Frictionomics

Our pursuit of convenience has evolutionary roots: we're wired to conserve energy because we emerged in a context of scarcity, not caloric abundance. That trait, once a feature, has become a bug that’s increasinglyu incompatible with the reality we live in today, overloading hardware that hasn't been updated since the early Pleistocene. No fast updates are expected, given the manufacturer's track record.

The 20th century was when many of the physical frictions in the human experience were removed — the proliferation of cars and public transit, the mechanization of housework, and the rise of industrial food. In many aspects, it was liberating and wonderful — who would rather hand-wash their clothes or swap the vacuum cleaner for a broom? But the result is that many people now leave the house and pay to move their bodies, and since 2010, the WHO has treated physical inactivity as a global epidemic. It took a long journey to understand what we were losing and to recognize how important those frictions actually were.

In the 21st century, we are on a trajectory of reducing other types of friction: relational, cognitive, and intellectual.

A great (and very Brazilian) example of how the price for the removal of some frictions comes later and in subtler forms is the dominance of asynchronous communication. When we communicate asynchronously, we don't have to make ourselves available on someone else's schedule, we can multitask, and we can advance our own goals autonomously. But then, in the cozy warmth of passive-aggressive emails or the ambiguous silence of WhatsApp, we stop discussing things that would take two minutes on a call or a face-to-face conversation — exchanges made far richer by tone of voice and body language, capable of avoiding conflict, nurturing relationships, and making communication genuinely, only by other metrics.

Friction reduction is a core part of the appeal of much of what we consume, from delivery apps to ready-made juices. Capturing value, in many cases, means shortening the effort between people and what they want. UX is a discipline fundamentally focused on reducing digital friction (so much so that measuring effort is crucial), sometimes with noble goals, other times not so much. The very process of building brands is also friction reduction: it's about shortening the path to memory and to the decision.

All of this suggests that friction reduction is a driving force behind design, marketing, and product and service innovation — the problem is how this reduction, in many instances, works against our interests both individually and collectively, especially in the long run.

Stop! Hammer time

The arrival of LLMs has been like a hammer that sees any and all human friction that can pass through the digital as a potential nail.

One of the recent targets is relational frictions. In June of last year, I wrote about the rise of synthetic relationships and the productization of our difficulty in dealing with alterity (one of the most-read pieces in this newsletter), but what happened since then is impressive. We have more “agentic” dating apps because swiping right has apparently already become too much friction, griefbots designed to ease the friction of processing grief, and Meta recently patented an LLM that simulates people posting after they have died — yet another major reduction in the friction of the work of Charlie Brooker, screenwriter of Black Mirror.

Going against the grain, some are beginning to recognize the risks of removing these frictions. China recently moved forward with a regulatory package against AI anthropomorphization—limiting how much systems can simulate human emotions and bonds. The principle is simple and clear-eyed: the ability to emulate personality, affection, and care is not genuine humanity, and this needs to be explicit at all times for those who use it. Their framing focuses on the risk of manipulation and inducing dependency — especially for the young and elderly — and not on the “human flavor” as on our side of the world. Can we really only solve the problems created by one sledgehammer with another — the regulatory one?

Intellectual and cognitive frictions: the illusion of instant mastery and the explosion of the Dunning-Kruger effect

Our relationship with technical and academic knowledge was already shifting, with the capacity to optimize the form and the emotions triggered (laughter, cuteness, but also hatred, outrage, and positive reinforcement of beliefs) becoming a key driver of visibility and consequently, financial success, at the expense of factual accuracy and lived experience. This brought forth two fundamental outcomes: the conflation of reach and authority (which I have already mentioned here, including in the context of human insights and market research) and the shattering of our understanding of reality.

The arrival of LLMs, beyond radically reducing the friction of producing text, also profoundly changes how we access more technical and specialized knowledge — in a wonderfully disintermediated, near-instant way. You send a photo of your open PC and AI helps you swap out the parts. You upload your lab results and it gives you a full breakdown of what's going on and answers every question you have.

