Insufficient forgetting

As part of an ongoing dive into the motion inherent in different types of systems—organic, machine, socio-technical—I’ve been reading the two volume Complexity, Entropy & the Physics of Information. Based on the proceedings of a 1989 Santa Fe Institute workshop, the text “explores the connections between quantum and classical physics, information and its transfer, computation, and their significance for the formulation of physical theories.”

One particular section from a paper by the volume’s editor, Wojciech H. Zurek, stuck in my mind. The full paper covered lofty concepts: algorithmic information content, the Church-Turing thesis, physical entropy and Maxwell’s demon. Yet it was the following passage which sunk an especially devious barb.

“The ability of living organisms to perform measurements and “profit” by exploiting their outcomes can be analysed in the algorithmic terms of the above discussion. Measurements decrease ignorance about the specific state of the system, but increase the size of the record necessary to encode the acquired information.”

Place this observation in the context of wrangling a human collective, and a question emerges: is the decrease in ignorance a measure provides worth the increase in the size of the record?

Conceptually, characterise ignorance as the absence of understanding of a system’s state and trajectory, and record size as the cost of capturing, comprehending and communicating available understanding. Taking a given measurement yields three possible outcomes:

  • The decrease in ignorance is superior the increase to the record size
  • The decrease in ignorance in equivalent to the increase in the record size
  • The decrease in ignorance is inferior to the increase in the record size

Naturally, all collectives want to maximise the instances of first outcome and avoid the second and third. However, what makes the above passage and the above framing of the question interesting is the possibility of forgetting. The ability to slash the record itself.

The lifecycle of a collective tends towards complexity, and as they become more complex the chances of using a measure to decrease ignorance at a rate superior to increases in their record size start to shrink; think law of diminishing returns plus island of knowledge dynamics. Formulated as a heuristic—and assuming a collective actually wishes to persist and extend its lifecycle—young collectives should prioritise remembrance; old collectives should try to forget.

Extended to a grander, more law-like (and likely erroneous) formulation: there exists a threshold past which further gains in collective complexity are impossible without memory loss. Death is, through this lens, insufficient forgetting. Organisms pass due to the accumulation of entropy, because of their inability to escape their encoded past; institutions, if they manage to have persistence beyond the mean, must inevitably rewrite their lore and forget where they come from.

Speaking of forgetting: Sonya Mann asked me, at the tail end of last year, to pick up the tinker’s arc within her Wanderverse project. I did as asked and and the first part (of three in total) is now published: Elka’s Welcome. Go check it out.