies are devoted to their A task, what they think they're about, but most don't have a good B process where they look at the A tasks and make them more efficient. Next to no companies have a C process which questions the tasks. "Are our goals still reasonable? Are our processes still reasonable?"

Kay considers the phrase "most human learning is remembering in context" and then brings in the analogy of the Grand Canyon. When at the bottom of the Grand Canyon you're blinded to reality - you don't consider climbing out of it, you consider moving along within it. Like an ant we're now locked into a two dimensional world without realizing it. Every once in a while we look up, see the blue sky, and realize we can escape from this world, realizing there are far more possibilities open to us. You'll frequently look mad until that new blue sky is established as a norm but that blue sky nonetheless brings you possibility.

The most difficult thing about the age of invention - invention is relatively easy with the right type of funding as the most important thing is pulling other people into your new world - but you must also realize even the new expanse, this blue sky above the Grand Canyon that you discover, is a gulley itself.

Kay paraphrased: "Anyone can build a dog house ... but to build the Superdome isn't just scaling that up 100 times."

Most ideas don't scale well. They're all the result of incremental problem solving. Xerox PARC tackled this, with limited resources, by considering the primitives of Objects, Interlinking (Internet), and Virtualization.

You need to solve a context, not just solve a problem.

Few pieces of engineering can scale to millions, billions, or trillions of components. Biology, with hundreds of trillions of cells, appears to be a still unique achievement. The Internet may be one of the only engineering marvels that gets even close to this concept. You can't scale and have centralized control.

The Internet does not have a center and it has grown by almost ten orders of magnitude by now. Your software breaks all the time yet the Internet is never broken. It has replaced all of its atoms and all of its bits at least twice since 1969. Your software could be like that. Your software could just be running eternally. Everything can be built out of a single type of entity that has functionality inside, provides services outside, and has something like a cell boundary on it.

Xerox PARC was the home for invention - the personal computer, GUI, WYSIWYG, object oriented programming, the laser printer, PostScript, ethernet, peer-to-peer / client-server networking, and about half the Internet. Twenty five researchers did all of this. The scorecard:

  • 9 1/2 inventions
  • 25 researchers for approximately 5 years
  • Equivalent of $12 million per year in today's dollars
  • Produced $30 trillion or more in innovation

The GUI was likely the most important innovation as it was the enabler for billions outside of technology to become part of the technological revolution.

For Xerox PARC they used ten years of context and vision to power five year invention horizons. Those five year invention horizons were followed by five year innovation horizons which were responsible for delivering the transformation to the marketplace. At Apple, if the wind was blowing right, each of these five year processes would take three years.

The visions for Xerox PARC were also powered by asking "It would be ridiculous if we didn't have ... in 30 years" and then skating to where the puck will be (Wayne Gretzky) within those invention horizons. With the right idea and right vision you can "buy your way into the future". With money but without optimization you can afford to do many experiments without having to optimize. If you can optimize however you can make "far future" apps. You can bring the future to the present without waiting for all of the technological world to turn.

Asking what the right threshold is for your project is an absolute necessity. Incremental improvement means nothing if that incremental improvement will never reach the threshold to change the underlying task or process. "Better" and "Perfect" are the two enemies of "What is Actually Needed". Kay names the Thacker / Ingalls "sweet spot" as picking a point just over that threshold which is qualitatively better than the accumulation of all the existing incremental progress. From there you identify what the new possibility space is, the new blue sky over your Grand Canyon, that you can now fully exploit. This enables you to think about the situation you're in better and not be overwhelmed with it.