Nuance and trade-offs

“You insist that there is something a machine cannot do. If you will tell me precisely what it is that a machine cannot do, then I can always make a machine which will do just that!”

— John Von Neumann

Many aesthetic complaints are about missing nuance. A show about crime scene investigation fails to capture what the job is really like. A fast food burger doesn’t taste like a burger from home.

But the elements that make a particular experience useful on a local scale don’t necessarily make it useful on a global scale. CSI might contain the same basic premise as the real thing — forensic analysis of trace evidence — but it trades veracity for coolness and compelling drama. The writers have to balance the audience’s suspension of disbelief against its attention span. The fast food burger might contain the same beef patty, tomato, lettuce, ketchup, and bun as the home-cooked version, but it trades complexity and freshness — not to mention the one of a kind context of one’s own kitchen and family — for consistency and affordability.

Kitchens make bad assembly lines. The territory doesn’t fit into a glove compartment. To mass produce a representation of reality or a complex sensory experience, some elements have to go. To make chords and maps widely useful, trade-offs are necessary.

One might assume that at any given moment the elements that didn’t scale, can’t scale. But the given trade-off is not necessarily the best one possible. That an element is missing may only be an artifact of the technological moment. For example, the fast food of the 1960s was subject to a very different trade-off window than the fast food of now, both in terms of quality and variety. Furthermore, the tendency of television audiences to tolerate — or even prefer — psychological realism and technical details has steadily risen since the advent of police procedurals.

Knowledge of better trade-offs — ones that hold on to as many desired elements as possible — is not centrally or evenly distributed. When presented with a chord or map, I might notice it lacks what I believe to be an essential element. It feels wrong because I believe the tradeoff was unnecessary. This assumes I know the conditions of the trade — the window the elements must fit through. Here is an opportunity for aesthetic vigilantism.

True, nuances are hard to articulate. But that’s the challenge. In doing so, the shape of the trade-off window must be respected. Specifying it might cause me to retract my complaint, once I see how it isn’t actionable. Or, if it still seems like an oversight, I can use whatever platform available to me to identify what is missing.

Unfortunately, nuance can feel sacred to some people, effectively blocking better trade-offs. The complainant may want others to have the better experience they envision while simultaneously resisting the sacrifices necessary to mass produce it.

Sometimes nuance provides only very marginal value. Consuming it anyway can act as a costly (and entirely symbolic) display of respect. In these cases, trying to efficiently “get the gist” is a sacredness violation. This tendency is evident in the belief that there are essential subtleties in fictional stories (and other types of art, such as poetry) that cannot be summarized or compressed. Thus the tendency to frown on reading the Cliff’s Notes versions. But, somewhat ironically, knowledge of facts and concepts can be obtained from summaries without shame.

What is the ideal result of absorbing every single nicety? One possibility is magical thinking: if the nuances in question are a particularly good illustration of an underrepresented aspect of the human experience (or of an ignored or oppressed group), then submitting to them fully may be considered a form of tribute — regardless of whether anyone who has had that experience or is from that group ever learns that the full submission took place.

Seeing stagnation, part 2: chords

This post is more unfinished than usual. Here’s what I’m wrestling with:

  • Pizza is a good example of a chord. Pizza is comprised of elements that combine appealingly but don’t describe reality, an excellent conveyance for superstimuli like fat and salt, and clearly bounded by its crust.
  • Pizza is an excellent solution to “many simultaneous complex problems.” Pizza facilitates social bonding by being a communal, handheld food, as well as an uncontroversial preference that nearly everyone can converge on. Pizza is also cheap to produce and, unlike many other forms of fast food (McDonald’s, Chinese) is good hot or cold.
  • Yet this and this seem like evidence that nobody knows how to vertically innovate on pizza. In other words, pizza is stagnant. Yet nobody seems bothered.
  • What would it take for pizza to become obsolete? Obsolescence doesn’t mean low quality — most pizza is already pretty bad. Nor does it mean substitutions that only partially cut into pizza’s territory, such as burritos. Obsolescence means a change in available options that causes everybody to collectively realize pizza is something they’ve been settling for, and don’t have to anymore. So they stop consuming it altogether. Obsolescence is automatic elevators instead of elevator operators, or electric light instead of oil lamps.
  • But how does that apply to pizza? The only unambiguous vertical leap I can imagine for pizza is if humans transcend biological form and food-based sustenance becomes optional.
  • Why should anyone care? The use of gimmicks to distract from disappointing quality (such as the pizza box made of pizza) is a phenomenon that occurs in all aesthetic domains. Does it make sense to call it stagnation? I’m not sure.




