Mutually hostile national fragments, and on the public's retreat into an ever-narrowing circle of neighbourly proximity

Initial title: The Corner Market and the Digital Fiefdoms — On Trust That Has Retreated Back Into the Neighbourhood.

An essay on platforms that have become infrastructure, on the slow erosion of the trust upon which they rested, on the fragmentation of the internet itself into mutually hostile national fragments, and on the public's retreat into an ever-narrowing circle of neighbourly proximity.

Translator's note. This essay was written in Romanian and adapted for an international readership. Where the original relies on local imagery — a small provincial market in north-eastern Romania, a family lineage of mechanics, references to Romanian intellectual tradition — I have kept the imagery but added brief explanatory notes for readers unfamiliar with the context. The Romanian original of this essay is available at petrucojocaru.ro.

I. The Machine That Still Works

The market on the corner has no official name. The people from the street simply call it the small market, to distinguish it from the big one, where one enters beneath a tin-lettered sign and where the stalls are lined up like the parking bays of a supermarket. At the small market, people do not come merely with a shopping bag. They come with a second of respite between two obligations, with the question "how are you?" addressed to someone they have known for ten years, with the suspicion that the tomato from the third crate is better than the one from the first, because someone brought it in from Cristești.¹

The vendor on the corner has a coffee machine — a classic La Cimbali with a manual lever, probably bought second-hand fifteen years ago. The chrome plate is scratched around the portafilter; the lever handle has acquired that particular lustre which only a hand pressing it daily can produce; the pressure is still within parameters. Her coffee is better than the one from the petrol station across the road, where a ten-thousand-euro automatic machine with a touch-screen stands unattended. Not because her beans are more expensive, but because the lever is pressed by a hand that knows, and the water passes through the coffee at the rhythm it should. It is, if you like, a small monument to what Constantin Noica would have called the utterance in things — the object which says, through its very functioning, what it is.²

When someone remarks, in passing, that the tomato from the corner tastes of coffee, it is not an insult, and it is not a laboured metaphor. It is a form of recommendation. The speaker means: there is something in that place which holds together the small things of quality. Good coffee and good tomatoes stand at the same counter not by coincidence, but through what we might call — in less scholarly language — a way of being of the place. A microcosm in which trust is not requested; it is observed. The vendor does not tell you the tomato is good; she lets you smell it. The coffee holds no barista certification; it has the taste one remembers. This place operates on a fund of pre-institutional trust — the kind the Romanian village knew for centuries, and which the city has preserved, almost without realising it, in its little islands of small commerce.

I recall this scene because in the 2026 Edelman Trust Barometer — the most extensive global survey of trust, now in its twenty-sixth edition — a figure appears which explains the scene without describing it.³ To the question of who still commands standing in the public's eyes, governments have lost sixteen points over five years, major news organisations eleven, and foreign business leaders six. Neighbours, family, and friends have gained eleven. It is the most brutal shift of trust ever recorded by this instrument: from We to Me, as the authors of the report put it. Or, more precisely: from institution to street corner. This is the hour of small markets. This is the hour of manual coffee machines and of vendors who know where the tomatoes come from.

What I want to recount in this essay is how we came to be here. How this immense shift was accomplished. How the digital platforms — conceived as bridges, as motorways, as a universal infrastructure of connection — managed to produce exactly the opposite: a general retreat into ever smaller, ever more closed, ever more suspicious circles. And, above all, what is lost when a civilisation begins to find its only source of trust at the vendor on the corner.

II. The Silent Mutation: From Application to Infrastructure

I have been using the word infrastructure rather freely. It deserves a moment of attention. In the classical sense, an infrastructure is what one does not notice until it breaks. Electricity, water at the tap, the road to the hospital, the telephone network — they operate so discreetly in the background that they constitute the conditions of daily life, not its objects. No one thinks in the morning whether the electricity will come; one thinks about it only when it does not. No one asks whether the water is potable; only when a notice arrives from the town hall about a burst pipe does one remember that someone, somewhere, maintains a complex system so that one may wash one's hands.

