Lowering the Barrier

Making programming (everything) more accessible

02 Sep 2020



It’s always tempting to have grand thoughts on a long drive. The world seems so vast! You have nothing else to do! Same thing on a long flight, especially if you touch down somewhere completely different from where you started.

The biggest question I have after driving a thousand miles from Illinois to Massachusetts is: why is it so hard to spot any gas stations from the highway while driving by Buffalo? They’re out there, presumably. You could find one on Google Maps, but that’s fraught. What happens if you get off and are slung on to another highway that takes you to the Canadian border? West of Illinois on I-80 you don’t find full-service rest areas every thirty to forty miles. You simply have gas stations and truck stops that put up signs of varying sizes, some you can see from space and others you see in time to cut across both lanes of traffic. I have not yet found a way to relate the highway signs to the quality of the bathrooms but that sounds like a problem machine learning can solve.

We drove by a lot of night construction in Massachusetts. Handling network traffic is different than car traffic but close enough for me to be curious. What’s the tradeoff of doing work on weekday nights versus in the daytime? Sure, you’ll have long backups in the day, but are construction crews more or less efficient at night? And I’ve seen some bad backups at night too (on our way into Illinois there was a long backup past O’Hare). Do Google and Apple share traffic data with the folks making these decisions? From the outside routing car traffic seems harder; cars have humans in them who will get impatient and choose alternate routes if they can. They’ll also probably get into more accidents in start-and-stop traffic, much like high request latency correlates with high error rates.