Who will taste the cherries?

By Liam Spradlin · Edited by Sam Dresser

On a sunny day in the summer of 2015, I found myself standing in the basement of an office building in Silicon Valley. The room was packed full of appliances, as if a commercial kitchen had been blown apart. Several stacks of wall ovens huddled together on dollies. Tangles of electrical cords popped up behind a broad stainless-steel counter. Rags and sponges dried slowly on the edge of a giant sink, over which a pre-rinse nozzle drooped like a willow branch. Nearby, a computer monitor showed Python developer tools next to a spreadsheet dotted with red and green cells. In the middle of the room was an improvised prep station on which there was an odd little machine. It looked like a tiny microwave designed to warm up a single cupcake. I later learned it’s a moisture analyser: the researchers use it to see how much juice is left in their samples of lamb, swordfish, chicken, pork and beef. I’m writing down everything I see: 10:45 am. The kitchen is full of smoke.

That basement room was a test kitchen, where two chefs and several researchers had been busy cooking a whole farm’s worth of meat in a set of prototype ovens. The company’s vision was to invent a cooking machine that could whip up Michelin-grade dishes at the press of a button, and the crew was tasting the food and filling out ‘sensory inventory’ forms that recorded qualities like colour, smell, temperature, texture and taste.

As the design lead for the oven’s touchscreen interface, I was there to learn from the chefs in order to build something that worked for them and for the ‘home chef’ or end user. The tension between their culinary intuition and the ovens’ logic was immediately apparent: the kitchen was full of smoke because one chef had attempted to cook three racks of lamb directly under a broiler to get the colour you’d get by ‘smacking it with a propane torch’. This instinct didn’t translate to a language the oven understood. I wrote down Smacking? and kept watching. The other chef, tapping away on another of the ovens’ screens, muttered: ‘These things are worthless. I need to be able to use my chef spidey senses.’

The chefs were hired to help program the ovens to follow their recipes. But they also had to pretend they were normal people cooking dinner for their families in order to see if the ovens could replicate expert techniques. On one hand, the chefs needed an interface with enough controls to make manual adjustments, adding to the oven’s corpus of training data. On the other hand, end users would need simplicity and the impression of smarts.

The company building the ovens viewed human intervention in the recipes as a liability. Having too many options would invite users to fiddle with the recipes, which would cut against the ovens’ smarts and risk producing a bad result. The product team talked a lot about users forming ‘bad habits’ if they manually adjusted recipes too much. ‘Change the ritual’ became a mantra written on whiteboards, sticky notes and task trackers around the office.

Still, I took the vision seriously. It was an early project in my tech career, at the height of the 2010s techno-optimist boom. I believed that, with the right information, the problems of cooking could be solved through clever design. But over the next decade of experience building interfaces at scale – from urban infrastructure to global design systems – I came to understand the test kitchen as a microcosm of a fundamental struggle in tech between human subjectivity and a belief in the computability of everything; between the human body’s senses – taste – and the view that data is a more valid form of knowledge.

Taste, both as a physiological process and an aesthetic preference, is subjective. It’s contextual and belongs to the individual. In the smart kitchen, as in so many of tech’s pursuits, this kind of inherently personal experience is (and must be) made uniform and repeatable. In other words, taste needs to become data – made into what the philosopher Albert Borgmann called a ‘device’ that abstracts the means and delivers the ends with little or no human input – so that human effort (and the home chef’s agency along with it) is abstracted away.

But an oven can’t taste. It has no idea what ‘smacking’ something with a propane torch means. As the designer Nitzan Hermon writes in his Substack, asking if an algorithm can understand something subjective is a category error, like asking a stone which movies it likes. It’s impossible to experience the taste of a cherry just by counting the petals on a tree. No matter how many data points are collected, data alone cannot reproduce – or even accurately represent – embodied experiences. Yet the oven relied on an assumption that the qualities of food and human taste are quantifiable.

The smart kitchen was my introduction to an ideology I call computism, which holds that the Universe is made of discrete, measurable bits of data – what philosophers call a ‘digital ontology’ – and extends that view into an economic and political system. According to computism, the world, including humans, can be computed: the bits that everything is made of can be represented and manipulated as data. The Universe can be governed, essentially, by code.

Tech leaders have made enormous claims about how computation can manipulate the world. Elon Musk says that AI will ‘rewrite the entire corpus of human knowledge’ and Dario Amodei, CEO of the AI company Anthropic, writes that graphics processing units will soon turn us into a ‘country of geniuses in a datacenter’. Others suggest that the Universe may be a simulation meant to bring about superintelligence, or that a future AI will be, literally, a god. Implicit in all these beliefs is that, in a world that can be predicted through computation, the most powerful person is whoever runs the computer. The datafication of our daily experiences and the increasing presence of predictive models are best understood through the lens of computism. When everything is computable, everything is predictable, optimisable and manipulable.

