Accounting for entrenchment to aide progress
When we say that natural selection is a ‘miserly accountant’, as Richard Dawkins put it, we frame evolution as penalizing any wrong move. But I see Dawkins as saying that natural selection is a accountant.
Not everything is an adaptation, meaning that every change produced by natural selection doesn’t necessarily have a function. Divergence comes from copying errors — some prove costly and fatal, some prove beneficial and increase survival odds, and some don’t seem to do anything. A ‘miserly accountant’ should see the last group as wasteful, as they don’t provide any obvious benefit. A thorough accountant, however, sees them as biological QWERTY keyboards.
The QWERTY keyboard emerged in the nineteenth century as a solution to the problem of mechanical breakdowns in typewriters, caused by hammers crashing into each other. In the early 20th century, August Dvorak invented a keyboard layout that has been found to be up to 60% more efficient, and while the Dvorak layout is the most well-known alternative to the QWERTY layout, a century later it is still an alternative at best. The more recent Colemak layout claims to enable a typing speed twice as fast as QWERTY. Why haven’t these more efficient layouts won out against the ‘less fit’ QWERTY layout?
The QWERTY keyboard is entrenched. A miserly accountant sees a great deal of lost productivity, while a thorough accountant sees the cost of switching to be counter-productive. QWERTY is less functional, but has an airtight reason for its continued use — the cost of switching or getting rid of it exceeds the cost it imposes. Improvements in speech-to-text technology lower the cost it imposes and make the cost of switching higher.
If we want to account for entrenchment, we have to count the carrying costs.
Robin Hanson notes that opinions entrench themselves by mixing with larger systems of thought.
We see entrenchment also all over our human systems. For example, at my university the faculty is divided into disciplines, the curricula into classes, and classes into assignments in ways that once made sense, but now mostly reflect inertia. Our legal system accumulates details that become precedents that many rely on, and which become hard to change. As our software systems accrue features, they get fragile and harder to change. And so on.
What we see as inertia or drift could be a case of negligible carrying costs. University structures, the legal system, and computer software each have their own brand of QWERTY keyboards — things that seem wasteful, but have a switching cost higher than an imposed cost and thus are the more optimal solution.
If we want to address a perceived inefficiency, it seems the best place to start is by being more thorough in our accounting.
One example of this is what Eliezer Yudkowsky calls ‘ Making Beliefs Pay Rent ‘. We build networks of beliefs that are connected only to each other — these are ‘floating’ beliefs, untethered to any real experience. They are not so costly that they cause you major problems (otherwise you would target them for revision), but the cost of rooting them out and removing them is also too high. In making them ‘pay rent’, you consider whether or not they aide you in making more precise predictions about what you will experience. This makes the accounting more accurate and you can systematically eliminate your weakest beliefs, just like natural selection eliminates the weakest organisms. In this way you pursue optimal rationality based on beliefs that are more fit (cost effective).
What other entrenched things can we make ‘pay rent’? Mancur Olson tells us that special interest groups get entrenched (“like barnacles on a ship’s hull”) during times of plenty and then go on the defensive during times of scarcity. In times of plenty, the economic pie gets bigger and the ‘drag’ caused by an entrenched group is minimal. The cost of removal is more expensive than the cost the group imposes. Benefits are concentrated, while costs are diffused and muted by tax exemption. Groups can persist on what appears to be drift or inertia, when in fact they benefit from sloppy accounting.
As a system or environment gets more complex, so too does the accounting. Our response to complexity is usually to ‘scale up’ — to add rules, regulations, procedures, etc. We might memorize a heuristic or enact legislation that contains additional rules, and this is usually detrimental to the work of accounting.
In his 1991 book The Rhetoric of Reaction, the political economist Albert O. Hirshman identified common rhetorical tropes used to defend the status quo. One trope is the ‘perversity thesis’, which says that well-intentioned rules and regulations ultimately exacerbate the problems they were designed to solve. This is true in cases where additions either overlook the importance of accounting or are designed to make accounting more difficult. The problem is not that government is incapable of producing efficient rules and regulations, but that what it sees as ‘ irrationality that can be fixed with legibility ‘ is actually ‘accounting that hasn’t been thoroughly investigated’.
Despite our fetish for procedures like enacting reform or ‘enhancing’ regulation, processes, transparency and accountability, we tend to stay away from actual accounting. Perhaps because it’s reminiscent of maintenance, or because numbers are boring, or it’s too complicated. Maybe it’s easier just to add — adaptations are cooler than mutations that came about just because they could. Memorizing heuristics and lists of biases feels more productive than refining beliefs. Addressing societal problems with new reforms, legislative agendas, or innovations is more exciting than refining 130,000 pages of regulations into something more legible or putting all of the variables that go into academic fragmentation into a spreadsheet.
Originally published at https://mattaurilio.com on January 27, 2022.