Seeing stagnation, part 1

In The Great Stagnation, Tyler Cowen argues that certain types of progress have slowed because the low-hanging fruit is gone. For example, the flush toilet led to substantial improvements in quality of life for the first generation to use it, but subsequent generations have gained comparatively less. Toilets that flush faster and use less water don’t measure up to the original leap from the outhouse.

What should we call innovation that feels substantial? According to Peter Thiel:

Vertical or intensive progress means doing new things — going from 0 to 1. Vertical progress is harder to imagine because it requires doing something nobody else has ever done. If you take one typewriter and build 100, you have made horizontal progress. If you have a typewriter and build a word processor, you have made vertical progress.

Vertical progress is what happens when a problem is transcended rather than solved — that is, when an entire solution context is made irrelevant by a form of “superfit.” Stagnation is what happens when this stops happening. In Cowen’s words:

It would make my life a lot better to have a teleportation machine. It makes my life only slightly better to have a larger refrigerator that makes ice in cubed or crushed form.

If it existed, teleportation would revolutionize transportation by unbundling time and distance from travel. Suddenly, concerns about fuel, roads, and collision safety would be optional, freeing millions of man-hours for more important tasks. In comparison, ice choices seem pretty weak.

Admittedly, this shift in perspective feels inappropriate. Whether cubed or crushed, ice that is conveniently dispensed is preferable to ice that must be pried out of trays (which must then be refilled). Comparing dispensed ice to teleportation may seem unrealistically demanding or even ungrateful. Nevertheless, such comparisons are sometimes the best way to express what feels wrong.

If we choose to see it, the same comparative unambitiousness is easy to find in aesthetic experience. Over the next two posts, I will attempt to make it apparent.

Two types of misfit

Imagine a nearly complete jigsaw puzzle. Only one piece is missing, but oddly there are three pieces left to try. The first doesn’t fit. Examining it, you realize it belongs to a different puzzle. Of the two remaining pieces, the first looks like it will fit perfectly, but no luck. Comparing these last two, you find they are nearly identical, except one is a defective copy of the other. You use the non-defective piece to complete the puzzle, leaving only the misfit. This last piece doesn’t fit in its intended spot, or anywhere else. It’s just bad.

A jigsaw puzzle is the ultimate tangible instance of a “solution context”*. The edges of unfilled-in areas provide immediate feedback to the rapid testing of multiple candidate pieces. Different types of misfit are also easy to distinguish. While one might lack an explanation for the defective piece, it’s at least clear that it fails to fit for a different reason than the piece from a different puzzle. It’s also clear that the one of two pieces fits somewhere, while the other does not.

Misfit is easy to detect when it’s as obvious as an ill-fitting puzzle piece, but such obviousness is rare. As the constraints of fit become more complex and abstract, the “edges” of the “gap” in the “puzzle” become increasingly illegible. In many contexts, it can be hard to tell whether misfit has occurred at all, let alone which type.

Consider chairs. To “fit” in a familiar context, a chair must satisfy (at minimum) the following conditions: affordable, easily moved, seat not too high or low off the ground, back support, accommodates human size and weight up to a few standard deviations. Because the solution context of a chair is still pretty concrete, both types of misfit are still fairly obvious. An ornate golden throne fulfills many of the conditions but is not affordable or easily moved. Yet a throne neatly solves the problem of king needing an impressive place to sit and receive subjects — it fits in a different context. A spindly-legged chair made from balsa wood might at first appear to fit, but instead breaks easily, crashing its occupant to the floor. One might argue this chair fits in a comedy context as a breakaway stool for pratfalls, but it’s too obviously flimsy to trick an unsuspecting sitter. It fits nowhere.