There's a genuinely great side to this: laypeople can protect themselves from professions that use information or technical asymmetry (mechanics, lawyers, and so on) to control the client relationship. But then we start to believe we know enough to make decisions on our own in domains we don't actually understand very well, and that, hey, consulting a specialist might not be necessary after all.

The problem is that this illusion only lasts until the moment we are confronted by someone who actually knows what they are talking about and isn't using a machine to statistically infer the most probable answer.

The results of these behaviors playing out at scale are already visible: a massive amplification of the Dunning-Kruger effect and of illusory exceptionalism which is when we want to commoditize the work of others but think it's unacceptable for ours to be commoditized. Since these limitations are inherently human, those on the front lines of exposure to this new context are disproportionately affected by them, starting with some technology leaders who expect to revolutionize sectors without a minimum understanding of the human processes and dynamics involved.

This dramatically increases the pressure on those who do specialized intellectual work.

Under pressure

In this pressure cooker that is the anxiety about where LLMs fit in the world and our more ambivalent and critical relationship with our online presence and social media, we are already seeking frictions deliberately in response to this reality in which we live. Perhaps we are collectively beginning to understand that the problems created by technology aren't necessarily solved by more technology.

In an earlier piece on the saturation of the attention economy, I wrote about a more critical look at our online presence and the emergence of initiatives and businesses built around "forced presence" and the offline. Since then, the temperature has risen:

• Zuck in the hot seat over the impact of his platforms on children and young people — and internal Meta studies that leaked show robust evidence of harm.

• In Toy Story 5, soon to be released, the villain is a tablet.

• Year-over-year engagement is falling across multiple platforms: Instagram (-26%), Threads (-18%), and LinkedIn (-5%).

• And to top it off, Nature published a study empirically confirming that yes, the medium is the message — exactly as Marshall McLuhan predicted.

From the early days of the influencer era around 2014, when Kim Kardashian "broke the internet" and we discovered that some people's everyday lives could be highly monetizable without intermediaries like gossip magazines, it took more than ten years for a broad and structured backlash to emerge.

But the criticism is expanding in scope. In education, countries that lead global indices are “de-digitalizing” classrooms, reviving handwriting and delaying the introduction of computers and tablets.

In gaming, there is a large and profitable wave of harder games that force players to figure out what to do on their own, both in the indie world (Hollow Knight and its successor Silksong) and in the AAA world (Elden Ring and other souls-like), in a kind of revival of a much older design philosophy like that of the first Legend of Zelda, which was totally exploratory and depended on curiosity, trial and error, and a willingness to get lost in another world. The friction was, in a way, the reward itself.

What happens if this idea gains social momentum?

The search for deliberate friction crosses ideologies and values

To illustrate how this search appears even in social groups with very different worldviews, I brought some practices that have been on the rise and being more discussed in Brazil in recent times.

What do the growing interest in Legendários (a Brazilian Evangelical men's movement focused on physical and spiritual trials), triathlons among executives, the practice of manual crafts such as ceramics, and urban garden and agroecology collectives have in common?

On the surface, they seem like distinct things with no apparent relation between them, especially if we look through the lens of external signaling (what it says about me to others):

• Agroecology and urban gardens — signals collectivism and ethical sense, value creation and subsistence “outside the system,” self-sufficiency.

• Legendários — signalsmasculinity through physical ordeals, reinforcement of traditional gender roles, and male socialization centered on activities, rather than conversations.

• Ceramics and other manual work—signaling of creativity and artistic sensitivity, appreciation of the artisanal (in contrast to the industrial / mass-produced) and of small scale workers.

• Triathlete executives—proving oneself and performance optimization, status signaling through participation in international events and physical capacity.

