Suboptimal Revolution

In Defense of Inefficiencies

Have I ever mentioned how happy I am that anyone reads any of this? Well, I am. Thanks.

Lately I’ve been thinking about human “depth”. By depth I mean that sense of a “whole world” contained within an expression—a world varied and arborescent, transformative and rhizomatic, full of rich topography: ingresses and recesses, peaks and valleys, dark and light, moments replete with digressions and recursions or stark and baring only the pithy bones of truth, a vibrant network of difference and notional relation. You know, depth…and how it is constructed.

And here’s what I’ve come to:

Human depth is generally a coincidence of suboptimal processes.

Inefficiencies, limitations, and ignorances which necessitate the erection of the unnecessary complexities we enjoy as “depth”.

I call this process of suboptimal construction depth scaffolding.

So, again, depth scaffolding is the idea is that human depth is constructed via suboptimal processes.

I reverse engineered this concept by considering the effects of optimization…

Optimization homogenizes human productivity.

Here we have that sort of obviousness that makes itself mostly invisible:

The optimal state of any optimization is inertia—behavioral sameness—that state which requires no further optimization. Optimization intends to minimize deviations from efficiency—and therein lies the crux. That we even imagine it is possible to optimize a product for consumption relays a tacit understanding that synaptic shortcuts to human consumption exist—that the consumer can be figured out. Likewise, that we would imagine we could optimize our own lives (the various “life hacks”) means we think we can be figured out to some optimal point of behavioral ideal.

Now, if we were as atomically idiosyncratic as we’d like to imagine, optimizing ourselves might preserve (or even enhance) human difference. But instead, the more we are optimized by the social algorithm, the more our pluralistic fantasy vanishes. By optimizing the synaptic shortcut to our consumptions, consumer algorithms have revealed human desire to be a flatland of sameness, shaving our mountainous vision of human difference down to a few plateaus of underlying uniformity.

And it is this underlying uniformity which accelerates optimization’s pipeline to homogenization in the form of another buzzword of virtue, democratization—and I say this as a guy who built my own direct democracy app.

There is a causal link between democratization and homogenization.

Democratization has an ironic relation to dynamic variety. The more democratization of a process, the less dynamic the variety within the process’ output. In short, the more democratization, the more homogenization. This etiology is again most obvious in the world of the arts—wherein the democratization of the creative process leads directly to the homogenization of the creative product.

You know those boomers who complain that music today all sounds the same? They are more right than wrong. Check out this study of 464,411 recordings across all genres of music from 1955-2010. Using algorithms it detected harmonic complexity, timbral variety, and dynamic range—and found that music has trended toward homogenization.

From the results of the study:

“These point towards less variety in pitch transitions, towards a consistent homogenization of the timbral palette and towards louder and, in the end, potentially poorer volume dynamics.”

While the study doesn’t get into the cause, it’s easy to piece together:

Although the democratization of the tools of creation would seem to be something that would increase variety, not lessen it (after all, the more unique individuals there are making music, the more unique music there would be, right?) the opposite is true: Technological optimization of the tools of production flattens any given domain.

A refresher on how this happens:

1. Optimizations across a given domain homogenize the outputs of that domain, as optimizing a process eradicates the processual inefficiencies and incidental deviations that produced domain variety in the first place. The technological democratization of tools of creation homogenize the products of that creation.

2. When you technologically optimize a given process, you make that process require less effort and less know-how while delivering the same effect. As people are more likely to engage in processes that require less know-how and less effort, optimization leads to the “democratization” of technological tools.

3. Humans copy each other—mimetic drive and status anxiety naturally lead to a flood of aesthetic imitation. The more humans engage in a given aesthetic, the more that aesthetic is likely to flatten. (Think “watered down” mainstream versus the potent asymmetry of the avant-garde.)

4. Mere democratization alone, without the aid of technology, leads to homogenization.

I’ll assume for a moment that the first 3 points are obvious to my readers and that the 4th needs more convincing... Or does it?

We are probably all familiar with the phrase “decision by committee”, and likewise all understand why it is not a phrase brimming with positive connotation—decisions by committee not only slow everything down, but water everything down.

Shazam! Bland!

Pop music today really is the population’s music. From major label A&Ring to terrestrial radio top 40s, the music business today relies almost entirely on user-driven metrics. Gone are the days of the intrepid A&R trusting their gut or taking a chance on an unproven act.

One jarringly circular example is the radio programmer’s reliance on the music identification app Shazam—which can tell you what’s being played on the radio—to determine what should be played on the radio. Think this might result in the same songs being played over and over? Ya.

As one writer puts it:

The big problem though is that it’s a self-fulfilling cycle—only popular songs get play, so they’re the ones that get popular, and then the relationship between big labels and big stations continues to devour itself like a Billboard Top 40 ouroboros.

It is easy to blame this sort of thing on the “powers that be”, but let’s continue to remember that this sort of market research is informed in the first place by discovering synaptic shortcuts to our enjoyment…

To illustrate this further, consider the effects of taste testing, audience testing, and focus groups on human production.

Food product optimization.

In what is literally called “food product optimization”, researchers used taste testing to optimize food products for what the industry refers to as “the bliss point”.

