Unplanned Obsolescence
My UtahJS talk, "Building the Hundred-Year Web Service", was put online this week! It's about how to build software infrastructure that lasts a very long time.
If you're not a software engineer (very unlikely that you're reading this blog if so, but it's possible) the first 11 minutes of the talk are non-technical, and have some insights about the future of the internet that I hope to expand on soon, in other contexts. So it's worth watching up until that point, before I start explaining to web developers how I think they should be making web services.
This is my first stab at publicly articulating something that has been on my mind for a while: is it possible to build software infrastructure with longevity and maintenance characteristics that rival those of physical infrastructure? A lot of people think this is basically impossible—our computational capabilities are increasing too rapidly, and our software is changing too frequently, that the only way to keep up is a constant and significant expenditure of resources. I disagree.
In the talk, I analogize this moment in software to the building of the Williamsburg Bridge. The Williamsburg Bridge was conceived in the 1800s, and built with horse-drawn carriage in mind, yet we've been able to adapt that structure to the needs of a 21st-century city (well, sort of; it should have more bike lanes, but that's a political problem). Maintaining that bridge today, we make use of the work of engineers who could not possibly have imagined what we're using it for; the invention of the automobile did not require us to "write" the bridge from scratch.