Infrastructure has a legal grammar of its own, acquired after centuries of conflict. The regime of common carriage — which obliges a railway to transport any freight at a public price and without discrimination — was invented in the nineteenth century precisely because it had been observed that, when a private actor controls an essential good on which the public depends, an imbalance of power arises which cannot be resolved by competition alone. A recent study from the Knight First Amendment Institute takes up exactly this argument for digital platforms: when a private actor controls the terms of access to an essential good on which the public depends for its flourishing, this places that actor in a position of arbitrary, dominating power over the public.⁴ Such power cannot be left unregulated without eroding the conditions of common prosperity.

The digital platforms have crossed exactly this threshold over the past fifteen years. The precise moment is difficult to date, but one can indicate its markers. When Facebook became the main channel through which an extended family communicates — grandparents in Romania with grandchildren in England, cousins in Spain with parents in Botoșani — the application ceased to be an entertainment product. It became social infrastructure. When Google Maps became the default means by which people find their way to the hospital, to the town hall, to a wedding, the application ceased to be a search tool. It became navigation infrastructure. When WhatsApp became the sole channel through which information is transmitted from the parish, from the school, from the tenants' association, the application ceased to be messaging. It became community infrastructure. When a small firm in Botoșani realised that, without a presence on Instagram, it could no longer reach customers under forty, the platform ceased to be a social network. It became commercial infrastructure.

Jean-Christophe Plantin and Aswin Punathambekar have called this process the infrastructural turn of platforms.⁵ What had begun as disruption — a word implying optionality, alternatives, market play — became substrate: something upon which other things are built, something upon which other activities depend. And Jo Pierson, in a 2021 study of messaging applications as European social infrastructure, has articulated the paradox that concerns us most: users feel obliged to appropriate these applications as part of daily life, whilst being made dependent on the corporate mechanisms of control over them.⁶ You cannot not be on WhatsApp, because your family is there; but being on WhatsApp places you inside the extractive machinery of a company which owes you nothing beyond the unilateral contract you accepted with a tick.

This mutation is silent because it did not happen through an act of parliament or in a single media moment. It happened through accumulation. As though at first you had wished to use an instrument, and after ten years you discovered that the instrument had become the condition of your work, of your relationships, of your civic identity. No one asked your consent for the leap from convenience to dependence. It happened simply. And when you wished to look, it was too late: dependence had already become infrastructural, yet the legal regime of infrastructure — with its obligations of neutrality, transparency, non-discrimination, and continuity — did not apply. You had infrastructure with the rights of an ordinary consumer. An equation which could yield only what it has yielded.

III. Enshittification: The Theory That Gave a Name to a Collective Intuition

In November 2022, the Canadian writer Cory Doctorow published a short piece on his blog in which he proposed a new term: enshittification. It is a vulgar word, deliberately vulgar, built on the verb which describes transformation into something without value. Doctorow repeated the term on stage, in articles, in interviews; he formalised it three months later in Locus magazine. In 2025 he published a book with that title at MDC Books, which has quickly become one of the most cited works on digital platforms of the decade.⁷

Why did this word stick so fast? Because it articulated what everyone had already felt: the internet no longer works as it once did, but no one could name the cause. Doctorow's formulation is brutal and clear: here is how platforms die. First, they are good to their users. Then they abuse their users to make things better for their business customers. Finally, they abuse those business customers to claw back all the value for themselves. Then, they die. The three acts of betrayal. It is a model with theoretical ambition, yet one which is verified empirically in almost any major platform of the past twenty years.

