A particular kind of technological optimism assumes that anything worth knowing can eventually be measured, modelled, and optimised.

Liam Spradlin calls it computism and begins his argument in a Silicon Valley test kitchen where engineers are trying to build an oven capable of cooking Michelin-quality food without the messy intervention of an actual cook.

From there, the essay becomes an argument about something much larger than cooking. What happens when subjective experience is treated as a problem of insufficient data? What happens when taste, judgement, and intuition are regarded not as forms of knowledge in their own right, but as imperfections waiting to be translated into something a machine can understand?

The essay begins with a kitchen, a chef, an oven, and the stubborn fact that an oven cannot taste.

It turns out to be a much bigger problem than it first appears.

Read: Who Will Taste the Cherries?

Liam Spradlin, Aeon, September 2026.

Why RareRead picked it

It gives a name to an assumption that is increasingly difficult to see from inside the systems built upon it: that if something cannot be measured, represented as data, and optimised, it is somehow an inferior form of knowledge.

And it leaves you with an uncomfortable question.

If we eventually succeed in making everything computable, what happens to the things that were never meant to be computed?

The internet made publishing almost effortless. Anyone could have an idea, write it down, and put it in front of the world. That was one of the great expansions of human possibility. The falling cost of producing writing is now pushing that abundance much further, making it possible to produce more competent prose, more quickly, than anyone could reasonably consume.

Our capacity to read, however, remains stubbornly limited. We still have the same hours in a day, so as the supply of writing grows, deciding what deserves to be read becomes increasingly important.

We have largely delegated that selection to algorithms. They rank what we see by signals such as engagement, popularity, recency, and similarity to what we have already shown an interest in. These are useful signals for predicting what people are likely to click on, but they are poor substitutes for judgement about what is actually worth reading.

The problem is that these signals reinforce one another. What attracts attention gets shown to more people, which creates more attention, which gives the system even more reason to show it. Popularity becomes evidence for popularity. Familiarity becomes evidence for relevance. The things that are easiest to discover are increasingly the things that have already proved their ability to attract attention.

This creates a peculiar distortion. A popular essay becomes more visible because it is popular, while an unfamiliar essay has to be exceptional before the system has much reason to surface it. Discovery becomes increasingly circular: we see what has already been seen, and what has already been seen gets seen again.

The best essay is not necessarily the one getting the most attention. It can come from someone with a hundred readers, receive almost no engagement, or have been published years ago. It can still be brilliant, and it can still be missed.

Reading selection works differently. It begins with the possibility that something is valuable, and is willing to make that judgement before popularity has made it obvious.

RareRead will make that selection once a week.

Each week, one essay or article will be picked because it has something to say that is worth spending time with. It might contain an original idea, a powerful argument, an unexpected perspective, or a way of looking at something familiar that changes how you see it.

It could be something published yesterday or something that has been sitting quietly on the internet for years. It could come from a well-known writer or someone almost nobody has heard of.

Popularity and recency are not the point.

The point is whether it makes you think.

When writing becomes abundant, finding the exceptional among everything that is merely good becomes a problem of its own.

RareRead starts there.