This narrative is not an invention of Silicon Valley. The ancient Greek philosopher Democritus proposed that the universe and everything in it was made of indivisible bits called atomos (‘uncut’, ‘indivisible’). This early digital ontology found new life in the Enlightenment when René Descartes took up this atomic model, incorporating everything physical into it, including the human body. In Descartes’s model, the body was thought of as a complicated machine imbued with an internal user – the soul – which was the only thing not made of bits. Later inventors would set out to prove Democritus and Descartes right. Jacques de Vaucanson’s 18th-century mechanical duck, a stepping stone to later mechanical looms, simulated eating and digestion – a creation that eventually led me to that basement room in California, attempting to simulate taste in a less fanciful mechanical object.

Since my time in the kitchen, I’ve observed computism slowly taking hold, and have had to grapple with how my own work – interface design – manifests it. Often, design projects centre on optimising how users move through the interface. Engagement, conversions, efficiency and productivity – a persistent fight against friction. Designers focus on measuring the actions of the user because outside of a research environment we don’t have access to their individual thoughts. To the computer, then, actions constitute the person – which means that the person, as anything more than a set of actions, disappears from the designer’s view. Over the years, interface design has shifted from a craft engaged primarily with human subjectivity (as all design is) to one seeking mechanical objectivity, free from the contaminating influence of human interpretation or judgment.

When I entered tech, many of my clients were adherents to the design consultancy IDEO’s ‘design thinking’: a repeatable series of steps for solving problems through design. This recipe, applied to all sorts of things from new apps to public governance, catalysed the formalisation of user experience (UX) design, offering the appearance of objectivity through data to risk-averse decision- makers. UX design, as the tangible manifestation of the code and data that powered the tech boom, was having a moment. And I was there on the ground, embracing UX design’s newfound proximity to power. The growing belief that design was a replicable process helped us sell its importance and helped us believe that design and technology together could change the world for the better.

Fast-forward 10 years, and design has succeeded in getting its seat at the table of business. But, to attain its status, software design had to change. Once design was accepted as a mechanism for quantitative growth, its artefacts (the interface) were broken apart so they could be measured. The elements of the interface – buttons, cards, lists – were broken down into various instantiations of shapes, colours and typography, each with discrete values. These constellations of visual qualities were codified in design systems. Design, through software, moved from a complex and contextual arrangement of interrelated structures toward a mechanical object with swappable parts.

Techniques like A/B testing – in which slightly different versions of an interface are served to different users to see which performs better – became standard in the digital interface as infrastructure improved. It was a technique adapted from the general concept of randomised controlled experiments and principles from behavioural economics. With exploding user bases, the ability to run experiments and tune the interface toward desired outcomes reached unprecedented heights. Over time, the interface became a one-way mechanism of extracting data. In the words of the critic Silvio Lorusso, ‘what was once a promising field rooted in problem-solving [became] a problem in itself.’

One illustration of this is a legendary anecdote about Google’s blue search result links. The Search team at Google tested 41 shades of blue to see which one got users to click on more links. The designer Douglas Bowman cited it as an example of the data-driven culture of tech that led to his resignation:

Yes, it’s true that a team at Google couldn’t decide between two blues, so they’re testing 41 shades between each blue to see which one performs better. I had a recent debate over whether a border should be 3, 4 or 5 pixels wide, and was asked to prove my case. I can’t operate in an environment like that. I’ve grown tired of debating such minuscule design decisions.

(One designer even told me that a typo led to an off-putting brown colour being included in the mix. The response from leadership? ‘If it performs better than the blues, we’ll keep it.’)

This kind of mechanical optimisation, delivered to users through design, abstracts the craft’s human element – the role of subjective taste – by focusing on quantified outcomes, losing meaning along the way. The interface becomes a ‘device’ (in Borgmann’s sense) as the shift from complex structures to isolated elements moves it further toward absolute quantification. The actions performed through software – communication, planning, learning – are reduced to computable outcomes, alienating users in the process.

The dominance of data permeates software down to individual user interface (UI) components: a five-star rating system, a text string, a dropdown, a hashtag. Experiences are collapsed into bits. Actions like ending a relationship or talking to your parents are now charged with social subtext based on what software you might use to perform them. Software’s ability to observe, model and respond to the user is turned toward creating or augmenting rather than facilitating the user’s intent.

Computism, in order to justify its pursuit of total datafication, promotes a model of reality that escapes the limitations of the human body and mind to see the world as it must really be: data. This leap requires a shift in how power operates, which the philosopher Byung-Chul Han explains in his book Psychopolitics (2014). As neoliberalism moves from physical coercion on a factory floor toward psychological optimisation on screens, users are driven toward their own exploitation.