Amazon began as a customer-centred online bookshop — good recommendations, low prices, reliable delivery, easy returns. It then moved into the second phase: to serve third-party sellers, it introduced a system of paid advertisements which pushed organic products ever lower in search results. In the current, third phase, Amazon extracts fifteen per cent in commissions, fifteen per cent in advertising, and further sums in logistics — until the small seller discovers that he works for the platform, not the reverse, and the customer discovers that the first search result is not the best product but the one which paid most to appear there. Facebook has gone through the same sequence, except that its business customers have been the news publishers and the advertisers, both of whom are today in visible distress. Google Search is in an advanced phase of decay, in which the first four or five results are advertisements, followed by an AI summary which harvests the answer without traffic to the source, followed by automatically generated content which copies — badly — the few who still write anything original. Twitter/X compressed the three acts into record time under Elon Musk.

What is important in Doctorow's model — and what makes it relevant to our discussion — is that he does not describe the nature of platforms, but the conditions under which platforms behave in this way. Doctorow identifies four forces which have historically restrained companies: competition (if you enshittify too much, customers go to a competitor); regulation (if the state limits your practices); self-help by users (if users can solve their own problems through third-party tools, data migration, interoperability); and the power of technology workers (if engineers refuse to build systems of extraction). Over the past twenty-five years, all four have eroded. American antitrust was progressively abandoned from Milton Friedman and the Chicago school onwards. Regulation has remained perennially twenty years behind the technology. Interoperability has been destroyed through Digital Rights Management, through legal threats, through deliberate proprietary standards. And technology workers have gone through, over the past five years, massive waves of lay-offs which have considerably diminished their bargaining power.

What has remained, then? What has remained is an environment in which platforms can extract and nothing stops them. And when can and nothing stops them are the baseline conditions of a market, extraction becomes strategy, not exception. It is the mechanics of the market, not the psychology of executives. It is what Yanis Varoufakis has called, in a book of enormous impact over the past two years, technofeudalism: the migration of industrial capitalism into a regime in which the great platforms function as fiefdoms — digital territories in which the feudal proprietor (Amazon, Google, Meta) taxes every passage, every transaction, every unit of attention.⁸ It is a description tightly bound to enshittification: if enshittification is the process, technofeudalism is the institutional result.

IV. Splinternet: When the Global Infrastructure Breaks Into National Fragments

There exists a second process, parallel to enshittification, which aggravates everything and which changes the very nature of what we call digital infrastructure. It is the process which journalists and researchers have come to call, with a portmanteau as ugly as enshittification, the splinternet — from splinter and internet.⁹ The idea is simple and troubling: the single, global, open internet which my generation knew between roughly 1995 and, let us say, 2010, no longer exists. In its place there have appeared fragments — the Chinese internet (the Great Firewall, WeChat, Baidu, Weibo), the Russian internet (RuNet, VKontakte, Yandex, increasingly isolated after 2022), the Iranian internet (filtered almost entirely), the American internet (Google, Meta, Amazon, Apple, dominant in the West), the European internet (technically the same as the American, but under an increasingly divergent legal regime). Between these fragments, on a scale of decades, the digital borders grow ever more real — incompatible protocols, mutually banned applications, closed ecosystems, criminalised VPNs, submarine cables re-routed for geopolitical reasons.

The term is not new. It was formulated at the end of the 2000s as a prediction. Today it is description. TikTok has been banned or threatened with a ban in the United States, Canada, India, and in many European countries. Facebook and Instagram have been inaccessible in China for a decade. Twitter/X is blocked in Russia. Western applications are filtered or modified in the Gulf. The European AI Act imposes obligations which mean that some American AI models are launched with delay, or not at all, in the European market. DeepSeek, the Chinese AI model, prompted in 2025 a wave of bans across Western public administrations. Each large bloc — American, Chinese, European, and to a lesser extent Indian — is progressively building its own digital infrastructure, with its own rules, its own national-champion firms, its own inward-looking public procurement.

For our discussion of trust, the splinternet matters for three reasons.

