(Halt and catch fire, Season 3, Episode 3)
We are in the late 80s, at a time when the commercial Internet had yet to be born out of the ARPANET and the NSFNET. The setting is a conference room at a small start-up in Silicon Valley that runs a bulletin board system, whereby users can connect via dial-up to exchange messages and trade goods (all the while being represented as sprites seemingly inspired by Maniac Mansion).
The managers of the start-up are discussing how to improve user experience by finding the right compromise between processing at the users’ computers and at the company’s servers. The ensuing conversation should resonate with today’s engineers and researchers working on the optimal functional allocation between edge and cloud of 5G cellular systems:
— “Okay, up next, how are we doing on the speed of the background graphics? Well, we’ll never get under half a second at 2,400 baud. We’re gonna need to use Huffman, or even better, Lempel-Ziv compressions so we’re not sending all the bits through.”
— “Okay, since when are we doing a graduate seminar on information theory? Our guys can’t handle that. No, no, no. That’s smart not sending all the bits through. We preload the most common backgrounds on the diskettes users already have and just send the catalog numbers. Okay, so just send the index to the scene.”
— “Okay, that’s good. I see that. That’s good, right? No complex coding. Well, you’ll have a limited set of images and the user will get tired of waiting for the same-old same-old, but, yeah, it’s great if that’s what you guys want to do. Great.“