We can see this directly in the mechanics of the interface. Something as simple as an endless feed with a slot-machine-like pull-to-refresh interaction doesn’t force the user to work by staying on the platform. Instead, the user is made to want to generate data. This distinction is key to understanding the insidious side of interface design, which exploits the physiology of the mind to maximise its reliance on the machine.

This, according to the philosopher Andreas Spahn, is how digital technology upends the traditional Cartesian subject-object model. Digital technologies, he argues, are no longer passive tools or inert objects upon which we act. Instead, they are becoming autonomous. When users outsource agency to the interface – allowing their kitchen, for example, to cook for them while they run errands – the interface begins to take on qualities of a subject while the user moves closer to a data-generating object.

The sociologist Maurizio Lazzarato’s writing deepens the critique of computism. He argues that, beyond reducing us to data-generating objects, capitalism uses that data to manufacture subjectivity from the inside out. In his book Signs and Machines (2014), he argues that technology bypasses our conscious behaviour to break us down into pre-individual components (like algorithmically detected ‘interests’) that can be made into measurable inputs. In this way, technology has the capacity not just to observe and respond to us, but to construct our responses to it. It’s similar to the way design itself has been broken down into atomic elements that can be snapped together and recombined to produce new interfaces. As if to illustrate Lazzarato’s point, many designers and engineers have moved increasingly toward feeding their design systems to large language models (LLMs) to produce simulacra of designed interfaces. Elsewhere, synthetic data is used to, for example, create synthetic user personas – imaginary people made of Lazzarato’s fragmentary, pre-individual information against which to test their ideas. The actual human can be abstracted out of the process entirely, leaving just a simulation. After all, in the computist model, humans can be understood as data, so data can be substituted in for humans.

This is what it means to treat the body and the mind as a device. Embodied subjective experience is abandoned in favour of measurement. Speaking to a consortium of longevity clinics in 2024, the entrepreneur Peter Diamandis said: ‘It’s either a hardware problem or a software problem, and we’re going to be able to fix that!’ If the body is complicated hardware, the mind is its buggy, degraded software waiting to be optimised and put to more productive work.

And that brings us to the logical conclusion of computism’s metaphysics: the idea that generative AI will transcend humanity to become a post-human subject. To the computist, human knowledge and culture are too contaminated by humanity to be objective. The pursuit of a ‘superintelligence’, a ‘country of geniuses in a datacenter’ or an emergent AI entity that will ‘fix the climate’ (as Sam Altman says) is fundamentally a project that seeks to debug human subjectivity. It’s about escaping humanity, handing the computer our responsibilities, our mistakes and our fears – an apology for the embarrassing mistake of letting humanity carry on in such a messy state for so long. Spiritual belief systems often endure because they bring a sense of clarity to the apparent senselessness of being human, suffering and dying. I see computism in the same way, except, instead of locating the cause of human suffering in a demiurge, a devil or a desire, computism locates it in simply being human.

The collapse of subjectivity into data is happening in real time. The dream of a smart home kitchen has morphed into paid services like Wonder, which assigns users three meals a day by algorithmically synthesising their fitness goals, preferences and hunger to choose what they’ll eat next. Forget cooking. Forget even thinking about what you want to eat – the app will handle it. And it isn’t limited to our bodies. Tools that used to act as spell-checkers now offer critiques from AI approximations of real human authors. Writing, an inherently dialogic act of sense-making (like design), is abstracted away. The tool tells humans how to sound like another human based on all the human writing it could access. A startup announced in June 2026 that even ‘taste’ is a data problem and, with extra training, generative models can create great works of taste.

The computist project works by co-opting human knowledge from the commons. The ‘corpus’ that technologists want to rewrite is the shared cultural output of billions of people over thousands of years. Computism sees this collective output as a complicated mass of data – an unclaimed natural resource to be refined, freed from contradictions and subjectivity, until the knowledge, in the words of the historians of science Lorraine Daston and Peter L Galison, is finally free from the knower. Computists believe this capitulation to raw information is the solution, that it will lead to the discovery of new truths – a utopian future in which the computer no longer needs an operator, and humans no longer need to be human or solve human problems. Over and over again – when we can’t figure out how to reconcile the subject and the object – we simply remove the subject. When, one day, only the computer remains, we can let it design itself, optimise itself and eliminate the friction, error, unpredictability and struggle that define our lives. But what then?

The real tragedy of computism is not just that computers could take over some human activities or exceed our intelligence. It’s that adherents could build a world in which every petal is counted, only to eventually realise that we abstracted away our ability to taste the cherries.