First, because it demonstrates retroactively that the internet was never a neutral infrastructure. It was an American infrastructure which presented itself, for thirty years, as universal. And the other blocs, when they came to understand this, began to build parallel infrastructures — not because they are hostile to the idea of connectivity, but because they no longer accept that the substrate of their communication, commerce, health, and national defence should be controlled by four or five corporations from a single country. From the perspective of Beijing, of Moscow, or even of Delhi, dependence on American infrastructure is not technical neutrality — it is strategic vulnerability. And the public of these countries, when this is explained to them, tends to agree: it is better to have an imperfect national platform than a perfect foreign one which can be switched off by a decision of the White House.

The second reason is that the splinternet fatally complicates any conversation about regulation. The Digital Markets Act and the Digital Services Act of the European Union are courageous attempts to impose an infrastructure regime upon platforms.¹⁰ But if the platforms are American, the users are European, and the regulator is the European Commission, structural tensions arise which no treaty resolves entirely. Meta has repeatedly threatened, in recent years, to withdraw from Europe if certain rules are applied. Apple delays the launch of some AI features in the Union. Google modifies its search results differently depending on jurisdiction. The consequence is that the European user already lives in a kind of sub-internet — a partially neutered version of the original products, in which some functions are absent, others are modified, others appear with delay. It is a splinter experience which most users do not conceptualise, but they feel it daily.

The third reason, the most serious for the substance of this essay, is that the splinternet confirms to the public, at the geopolitical scale, exactly the same intuition which enshittification confirms at the everyday scale: there is no neutral digital infrastructure. Every platform belongs to someone. Every protocol serves someone. Every submarine cable passes through someone's waters. And if tomorrow the geopolitical relationship between that someone and your country turns sour, the infrastructure you thought you leaned upon can be cut off, filtered, sabotaged — or, worse, remain apparently functional whilst silently hostile.

The public reacts to this reality with the same reflex with which it reacts to enshittification: retreat. If the global infrastructure fragments into national blocs, and the national blocs no longer guarantee neutrality, then trust migrates not only from the great platforms to the small circle — it migrates also from the great global network into the local network. The WhatsApp group with the neighbours on the stairwell is more reliable than the global news network. The Telegram channel of the parish priest is closer than the YouTube channel of a Los Angeles influencer. The bulletin of the town hall — even the paper one, taped to the stairwell door — is more credible than a viral tweet. Geopolitical splinternet and cognitive insularity are, in the end, manifestations at two different scales of the same phenomenon: the retreat of trust from a global system which no longer merits trust, towards smaller, more familiar, more controllable units.

It is a disquieting symmetry. States fragment into blocs, the blocs fragment into national ecosystems, the national ecosystems fragment their public into circles of neighbourhood. Each level repeats the movement of the level above it. It is as though an entire civilisation were practising, simultaneously, the same discipline of reducing its radius of trust — from global to national, from national to local, from local to personal. At the end of the line, at the smallest scale, waits the vendor at the small market, with her La Cimbali, as the last credible institution.

V. The Perception of Facts in Practice: What Erosion Looks Like Day by Day

For those who live inside the digital infrastructure, erosion does not arrive as a theory but as a suite of small daily annoyances which, taken together, produce fatigue and, in the end, retreat. It is worth listing them, since each is a direct example of the mechanism described above.

The search engine which no longer searches. Open Google today and search for a simple recipe. The first three results are advertisements; there follows an AI summary which harvests the answer from three sources without attribution; there follows a blog article which opens with a thousand-word personal story about grandmother in Ohio, followed by the recipe itself buried beneath seventeen advertising banners; there follow five auto-generated pages which copy, with errors, the first results. The user's perception: search has become an obstacle between question and answer, not a tool which brings them together. What was once a service — has become a tollgate.

The social network which no longer connects. Open Facebook today and you will see, in this order: an advertisement, a post by a friend which the algorithm decided to show you three days late, a paid "news item" from a publisher trying to survive, an advertisement, a public group to which you never subscribed, an advertisement, an auto-generated sensationalist video, an advertisement. Friends glimpsed among advertisements as through fog. Cory Doctorow, speaking at the DemocracyXChange summit in the spring of 2026, put it directly: citizens are "trapped in systems designed to extract from them rather than serve them". Trapped is the key word. They cannot leave, because the community is there. But by staying, they are drained.