But they share something deeper, when you look from the inside:

• Being offline

• Self challenging

• The emphasis on the sensory experience over the intellectual and, also, on active involvement instead of passive consumption

• 100% focused attention as a base requirement

• Unmediated socialization

• Identity displacement and temporary reset of social roles—during the the activity, I am no one's father or mother, I am not a customer or supplier, I am not a boss or employee.

These are practices that function as a temporary suspension of our ordinary lives. In anthropology, this has a name: liminality.

But this doesn't necessarily cut across socioeconomic lines

What these activities also have in common is that they assume significant availability of money or, at the very least, time, two things that are particularly unevenly distributed in Brazil.

It is essential that we remember how much this reflection on intellectual and cognitive friction can perfectly be representative primarily of our bubble of intellectual work and the corporate world, both historically restricted to a narrow elite in Brazil. The world outside it may perfectly fine with embracing what little leisure they have with less guilt and have no desire to climb any mountain.

To argue with data: we can see this in practice with the adoption and use of LLMs, which has a huge SEG bias—it’s 32% among Brazilian internet users (criterion: used in the last 12 months / any purpose)—but there’s 69% in A, 54% in B, 49% in C and 16% in D/E SEGs. The TIC Domicílios 2025 (the Brazilian ICT Household Survey), the source of these data, understands that this represents about 50 million people, comparable to the population of the state of São Paulo, which has 46 million—it is far from being “everyone”! Besides this, professional use is much more concentrated in A than in the others.

Note: Base / universe = internet users | Professional, Personal and Research bases = LLM users.

A key methodological aside here. Unlike other studies recently released with very different figures, this survey is probabilistic (that is, it uses controlled randomness in sampling, making it far more representative and allowing for a calculable margin of error—0.8% in this case) and based on in-person, household data collection (which ensures much broader and more controlled coverage—it is not limited to those who volunteer to participate!), unlike online panels.

These panels, by definition, overrepresent certain groups: heavier internet users, people in metropolitan areas, younger populations, and socioeconomic segments for whom participation incentives are more appealing. As a result, studies based on them may end up describing their own panel composition better than they describe Brazil at large. This is why such methods are not used for electoral or opinion polling, or in any context where the goal is to measure the country — not just those willing to answer an online survey. We’ve seen large discrepancies like this before: recently in the UK, with a supposed revival of religiosity, and in Brazil, with data on alcohol consumption among young people.

This is another practical example of how the reduction of friction (cost, deadlines, complexity of the operation, discernment of buyers, honesty of sellers) can also bring unintended effects elsewhere. The rhetoric of “good enough,” “objectivity is the enemy decision-making” and the hype around synthetic data — vastly modeled with studies done with convenience samples — can accidentally amplify bubbles, fabricate consensus, increase FOMO and distance us from objective reality (and objectivity is a pursuit, not a destination). And this is in an informational context already highly fragmented, politically polarized, distributed and absorbed according to our previous beliefs. Friction that brings us closer to reality is worth preserving!

Who are we talking about then?

To help place this phenomenon socially, it's worth recalling that György Lukács argued back in the 1940s that the existential is a bourgeois concern and that only about 17% of Brazil is AB SEG, depending on the criteria.

Most employment here is tied to agribusiness, retail, construction, and in-person services like restaurants, hair salons, and logistics. Even in office work, most roles are operational and administrative rather than purely intellectual, in both the logical and creative senses.

But the real highlight here is something else: the greatest impact tends to arrive first precisely in the segment that builds social distinction by contrast to manual and operational labor. If the most cataclysmic scenario actually materializes and certain types of intellectual work have their market value destroyed, we may witness a plot twist no Russian revolutionary, however visionary, could have imagined: the temporary economic and prestige decline of bureaucratic work in favor of the physical labor. In countries where adoption is moving faster — with more workers in advanced manufacturing and more IT and finance in services — there are already signs this may be happening.

A shorter cycle from idealization to backlash

However remote that possibility may seem in the short term, the tension and anxiety are palpable — especially in a country like ours, where this kind of work has almost always been a social marker and physical labor has always been undervalued. This real risk of economic decline and of identity and symolic erosion fits what sociologists call anomie and what Alvin Toffler called "future shock."

