(2024) really. Saw this years ago.

Xerography was once a very touchy process. It took insane complexity and skilled technicians to make xerography go in the selenium drum era. The basic idea of xerography is that you charge up a photosensitive surface which is discharged by light. So the original is projected onto a selenium drum, and the light areas are discharged. The remaining charged areas will attract toner, which is then rolled onto paper. The final step is heat-fusing the toner to the paper. Then the drum is cleaned and recharged for the next page.

It's amazing that this works at all. It barely did in the selenium drum era, which is what this article is about. Selenium is very soft, and not a very good photoconductor. Around 1990, selenium was replaced with multi-layer organic photoconductors, and the process became much more robust. That resulted in smaller, cheaper, and less maintenance-intensive xerographic copiers and printers. Now, anybody can change the toner cartridge.

That's why Xerox needed so much technician know-how. The selenium drum machines needed a lot of tweaking. Sort of like 1950s cars or TVs, which needed lots of screwdriver adjustments. As the technology matured, it became idiot-resistant, or at least field-replaceable without adjustments.

This is how technology progresses, from fussy to robust. At the fussy stage, you need people who really understand how it works and how it breaks. At the robust stage, you have minimally trained part changers. They're viewed as under-qualified by the old guard, and as cheaper by management. Watch for this pattern as new technologies are adopted. It's happening to programming right now.

The amount of knowledge in the heads of the engineers and socialized through storytelling with each other is immense, and immensely valuable. Really smart customers know this too, and some would refuse to reorder or renew their maintenance contract unless they got guarantees that “their” engineer would continue to be assigned to them.

I once had a shouting match with senior management of an East Coast company I worked for because they wouldn’t do what was necessary to avoid an experienced Sales Engineer from leaving. To them these were “headcounts” you just replaced when they left, zero understanding of the years of accumulated undocumented knowledge that would walk out the door. Literally the difference between successful deployments and catastrophes that required expensive fixes and escalations. MBA thinking at its finest.

Excelent read. Xerox field technicians + Xerox PARC, a case of knowledge driven comunities, complete with it's fight with Xerox corporate.

I've read an earlier draft and loved it. Every chapter of that book is excellent.

"He revealed Kaplan’s and Newell’s pedigrees, and the Xerox execs sat speechless.” When I first got started in UX/UI, a future XPARC staffer told me to read some of both of their papers. I loosely saw the connection at first, but as my work continued, the more it directly influenced how I spoke, how I taught, and eventually and reluctantly, how I thought. I said in the previous post, that this should be a static post, in a Best of Hacker News, and I still truly believe it.

We are doing here, and most brilliantly, doing what Xerox technicians did back then, and that the scholars that worked in the infinite corridor at MIT have done, and do now.

Thank you!