The online marketplace which sells fakes. Buy a charging cable on Amazon. There are fifteen options, apparently identical, with random names (BZKCHZO, JXRTYUI, KMPQWER), all with 4.5-star reviews. Three in five will be Chinese counterfeits which will break in six weeks. The "original" — thirty per cent more expensive — is recommended through a paid advertisement. Legitimate European sellers have progressively disappeared, since they cannot afford Chinese economies of scale plus the Amazon tax. The consumer pays more for ever poorer quality, in an infrastructure where he can no longer distinguish. It is, at a small scale, precisely what happens at the geopolitical scale — dependence on infrastructure whose components lie beyond borders one can no longer verify.

The advertiser who himself becomes raw material. This is one of Doctorow's finest observations. Advertisers were, in the second phase, the apparent beneficiaries of extraction from users. In the third phase they themselves become the merchandise to be squeezed. They are pushed towards "full automation" — "let the algorithm optimise your spend" — which simplifies buying but reduces transparency. The small advertiser pays more and more for an ever less verifiable click, in a system where the platform controls the measurement, the reporting, and access to the audit. The advertising market has become a casino without external audit.

The streaming which erases what you paid for. Disney withdraws series to avoid royalty payments. Warner Brothers removes entire films from the library for tax optimisation. Netflix quadruples its price over seven years whilst eliminating shared accounts and narrowing the catalogue. Doctorow observes that it used to be far simpler to find the right combination of streaming services; now it is a game of confusion in which you pay for access to content which may not be there tomorrow.

Here, had my father been alive, he would have uttered one of those short mechanic's sentences I can still repeat: "an engine you have repaired stays repaired. A service you have paid for may not exist tomorrow."¹¹ This is the fundamental distinction between the world of objects and the world of platforms. In the world of objects, your work sediments. In the world of platforms, your work is leased from a proprietor who can rewrite the contract retroactively. Feudalism in its classical form had at least the advantage of stability: the lord remained lord, the serf remained serf, but the labour contract — the tithe, the corvée, the rights to firewood — did not shift overnight. In digital feudalism, the contract is rewritten with each update. And in the age of the splinternet, the map can also change: the app you used yesterday may be illegal tomorrow, in this country or that, depending on which minister has signed which order.

VI. The Figures of Erosion: What the Barometers Say

All these observations — which might otherwise be dismissed as the mere lament of a fatigued user — are now confirmed by the most serious global survey of trust. The Edelman Trust Barometer has measured for more than twenty-five years, across nearly thirty markets, the state of public trust in the four major institutions — government, business, NGOs, media — and in the sectors that compose them. The 2026 edition, presented at the World Economic Forum in Davos in January, tells a story worth quoting in some detail.

For the first time in the history of the survey, Edelman finds that mistrust has moved beyond polarisation and reached insularity. Seven out of ten respondents worldwide report themselves reluctant or refusing to trust someone with different values, approaches, or sources of information. This insularity is highest in Japan (90 per cent), Germany (81), the United Kingdom (76), Canada (73). This is not merely bubble — the word that has been trivialised by overuse. It is a global-scale instinct of cognitive self-preservation: the public retreats into circles of familiarity, refusing, through mental economy, contact with alterity.

The causes of this retreat, as Edelman identifies them, are the five events which most affected trust over the past five years: inflation (54 per cent), the growing prevalence of misinformation (50 per cent), the COVID-19 pandemic (43), trade wars (37), and — the figure crucial for us — the increasingly extensive use of generative AI platforms (37). Digital platforms enter, then, explicitly and by name, among the forces which destroy public trust. And in the case of AI companies, the collapse is concrete: trust in them has fallen globally from 61 to 53 per cent in five years. In the United States, from 50 to 35 — a fifteen-point drop, comparable to the post-2008 collapses of trust in banks.