Clothing is trickier than chairs. Consider the ironic thrift store t-shirt. Candidate shirts must be broken in but not tattered. They should also fit in the literal sense — baggy can be stylish, but too baggy is bad. More saliently, they should feature text and/or design elements that can be amusingly recontextualized. For example, a shirt commemorating a 1993 insurance company employee picnic, featuring a cartoon hot dog wearing sunglasses and sunning itself. A plain black tee won’t work because it’s devoid of ostentation, yet it’s perfect for bussers and stage crew who need to not stick out — it fits in a different context. But a brand new, oversized shirt with uninteresting colors and patterns, or with bland text that’s impossible to “flip”? These are left on the rack until they finally go wherever thrift store dregs go in the end, unwanted by all.

Finally, a highly abstract problem: the evocation of a subtle (but universal) human experience, specifically the sinking feeling of reality battering against delusional confidence. A good solution should produce a nearly identical experience for every user: cringing relief at not being in the aforementioned position mixed with spontaneous recollection of times when the user was in that position. One famous example is Robert De Niro’s portrayal of Rupert Pupkin in The King of Comedy. Any one of Daffy Duck’s numerous instances of punished hubris might seem to fit, but are too painless. However, they fit in their own context like a mild but still flavorful version of spicy dish. An abstract metal sculpture titled “The Sinking Feeling of Reality Battering Against Delusional Confidence” stretches the definition of “evoke”. The abstract nature of the solution context makes failure plausibly deniable. If one stands to gain socially from claiming the sculpture evokes the target feeling, then there is incentive to lie about one’s reaction. This sort of dodge would be a lot harder to get away with if it were as concrete as a flimsy chair.

Solution contexts can be ambiguous. If you’re trying to solve A, a great solution to B may seem worthless. But that doesn’t mean that every bad solution to A is actually a perfect solution to a different problem. Some solutions are just bad.

*Thanks to Sarah Perry for this term.

Taste is sensitivity to fit


Alice and Bob decide to check out a new restaurant. Alice orders the signature burger, which is topped with a fried egg and served on a toasted English muffin. Bob, accustomed to plain patties on a bun with ketchup, is disgusted. To him, the combined textures of slimy egg and crunchy English muffin seem cacophonous. In fact, everything on the menu strikes him as ostentatiously “creative”. He grumpily orders only a side of fries. Noticing this, Alice teases Bob for having bad taste, which he quietly resents. Bob likes what he likes. How can his taste be “bad” if it reliably predicts what will and won’t give him pleasure?

Later, at Alice’s apartment, Alice and Bob argue about what to watch on TV. Alice wants to watch reruns of The Big Bang Theory, which to Bob seems beyond the pale. He derides her bad taste and demands they torrent the latest episode of Nathan For You. Alice grudgingly relents, then squirms and winces throughout. She can’t help wondering if Bob is pretending to enjoy himself in order to seem cool and different. If good taste means preferring discomfort to comfort, how “good” can it possibly be?

Alice and Bob are friends, but sometimes they don’t understand each other’s taste, and this causes conflict. Why?


Sarah Perry, once again:

The concept of beauty in diverse domains has a unifying, definitive feature: it reflects the detection of fit between parts of a system. Beauty presents to us as a mystical quale; this is because a beautiful form is a solution to many simultaneous complex problems. Beauty in nature, art, music, architecture, mathematics, and even human faces is a response to the detection of fit.

Ability varies. Just as some people are better at hearing when a musical instrument is properly tuned, some are better at sensing whether a given solution to an aesthetic problem is correct. Just as the phenomenon of fit exists, so does variation in the ability to detect it.

Sensitivity to fit is:

