Why Quasioptimal?

Why something arcane and difficult to pronounce? Quasioptimal is my imperfect solution to an explicit design problem. I spent a few afternoons workshopping it and tossing some other names around, and this is the one that tickled me enough to start recurring.1 I wanted brand name that was (1) a suggestive word or phrase that (2) resonated with me, (3) was general enough to put to any number of evolving uses, and (4) wasn’t already in wide use.

These constraints were informed by some examples I know well from the podcasting world and the blogosphere. My entry point into podcast listening back in high school was Hello Internet, which at one point or another featured an explicit discussion of how they came to their name that sketched some similar considerations above list of design constraints.2 From that show I branched out to Cortex, then to ATP, Upgrade, and Reconcilable Differences. I don’t listen to any of these shows these days, but they formed an indelible part of my late adolescence, and I think back on them with great fondness, as you might an old friend from grade school whom you don’t see much anymore.

The most direct model for the name Quasioptimal came from John Siracusa, a cohost of ATP and Reconcilable Differences, who writes a blog called Hypercritical.3 I think a title like Hypercritcal is great because it gestures at an approach and tone characteristic of John Siracusa without being topically restrictive, leaving him a lot of leeway to write about whatever he’s interested in. It would be weird if the China History Podcast, of which I’m also a listener, started producing shows about the electric vehicle market in Sweden or the pop music scene in Uzbekistan or Lazlo Montgomery’s decision-making process when buying a sofa. But Hypercritical might cover anything from the technical details of Apple products to Destiny music videos to broader musings about the tech industry, and anything else besides. It ties the brand to the voice and approach of the writer, not on the content of their writing.

For a project like mine, the content of which is developing and evolving (slowly, alas), finding my version of Hypercritical that could grow and change with me without losing its relevance or feeling like a burden down the line was crucial.


To spell it out for those who didn’t subject themselves to five years of Latin in high school, quasi- is a concatenation of quam and si that means “as if” or, more loosely, “almost” 4. The -optimal root conjures up the sequence of positive “good” (bonus), comparative “better” (melior), and superlative “best” (optimus). In the highfalutin Latinate register of academia, quasi- gets plopped onto the front of all sorts of words as a qualifier: people speak of “quasi-fraudulent” behavior and “quasi-monopolies” and “quasi stellar radio sources” (quasars).

A quick google shows that it is indeed a word that gets used a fair bit in physics and mathematics, but the first couple pages of show that it is not in popular use. The n-gram data reinforce my impression from the search results: quasioptimal is tossed around as a technical term in academia, but as far as I can tell it hasn’t been claimed elsewhere. Score!5

Reification Frustration: Models, Metaphors, and Theory

To call something quasioptimal is to say that it’s almost as good as the best you can theoretically do, but not quite. The core claim I will be making time and time again, distilled to its barest form, is that close is good enough, and often it is better.

The thrust of the argument runs like this: a theory is a compression of reality, not the real thing. We construct theoretical models to solve problems in the real world, but we are always leaving most of the information behind by design so that we can work our problems in a more tractable space. Broadly speaking, optimizing our behavior according to our models will yield good results so long as the disparity between our model and the real world is not so big in the context we care about.

The trouble I’m claiming we constantly find ourselves running into is that we reify our models: we forget that we’re using a model at all because the model is all that we can see and reason about. Furthermore, for most optimization problems in the real world, we get diminishing returns for our efforts as we approach the optimum. Near the optimum, we’ve extracted most of the insight that the model has to offer, and the divergences between our model of the world and the real world start becoming much more important than what the model says.

In its broadest conception, theory is a set of models, metaphors, concepts, and modes of argument that we construct to stand in place of the infinitely complex world we are faced with. We don’t have the bandwidth or computational power to reason about the world in all its complexity, so we compress it into a more tractable set of objects that we can deal with. This process abstracts away much of the detail that we hope doesn’t matter too much to what we’re trying to do.

This theoretical leap is at least as old as Plato and probably much, much older. The ability to look past our sense data, to perceive and act on abstract patterns, and to nimbly shift among these abstractions is in some ways a defining set of characteristics of general intelligence, at least as humans experience it.

Perhaps the way I’ve phrased it makes it sound a bit academic and, well, theoretical, but the construction and deployment of what I’m calling theoretical models is in fact so pervasive in our everyday lives that we hardly ever recognize that we’re doing it. Walk down the street and you’re modeling the world around you in every instant. The angle of incline or decline of the pavement, the texture of the ground you’re stepping on, that pattern of photons flashing on your retina that you call “tree” or “car” or “house.” All of these collections and groupings are conceptual abstractions from the infinitude of detail and richness we must navigate in the world.

Quasioptimal is my attempt to reorient attention toward the models we deploy in a way that respects what they have to offer without forgetting that they are models—not the real thing—in the process. My fanciful hope is that Quasioptimal, which is itself a collection of concepts and ideas that might pretentiously call itself a metamodel, can withstand and resist its own reification by continuing to call attention to the importance of our conceptual processes.

  1. The closest runner up was something about the concept of meliorization as contrasted with the idea of optimization. Both “words” betray the years of my adolescence I spent nerding out in Latin class. 

  2. Requiescat in Pace. 

  3. To be honest, I haven’t actually read much of Hypercritical; I just really like the name. 

  4. The post facto Latin grammar refresher I consulted: [https://en.wiktionary.org/wiki/quasi-]. 

  5. Looking at this chart, I can help but be fascinated by the clear hump-shaped patterns for “minimum,” “maximum,” and “optimum.” They reach their peak sometime between 1940 and 1960 after steady inclines, plateau for a while, and then decline quite precipitously. Why? Are they being replaced by different sorts of language? Part of me can’t help but associate this with what Brad DeLong calls the “Neoliberal Turn” starting around the late 70s that saw the rise of Reagan in the US and Thatcher in the UK. My pet theory is that the neoliberal economists and their hangers-on did a lot of talking (and in this case writing) about theoretical maxima and minima before they were in power, and, once they were, they found limited utility in the deployment of these concepts in the real world. But all of that cum grano salis—it’s based on nothing but a hump and a hunch.