The migration of trust — formulated in the report as a movement from We to Me — has precise targets. National government leaders have lost sixteen points of net trust. Major news organisations have lost eleven. Foreign business leaders, six. In exchange, neighbours, family and friends have gained eleven. Add this figure to the observations on insularity and you obtain the complete portrait: the public is retreating from the large-scale institutions — which implicitly include the platform-infrastructures — and is recomposing a map of trust restricted to the circle of proximity.

There is a particularity of the figure on AI which merits emphasis. Distrust in AI is settling in before the AI infrastructure is fully built. Generative AI is still in a phase of technological expansion — ChatGPT, Claude, Gemini are only three to four years past their public launches — but the public reacts to it as to an already compromised service. It is, if you like, a collective immunity dearly acquired through the lessons of 2015–2025. The public has learned that the promise of universal infrastructure predictably converts into a unilateral contract rewritten. It no longer wishes to be a beta-tester for a contract that will be rewritten in two years' time. It is a form of popular wisdom, of the kind the Romanian village once expressed as "good watch turns bad danger aside", transposed into the age of AI.¹²

VII. Why Trust Has Not Evaporated but Migrated

Here intervenes the sociological mechanism which most concerns me. When an infrastructure becomes hostile, people do not simply abandon it — the dependence is too great. Facebook remains on the phone, Google remains the default search engine, WhatsApp remains the sole channel for speaking to the extended family, Instagram remains the only shop-window through which the small Botoșani firm reaches customers. But trust — the cognitive and affective resource which made a platform an infrastructure — is retreating. It is retreating somewhere. The Edelman data tell us exactly where: into the circle of neighbours, family, friends. Into the small market on the corner.

It is a paradoxical retreat, since it is being accomplished on the platforms that have produced the hostility. The WhatsApp group with four former university classmates has become more important than the main Facebook feed. The Telegram channel of the parish priest is more credible than the seven o'clock news. The neighbourhood group of thirty members — more relevant than the town hall. The platform has remained as a transport layer, but social meaning has migrated into an ever smaller, ever more closed, ever more impermeable compartment. The modern user does not use WhatsApp because he trusts Meta, but because he has no other option for speaking to his mother; and within WhatsApp he grants trust only to a cell of ten persons whom he knows personally.

It is, if we may risk the formulation, an interior splinternet. Just as the global internet fragments into national fragments, so the feed of each user fragments into affective fragments. The small WhatsApp circle is a little autonomous republic, with its own borders, its own rules of entry (explicit invitation), and its own filter against the "outsider". The user reproduces, at the scale of his personal screen, what states are doing at the scale of the continent. If, at the geopolitical level, the splinternet is a strategy of national sovereignty, at the personal level the interior splinternet is a strategy of cognitive survival.

Byung-Chul Han has described somewhere the loss of the Other as a mark of the digital age: we can no longer bear alterity, because we no longer have the psychic infrastructure for it.¹³ Horia-Roman Patapievici, in The Recent Man, described the same phenomenon from an entirely different vantage — the loss of the hierarchies of meaning as a consequence of late modernity.¹⁴ What neither anticipated is that platform-infrastructures would be the concrete mechanism through which this retreat is accelerated. Because the platform promises universal access but delivers an algorithm which encloses you in an ever tighter circle of resonances. And when the tight circle itself becomes hostile — because it is traversed by advertisements, bots, misinformation, extraction — the only refuge is lower still, in the truly small circle: those you have known your whole life.

It is, if you will, an involuntary nepsis. The Fathers of the Philokalia named nepsis the watchfulness of attention, the sobriety of a mind which chooses what to let in and what to repel.¹⁵ It is a spiritual discipline requiring years of exercise, a supporting community, and a pedagogy. The modern user practises today an involuntary and exhausted version of the same posture: he no longer lets in anything but what comes from the small circle, because everything else has hurt him too often. He is not retreating because he has arrived at discernment. He is retreating because he has been fatigued to exhaustion by a milieu that wanted to extract everything from him. The Fathers of the Philokalia would have understood the gesture, but would not have praised its origin. Genuine nepsis comes from freedom; the involuntary kind comes from the trauma of dependence.