Via Wikipedia:

The bliss point is the amount of an ingredient such as salt, sugar or fat which optimizes deliciousness (in the formulation of food products). Pioneering work on the bliss point was carried out by American market researcher and psychophysicist Howard Moskowitz, known for his successful work in product creation and optimization for foods ranging from spaghetti sauce to soft drinks.

Combinations of sugar, fat, and salt act synergistically, and are more rewarding than any one alone. In food product optimization, the goal is to include two or three of these nutrients at their bliss point.

Of course, this process of optimization is industry-wide, resulting in the same general salt/fat/sugar ratios across all food products, ultimately homogenizing the landscape of taste itself.

While we can blame market researchers like Howard Moskowitz, it is important to remember that these researchers are not foisting these tastes upon us, but exploiting the very synaptic shortcuts to our enjoyment that we willingly expose via taste testing.

In pure operational terms, taste testing is a process in which the company cedes some of the decision-making to the people—a literal democratization of the process of production. This democratization entails, first, the optimization of the food, and second, the homogenization of food generally.

Happy endings.

Audience testing of movies began in the late 1920s. Just as with taste-testing, audience testing should be viewed as a democratization of the creative process of cinema. And just as with taste-testing, it is a process which both optimizes the product and homogenizes the sector.

As a film composer, this one hits close. Most things I’ve scored were more interesting, idiosyncratic, even groundbreaking, before audience testing. After re-cutting based on audience notes, the art comes back less interesting, less groundbreaking, less idiosyncratic—more average. And there is an obvious, rather mathematical reason for this—the more opinions one takes into consideration, the closer one gets to an average of all opinions.

As this process is adopted across a domain, over time one product more or less comes to represent all products—where all products are “the average”.

Here is a short list of movies that changed dramatically after audience testing (most such changes are protected by NDA’s, so what you see on this list is not even the tip of the iceberg). Most changes to movies due to audience testing are, of course, changes revolving around making endings less dark, more happy.

Driving the optimization of movies today is the hedge-fund-ification of Hollywood (many studios went public in the 80s). In a Wall Street effort to avert risk, the trend toward sequels and superhero movies has exploded, while character-driven stories are largely deprived of the big funding. By way of example compare the top 10 of 1972 and 2022:

1972: 1. The Godfather 2. Poseidon Adventure 3. What’s Up Doc? 4. Deliverance 5. Deep Throat 6. Jeremiah Johnson 7. Cabaret 8. The Getaway 9. Last Tango In Paris 10. Lady Sings The Blues.

2022: 1. Top Gun: Maverick 2. Black Panther, Wakanda Forever 3. Dr. Strange In The Multiverse of Madness 4. Avatar: the Way Of the Water 5. Jurassic Wold: Dominion 6. Minions: Rise of Gru 7. The Batman 8. Thor: Love and Thunder 9. Spiderman: No Way Home 10. Sonic the Hedgehog 2.

Literally every movie in the top 10 of 2022 was both an action movie and a sequel—and every one of them had myriad audience test screenings (links to evidence of test screenings on names above). Meanwhile, the top 10 of 1972 contain some of the most daring, character-driven movies in history, and even a pornography without proper distribution in Deep Throat.

And again, before we go thinking that its all the man’s fault, lets take a look in the mirror and acknowledge our drive toward the inert inertia of the depth-less…

As shoddy-but-fun evidence, consider the so-called “soap opera effect”. It occurs when a film shot at 24 frames per second is shown on an HD television set to refresh the frame 120 times per second. As a result, the TV generates fake frames and interpolates them between the real frames, creating a strange, synthetic effect to the motion of the film. I hate it. Film directors lament it. It looks absolutely atrocious… But guess what? Many users prefer it. Check out the youtube comments on this side-by-side comparison.

Deep Enjoyment

In the name of inefficiency and deviation, here is a relevant tangent… (and then I’ll present some happy hippy hope stuff, promise).

As discussed, homogeneity creates a flatland of non-dynamic topography within any given domain—no valleys, no peaks, just plateaus and plains. But what is the human compulsion toward this flatness? After all, a non-dynamical space is a dead space. And yet the algorithms of our enjoyment are increasingly exposing our compulsivity towards the homogenized, dead experience. The more our synaptic paths to enjoyment are hacked, the more we are revealed to desire the non-dynamical, the dead.

Iterating on psychoanalyst and infamous obscurantist Jacques Lacan, I call this phenomena pop jouissance—not merely “enjoyment with a deadly reference”, per Lacan, but straight up enjoyment of what is dead.

Lacan aside, why would we enjoy a non-dynamical state? Really. And why would the democratization of things trend toward non-dynamism? Toward inert inertia? Toward…death?

I may be inert but I have inertia.

Inertia is changeless; inertia is thereby eternalizing; inertia is thereby escape from the mortal condition. This is the premise of the death drive: a drive to return to that static, worry-free state of changeless inertia from whence each came. In very practical and physical terms, that inert place from whence we came is the womb. The womb-state, floating in the self-same liquid of the amniotic sack, all things in homeostatic equilibrium, precedes what Freud would call the "trauma” of the interior/exterior split of birth—a trauma which itself marks the awakening of the drive to re-produce that inert state.

