Bjorn Westergard: “Do Computers Follow Rules Once Followed by Workers?”

I’m extremely excited to call attention to this new Think Piece in the IEEE Annals by Björn Westergard, “Do Computers Follow Rules Once Followed by Workers?“. Here’s a sample:

If the “exhaustiveness” of work rules is simply defined as their susceptibility to “codification,” there is no distinction between explanans and explanandum, and the ALM theory is explanatorily vacuous (as would be the case if Susskind’s “routinizibility” superseded the ALM theorists’ “routineness,” as he proposes).

The ALM theorists, on the other hand, require but do not provide a criterion of specificity (“exhaustiveness”) distinct from the aforementioned susceptibility that is applicable to rules for (e.g.) identifying bird species and multiplying numbers alike. But as sociologist Kjeld Schmidt points out, “the criterion of adequate specification of skilled performance is surely whether the specification serves the purpose for which it is given.” Pace Autor, experienced chefs can provide rules specific enough to guide apprentices in cracking eggs (“no, no, strike the flattest part of the egg, like this!”). It is a mistake to think that “even if an account is satisfactory to the practitioners (masters and apprentices alike) for their practical purposes, something can be construed as unsaid.” An engineer writing a program for an egg-cracking robot might define the “flattest” part of the egg in terms of the second derivative of a curve fit to a raster image of the egg, etc.—but this is not a more precise statement of the chef’s rule. It is different rule for a different (but related) purpose (writing software to crack eggs) (cf. Shanker on algorithms).

A couple of years ago, Björn raised the issue in SIGCIS of the links between packet switching broadly understood (packetization, routing, etc.) and the labor practices from the telegraph system. Luckily for us, he agreed to write it up in the Annals. This piece is one outcome of his line of reasoning, a broader historiographical/theoretical issue that should be addressed in order to make sense of his first question about packet switching.

About a year after Björn’s question re: packet switching, and certainly influenced by it, I gave a SIGCIS talk on human routing algorithms in the telegraph system (‘when routers were women’), in an effort to historicize one of my upcoming book chapters. Talking with him after, I quickly realized that in my research I had only half caught up to his own understanding and research on the matter. It’s likely that my ability to frame the question that way had been influenced by seeing his reasoning along the way. Talking with him after, his take on this issue is transformative in terms of our historiography of networks and certainly my own understanding of them.