Let it be said, in fairness, that this retreat has also a healthy dimension. The rediscovery of neighbour, of extended family, of parish, of tenants' association, of small markets — these are goods in themselves. They reconstruct, at a small scale, the social fabric which the ideology of digital globalisation had weakened. The WhatsApp group with the parents from your child's school is a form of real associativity, producing concrete mutual aid: one borrows a car, shares a recipe, recommends a good doctor. It is, mutatis mutandis, what Tocqueville was so astonished to discover in the America of the 1830s as the source of democratic vitality — the small, voluntary, free associations. If retreat into the small circle produced only this, we could celebrate.

VIII. What Is Lost: The Hidden Cost of Insularity

But it does not produce only this. The retreat of trust from infrastructure carries a cost which the small circle cannot cover, and it is important to name this cost precisely.

Hard public decisions — health, education, collective safety, defence, environment — require a fund of trust beyond the circle. Vaccination, the observance of traffic laws, the payment of taxes, engaging a doctor one does not know personally, boarding an aeroplane flown by strangers, crossing a bridge built by unknown firms — all presuppose an institutional trust that neighbourhood cannot replace. If I inform myself about a medical treatment exclusively from my neighbour, or about a candidate exclusively from a Telegram channel of the parish, or about a law exclusively from a brother who works in the ministry, I am withdrawing not only from the hostile digital infrastructure but from the complex society in which hard problems are managed. I arrive, at the individual scale, at a form of cognitive primitivism — I accept only what comes from the small circle, and I reject the rest as up to them, an expression which in Romanian already denotes a form of civic resignation.¹⁶

It is what Ion Luca Caragiale, with his keen eye for Romanian types, described in his Sketches — although he had not yet the modern vocabulary of institutional trust.¹⁷ The suveica — the "shuttle-back-and-forth" — among the near ones as a substitute for institution, the favours circulating among acquaintances in place of predictable public service, the information passed by word of mouth because "the paper lies anyway". All these are comforting, they produce social warmth, but they also produce a society which can no longer take large decisions. A society in which every extended family is a little isolated republic cannot build a motorway, cannot reform a health system, cannot fight a war of defence, cannot respond to a pandemic. Insularity, however emotionally comfortable, is collectively pathological.

At the geopolitical scale, the equivalent phenomenon occurs. States withdrawing into their national splinternets gain apparent sovereignty and lose real capacity for coordination. A splinternet that seals itself tight may protect against external disinformation, but it also forbids scientific dialogue, academic collaboration, cultural exchange, humanitarian flows. And at the limit, hostile digital blocs become hostile political blocs — and political hostility historically prepares military hostility. It is a consequence few discuss openly, but it hovers over any conversation about digital fragmentation: divided technical infrastructure is divided political infrastructure, and this is the antechamber of graver confrontations.

Here lies the risk: retreat into the small circle is not health recovered, it is the symptom telling us that the public infrastructure — including the digital one — has failed. It is the fever telling us that the social organism is ill, not the sign it has healed. And if we do not reconstruct the infrastructure in a regime that merits trust, the small circle will tighten ever more, until it can no longer sustain even the interior lives it contains. For the interior lives — family, friendship, neighbourhood — depend also, in the last analysis, on a framework of public safety and predictability which they cannot produce alone.

There is a fine line, here, between what we might call healthy localism and pathological insularity. Healthy localism preserves the small circle as a source of meaning while remaining open to participate in larger structures. Pathological insularity refuses the larger structures and remains closed in the circle until it suffocates. The sign of passage from the first to the second is totalising distrust — the moment when one no longer believes anything that comes from outside, regardless of source. The Edelman data suggest that we are approaching, as a civilisation, this passage. And if we accomplish it, the consequences will be far graver than the discomfort of a Google search saturated with advertisements.