And so we find ourselves compulsively reproducing this or that inert state, this or that optimization, this or that homogeneity.

Suboptimal revolution.

1. Natural Asymmetry

If there is any pink elephant hanging around this piece it’s probably this: if democratization of the creative process yields homogenization, then is the solution a kind of asymmetrical elitism?

Yes and no. I’m against active gatekeeping and against synthetic societal asymmetries. But only the naive or homicidal are against passive gatekeeping and natural social asymmetries.

It just so happens that where processes are not technologically optimized—when a process requires sufficient effort and specialized knowledge—a passive gatekeeping naturally occurs, and social asymmetries incidental to the required proficiencies of the domain naturally arise. Thus suboptimal “inefficiencies” (particularly technological ones) produce the natural occlusions associated with effort, time, and degrees of difficulty (such natural occlusions produce the only sort of “elitism” worth defending).

Apropos of the above, check my footnote on the call to “write less fancy”, and its relation to Gen Z patois.¹

2. Suboptimal Tech

As far as I can tell, technological inefficiencies and cognitive limitations are the mines from which we have produced everything we think of as cultural depth. The erroneously explanatory myths and inventions of religions, the porous and contradictory texts of philosophers, the stretching wildernesses of art, the colorful speculations, the confounding poetry—all the natural result of a clumsiness born of primordial blindness. A blindness which sees to it that even the most deft pen is a perfectly clumsy tool, sending brilliant minds groping, exploring, imagining, inventing.

This goes for everything from food to art, horticulture to sailing.

Even when technological innovations move a domain forward, the depth comes less from the optimal aspects of the technology, and more from the suboptimal aspects of it. Random examples: the grain of film, the distortion of the microphone, the saturation of tape, the pixelation of 8-bit, and so on; the roughshod bronze cut dies used to extrude premium pasta; the supreme taste of the inefficient brick oven; the leeching, inefficient wood barrels used to age liquor; the sun-dried coffee; the weak edges of spray paint; the brush marks left by paint brushes, the irregular stamping of the typewriter, the occasional blots of the quill, the roughly hewn look of parchment, the stone-ground corn and stone-milled flour; the physical book (as opposed to the optimal e-book), and on and on…

I’m not declaring war on technology. Rather I’d like to think of this as techno-ontological design motivation: to build and use technology with the intension to erect depth scaffolding; designed for suboptimal behavior, designed to explore.

I want sub-optimization built in to the tech.

Currently tech doesn’t really do that. You can increase the temperature parameters on LLMs like ChatGPT, and you can find suboptimal workarounds. For instance, instead of using ChatGPT to optimize my ideas, I have been using it to sub-optimize them by asking it to point out my common, sure-fire tropes (optimizations) and offer novel ones (sub-optimizations) using high temperature parameters. (Novel tropes are going to have less certain effects on an audience). I’ve tried this out with ChatGPT4 and it works OK—I call it liminal editing.

But we can do much better, er, worse. ChatGPT itself is built purely to optimize. The workarounds are very limited, and this describes tech generally—making suboptimal interactions with tech increasingly difficult.

Yes, we can create popular algorithmic interfaces that flag lazy optimizations and prompt US to be more interesting. Yes, we can and should sub-optimize our approaches to everything from child care (take away the iPad) to cooking dinner (throw out the microwave).

Dear suboptimalists…

I can already hear the comments balking at the notion of embracing the “suboptimal”.

“Why not “self-optimal!”

Clever! But there are two reasons why not. First, because whatever is self-optimal self-homogenizes, self-flattens. Second, because we need no help with self-flattening. We have been spending our lives, since birth, attempting to self-flatten, remember, to return to the inert state of womb-ish inertia. What we seem to need help with is the courage to kick ourselves out of the inert state to continue to live dynamic lives.

Collapsing into a self-same uniformity (whether its you or your society) is inevitable. It is the third and necessary part of the dialectic (sublation). It can also feel amazing. That sense of inertia, sameness, changelessness; that eternalized, return to womb-ishness, that suspension of transience, that total equilibrium. The problem comes when we refuse to leave that state. Refusal of change is the premise of fundamentalism, yes, but also the denial of death and the negation of vitality.

I think it’s important to incorporate the self-confrontational language of sub-optimization (especially for those of us attached to purely self-affirming language), to become deft sub-optimalists, able to abandon the optimized, homogenized path at will. 

Well, that’s just about it for now…

Thanks for reading.

1. Calls to “write less fancy” bore me. Not because I don’t appreciate the use of communicating simply or broadly, nor even because some notions are actually complex and nuanced (believe it or not) or (gasp!) even technical, but because complex writing narrows not the scope of the audience, but the scope of the willing audience—and it is precisely that it creates an asymmetry based on will that it is of fundamental service to the dynamic topography of (an otherwise flat) society. And this doesn’t just happen with fancy pants language. Go spend some time on Gen Z Instagram pages and cop the greedy bloom of newfangled language developing by the hour, and you will feel immediately on the ancient outside of a youthful inside thing—and that, too, contributes.