When smartphones and social media arrived, public reactions were different: the mood was one of near-universal excitement. iPhones were aspirational objects and watching Apple launches was collective entertainment. The downsides of these novelties may not have been as immediately visible or as frightening.

Even in research and insights, people were excitedly declaring that quantitative research would wither, that representativeness was dogma, that social media would become the world's biggest focus group... 🤷🏻

That's not to say there were no critics — some of them quite visionary. The first public mention of "crackberry," the term mocking the addictive potential of smartphones, dates back to 2001. The tech press called Steve Jobs's charisma and stage presence a "reality distortion field," especially when he sold as revolutionary things the competition had been doing for years.

An archaeological record of the hype and idealization of that era — if you were working in marketing in 2009, you've probably seen this video

But now it's different — in part because the adoption of LLMs isn't happening entirely voluntarily. It's being driven by competitive pressures of all kinds: top-down mandates, some even tied to promotions, productivity demands, aggressive cost-cutting, fear of being left behind. A lot of people are being pushed, not walking in willingly. Many critics use precisely this dynamic as their argument: if the technology were entirely trustworthy or desirable, perhaps no one would need to be pressured to use it.

On top of that, people's individual experiences vary wildly. There are complex tasks where the performance is remarkable and simple ones where the results are consistently frustrating. This highly heterogeneous adoption dynamic tends to generate far more resistance, criticism, and early concern about negative, even catastrophic, impacts of all kinds.

If technology adoption cycles generally move through stages: first hype and idealization, then normalization and mass adoption, and finally disillusionment or backlash — with LLMs this timeline is far more chaotic: the idealization, the fear, and the pushback are happening almost simultaneously, which means we're already weighing what we have to lose.

If technological adoption cycles in general pass through stages—first of hype and idealization, then normalization and mass adoption, and finally disenchantment or backlash—with LLMs this timeline is more chaotic: the idealization, the fear, and the contrary reactions are happening almost at the same time, which makes us already weigh now what we have to lose.

What comes next: a renewed appreciation of process and the arms race for the “human touch”?

One of the spaces where these negative reactions are most visible is in LLM-generated content on social media and even in journalism. "Sloppification" and the criticism that comes with it are already part of a broad conversation — especially when commercial interests are involved — a discussion that seems to be more about how to use these tools than whether to use them at all.

In luxury and art, process has always been an inseparable part of building narrative and value: Nappa leather, the walnut trim of vintage Jaguars, the hand-dyeing of Japanese selvedge denim, dishes prepared from scratch with hard-to-source ingredients in fine dining, the frame-by-frame animations of Studio Ghibli. These are the things that, once you understand the effort involved and the role it plays in the final result, you come to appreciate even more.

Costly signaling is exactly that: when we use markers that show others that what we made or communicated involved serious effort or expense. It's an important concept in game theory, behavioral economics, and negotiation. When something appears expensive, labor-intensive, or sophisticated, it shapes how people assess its value. But this mechanism can operate in different ways:

• It can negatively affect the perceived value of something: you attend an event and receive a cheap promotional gift — rather than improving your impression of the brand, you leave disappointed. Their effort didn't translate into a positive perception.

• It can be used to create the illusion of effort: as in the semantic hollowing-out of the word "artisanal," applied to clearly industrialized products like sliced bread and frozen pizza. Same goes for that LinkedIn post that deleted the em-dashes but kept all the LLM tics that we've already learned to recognize - is robotic text the crappy swag bag of intellectual work?

Advertising, for example, is more persuasive when there is costly signaling. But we're in a moment where the focus on volume and flooding the zone is so intense that we fail to notice: if the perceived effort is zero, the perceived value also tends toward zero. Will this push us toward appreciating genuine intellectual effort more?

The problem: is it possible to do costly signaling in something that isn’t necessarily physically perceptible?