  • Unequally distributed. Those with high sensitivity are often artists, curators, or critics. Those with above-average sensitivity may simply be good cooks and natty dressers. Those with average sensitivity tend to default to whatever solves social problems for them, such as increasing their status. They are unlikely to notice if an aesthetic experience doesn’t work in other ways.
  • Expressed as selection or creation. Most aesthetic problem-solving is just picking the best option from existing choices. But sometimes the best option isn’t good enough, and this may compel those with exceptionally high sensitivity to create something new. Too many subpar burgers might push Alice into operating a food truck that sells her unique recipe. A lack of good TV comedy might propel Bob into writing and producing his own YouTube series.
  • Socially motivated. Lack of interest in a problem can indicate that one is unlikely to benefit socially from solving it. Alice cares about burgers and Bob cares about comedy because both have subconsciously figured out their aesthetic comparative advantage — the stuff they feel valued for knowing better than others in their “tribe”. This advantage can be superficial or even illusory, but it doesn’t have to be. The pursuit of new aesthetic experiences can be like hunting for a new “food” source for the tribe.
  • Not necessarily domain-general. Sensitivity in one area does not guarantee it in other areas. Alice knows burgers, but she may dress functionally and without flair. Bob on the other hand may have generally high sensitivity, leading him to subtly manipulate the aesthetic quality of his environment by surrounding himself with particular objects, colors, even odors. But his choice of burgers belies this.
  • Not the same as understanding. When someone feels strongly about something, but can’t explain why, fights happen. Alice and Bob’s mutual teasing can be interpreted as an attempt to put the other party on the defensive. This way they can avoid the responsibility of arguing for their view.
  • Testable. Those with high sensitivity know the difference between “this works for me” and “this just plain works.”  If Alice and Bob are both employed as burger testers for a large corporation, Alice will consistently output better judgments than Bob about which burger recipes will be popular among a particular demographic. As members of a test audience, Bob will be better than Alice at predicting which TV comedies will be most prestigious.


Now that we see the problem-solving underlying aesthetic experience, taste disputes become less muddy. There seem to be two primary types:

  • “Your problem doesn’t matter.” Problems can be ranked according to different values. For example, if A is closer to being solved than B, then B may be more important. Bob’s preferred burgers are in every supermarket and are therefore easily accessible, while Alice’s preferred burgers are comparatively rare. There may be a way to cheaply mass produce the burger experience she prefers. It follows that attracting attention to the problem is a good idea. Instead of seeing things this way, Alice and Bob dismiss the other’s disinterest as mere bad taste.
  • “My solution will work better than your solution.” Shared problems can have contested solutions. For example, Alice and Bob are in a band together. Alice may not care about comedy and Bob may not care about burgers, but both care about music. Their preferences are pretty similar, so songwriting usually flows naturally. However, they occasionally argue about whether a particular chord progression or lyric will achieve a particular effect. They could A/B test different versions with a select group of fans. Instead the decision is usually made by whoever is more forceful.

Because Alice and Bob don’t see taste disputes in terms of sensitivity to fit, they’re unable to see when their respective sensitivities are similarly sophisticated, if pointed in different directions.  Don’t be like Alice and Bob! Don’t ignore opportunities to find each other interesting.

Chords and maps

In her excellent essay “Beauty Is Fit”, Sarah Perry redefines beauty as indicating a successful problem-solving strategy, saying it “has a unifying, definitive feature: it reflects the detection of fit between parts of a system.”

In art, I’ve noticed two basic varieties of fit: aesthetic fit, which I call chords, and abstraction fit, which I call maps. Both take advantage of “the nervous system of appreciating organisms” but in importantly different ways.

Chords are elements combined in a way that is appealing, but not because the combination describes reality. Chords exploit the many evolved sweet spots of the senses. They can be comprised of “real” things but prioritize creation of an experience over transmission of knowledge. Chords can be consonant or dissonant — the sum of their parts can elicit pleasure or irritation, or even revulsion. Chocolate and peanut butter fit better than chocolate and ketchup.

Maps describe what exists. They exploit the evolved need to understand how reality behaves. They can be aesthetically pleasing but they put the task of abstraction first. Maps “fit” when they achieve compression — when they eliminate redundancies in a pattern of real-world relationships without sacrificing essential features. Poor map fit is usually due to bad compression (irrelevant information that feels like fat on the bone) or outright misrepresentation (features that don’t appear in the abstracted territory). The Wire fits better than unedited surveillance camera footage or CSI: Miami.

Sometimes chords and maps are hard to tease apart. Fit feels like “yes, this is correct!” but doesn’t say why, so it’s easy to misinterpret a chord as a map and vice versa.

Chords are sometimes confused with maps because of their power to trigger associations. Food may taste similarly regardless of where we grow up, but the experience of eating a particular food may be more or less complex depending on accumulated personal and cultural context. Given relevant experience, food can conjure thoughts of home, status, or events in one’s life, but food by itself is generally incapable of triggering associations that aren’t already present.

