The Mark Cuban-Daniel Beunza debate from earlier this week deserves further comment. Beunza argues that “behind every seemingly autonomous algo[rithm] there is a person.” Mark Cuban argues that people aren’t so in control of algorithms as Beunza makes them out to be, warning “when algorithms have to interact with each other, the owner controls only their own.”
The debate is worth highlighting, because I find them both right, and because I think what they are talking about applies to more than algorithms and maybe to most codified knowledge in general.
Beunza’s main case is not that algorithms can be reduced to people, but that there is a reflexive interaction between a programmer and a program — “we need to look at the combined “assembly” of humans and machines,” Beunza writes.
Cuban makes a different argument, but it’s no less interactional. Cuban’s focus is less on the person-program interaction,* and more on the interactions between program and program (and program and program) that produce the information environment. In Cuban’s view, a person’s control of an algorithm is limited, because the nature of that algorithm will only be clear in interaction with an environment the person does not control.
Taken together, the arguments paint a picture of a complex model with constant feedback-loops, or reflexivity:
*Cuban doesn’t ignore the role of the person as a fallible programmer, especially when the programmer “acts like” the algorithm exists in a “vacuum.”