If intellectual work no longer carries visible markers of the effort involved, and reaching some outcome stops being a competitive advantage, does the differentiator become the process? Recent evidence from music: while 97% of people are unable to distinguish music made with generative tools from conventional music, 73% believe this use should be openly disclosed, 45% would like filtering tools, and 40% would always skip tracks produced that way.

The process, in human learning, is what ensures we internalize experiences and grow — even in things that might seem like a waste of time or whose importance is hard to prove. The falls and the gradual recognition of what physical balance feels like when learning to ride a bike. The calluses that form on your fingers from practicing guitar. The darkening vision and near-fainting as you build physical conditioning in a martial art or a demanding sport. The paragraphs you need to re-read several times in a difficult text. Staring at a blank screen or page for a long time before writing. All of these processes produce tacit knowledge that can't always be put into words.

Martin Heidegger said that language is an incomplete system for transmitting information — a discussion that becomes acutely relevant in the moment we're living through. Talking about something is not doing the thing, and not necessarily learning about the thing either. There are sensations, emotions, and experiences that language is insufficient to convey, and that is an inseparable part of human experience. This explains both a possible ceiling in LLM capabilities and the new venture from Yann LeCun, who wants to model the world directly, not through language.

Now, if it's clear that we value these frictions and efforts — even when they can't be transmitted through language — and, as I argue here, we'll increasingly reflect on which ones are worth preserving and which ones are worth investing time and money in, a new arena emerges: signaling itself. If the value of what we do isn't immediately recognizable, we look for substitute signals to demonstrate that it exists: degrees, certifications, years of experience, association with recognized institutions, or humanizing traces that indicate the effort was real.

And speaking of the future…

Technological uncertainty is high, but the human dynamics are visible and reveal the incentives

We're in the middle of a social experiment that stirs strong emotions. How we see the issue increasingly depends on how we see technology: fatalistically, enthusiastically, skeptically, cynically, or even catastrophically — which makes the discussion more polarized and more grounded in belief than in fact. This plays out even among some of the brightest minds in the world across different fields of knowledge. There's no scenario in which everyone is right at the same time, and at this level of divergence, some of the mistakes will be very costly.

Rather than just debating technology's potential, perhaps we should pay closer attention to the human dynamics around it — for example, how gaining access to the right people and to capital involves echoing a certain discourse, or where the preaching diverges sharply from the practice among key figures in tech. As always: cui bono and follow the money.

So what now?

• There may be room for brands that sell as a service or participate in a relevant way in frictions considered productive, valuable, or rewarding.Shall we work on identifying which ones those are?

• “Offline is the new luxury” is a cute framing that tells only half the story. Historically, low-cost products like powdered juices and free-to-play games are invariably pushed to those with less choice. This can transform AI slop into the cultural and cognitive equivalent of instant noodles. If it happens, we have a real chance of having cognitive inequality expanded—which would also create a promising market for those who can mitigate this impact.

• Don't assume your audience / leads / clients / consumers are suckers who can't tell your brand voice or your LinkedIn posts sound robotic and riddled with LLM tics. The temptation to crank up the volume in the knife fight for attention we're all stuck in is real — but if everyone does the same thing, the collective effect is a disaster. The higher the proportion of garbage content, the worse or smaller the audience for that channel gets. Game theory in action: if everyone pees in the pool, who still jumps in? What happens when every channel out there turns into a 24/7 infomercial?

• Attention is not trust, attention is not relevance, attention is not intent. A practical example: we spend a lot of time thinking about and being exposed to certain public figures. Does that increase our favorability toward them? In some cases, it's quite the opposite — as the internet has put it perfectly: "Everything I know about this person I learned against my will."

Wrapping up: in a time where, in the competition for attention and capital, people will even claim that consciousness, one of the hardest problems in science, is already in the machine, this month's recommendation is Michael Pollan's interview on consciousness and on which frictions may be worth preserving. Thank you for reading to the end, and see you in the next edition!