For both chords and maps, fit depends on the problem being solved. In much fiction, selective distortion of known patterns draws attention to features that don’t fit. The premise “everything is the same as real life except some people are telepathic” must respect facts about reality, but only enough to solve the problem of being entertaining. This map/chord hybrid is called “suspended disbelief” and, handled properly, can be quite pleasurable.

Maps can be more or less ambitious depending on how much they attempt to compress. Technically accomplished portraits and still lifes are less ambitious than realistic portrayals of human emotions in films. In turn, emotions are less ambitious than the compression of entire political systems.

For example, films of the Person Without Lots Of Money Has To Make Difficult Choice genre are often accurate maps of a single person’s experience, and as such, encourage a particular sympathetic reaction in the viewer. Rarely, however, do these films zoom out to see the protagonist’s role in multiple nested systems, a change that can make sympathy more difficult.

I’m not totally confident that “maps vs chords” is the best name for the phenomena described above, but I’m pretty sure the distinction between mostly pleasing and mostly accurate is real. What do you think? Comments welcome.


Let’s say I need directions to a location across town, but for some reason Google Maps only offers two scales: 1000 miles per inch and 10 feet per inch. That’s the difference between national borders and half of a house.

In reality, most searches are for directions to area businesses like Walmart or Starbucks, which require scales roughly between 2 miles and 1000 feet per inch. Our hypothetical Google Maps isn’t wrong, but it isn’t helpful either.

For most human needs, neither is the study of aesthetics.

Philosophers draw and redraw national borders by squabbling about what qualifies as art, while neuroscientists peer at isolated building blocks, investigating how brains process stimuli like shape and color.

But when ordinary people reach for a concept to explain their experience of a movie, meal, or sunset, it isn’t there, so they default to words like “awesome” or “sucked” or “meh.” Even professional critics use language that is more poetic than exact.

Given the poor correspondence between the richness of aesthetic experience and the available maps of that experience, it’s no surprise that aesthetic judgments cause heated arguments.

Aesthetic experience isn’t inherently mysterious; it’s as real as your local Walmart. It’s just really hard to measure.

What would be required to map aesthetic experiences like we map streets? To resolve aesthetic arguments as easily as we resolve arguments about the location of Walmart? First, we’d need access to advanced neuroimaging technology. Then, assuming sufficiently high-resolution scans, we’d have to figure out which patterns of neural activity correlate to various experiences. Solving the technical problems of scanning the brain would be just the beginning. Disentangling cause and effect could take decades.

Which brings us to the project of this blog. While others are naming aesthetic continents and studying individual blades of aesthetic grass, we will try to identify landforms like mountains and rivers. Someday we might get to Walmarts.

Aesthetics obviously also has a social dimension. What we choose (Walmart or Starbucks?) is information about who we are. But, confoundingly, who we want to be also affects what we choose. Both affect what becomes popular or stays obscure. We will explore the dynamics therein.

Some questions we hope to tackle:

  • What people like is notoriously variable, but not infinitely variable. Some experiences are almost universally desired, others reviled. What causes human preference to cluster in the space of possible experiences?
  • Some experiences are primarily sources of sensation, while others attempt to describe reality. Is there a bright line between the two? What criteria limit effectiveness in either category?
  • Good reductionism entails carving reality at its joints. Where are the joints of aesthetic techniques? Is it possible to compare techniques along objective criteria? To establish which is a degraded or more functional version of which?
  • How do people use aesthetic experiences? Which uses are commonly considered high or low status? What makes a particular use more social than personal? What does it mean to have a “fetish”?
  • How does taste work? Why are some people said to have better taste than others? Is there a universal hierarchy of taste?
  • What is the value of connoisseurship? All else being equal, what advantage does a connoisseur have over a non-connoisseur besides “they get to enjoy more stuff”?
  • Aesthetic judgments are extremely abstract representations of real (objective) internal events. Given unequal knowledge of the processes underlying those events, is it possible for A to furnish a more accurate description of B’s experience than B?
  • Mass-production lets more people experience more things, but often at a lower quality. What does “quality” mean? Is it illusory? What factors affect its distribution?