IX. What a Trustworthy Digital Infrastructure Would Mean

The question is not easy to answer, but neither is it impossible to formulate. Cory Doctorow proposes, at the end of the argument, a combination of four reactivated forces: competition through antitrust, regulation through frameworks such as the European DMA and DSA, self-help through mandatory interoperability and data portability, and the renewed power of technology workers. Interoperability, in particular, is the key — because it reintroduces the right of exit: if I can migrate my data, contacts, history to another platform in one click, then the original platform can no longer hold me hostage. The system becomes competitive by its very legal design.

The European Union has taken, through DMA, DSA, and the AI Act, an important step in this direction. These are the first legislative texts in the world which implicitly recognise the infrastructural nature of the great platforms and attempt to impose the correlative obligations: algorithmic transparency, data portability, interoperability of certain essential services, independent audit, moderation duties. It remains to be seen how much of the letter of these texts will survive practical application, which will be borne by national courts at variable rhythms. But the orientation is correct: digital infrastructure must enter a legal regime akin to that of classical infrastructure, with explicit public obligations.

Here, however, arises the splinternet dilemma. European regulation is, for some, a legitimate act of sovereignty; for others — including the American corporations and some governments — one more step towards the fragmentation of the global internet. Each bloc that imposes its own rules produces one more shard. How do we exit this paradox — how do we regulate infrastructure without contributing to its definitive fragmentation? One possible answer, articulated by a strand of European doctrine, is internationally coordinated interoperability: common standards, obligatory for all major platforms regardless of origin, negotiated through multilateral instruments similar to those in telecommunications. It is a vast ambition, presupposing a geopolitical cooperation the current climate hardly offers. But it is the only path that avoids the double trap: unregulated extractive platforms on the one hand, and hostile digital blocs on the other.

There is also a dimension Doctorow's programme touches only tangentially: the cultural dimension. The reconstruction of a trustworthy digital infrastructure is not accomplished by law and interoperability alone. It presupposes that the public will accept, once again, to grant trust to a complex public infrastructure — opaque by its technical nature, but transparent by its legal regime. This is something which cannot be achieved by communication campaigns, but by repeated experience of non-disappointment. And this takes decades. Not years. Decades.

It is possible, then, that we live in a period of long transition, in which the small circle will remain the reserve of meaning, and the reconstruction of public infrastructure — digital included — will be the labour of twenty or thirty years. It is a sober perspective, not a desperate one. We are not at the end of a road; we are in the middle of it. The middle of the road, the hardest, is the one that requires maturity: not to be seduced by nostalgia for the infrastructure lost (which was never quite as good as we remember), nor by substitute utopias (Fediverse, radical self-hosting, complete withdrawal), nor by insular resignation ("well, that's the world now"), nor by the temptation of the hard splinternet ("let us build our platforms with our rules, and leave the rest of the world with theirs").

X. Closing: The Utterance of the Machine

I close where I began. At the small market on the corner, at the La Cimbali with its scratched chrome plate, at the vendor who presses the lever with a hand that knows. This place is not the answer to the problems of digital infrastructure. It would be ridiculous to say so. But it is something more important: it is a sample. It is an example, preserved in the life of the neighbourhood, of what it means for trust to function without being requested. A microcosm that, by its very existence, reminds us that it is possible to build environments of exchange — commerce, information, meaning — in which the actors comport themselves in such a way that trust comes of itself.

If this can be done in small commerce, it can be done on larger scale too. Not easily, not quickly, not without strong institutions. But it can be done. And if we come, one day, to rebuild the public digital infrastructure in a form worthy of trust, the model we shall use is, in its essence, that of the small markets. Not by dimension — public infrastructure will always be large — but by moral disposition: the place where what is on offer does not lay traps for you, where the best product is the first in the window because it is the best, where those who work in it comport themselves in such a way that trust does not need to be defended.

My father, who was a mechanic, would have put things more simply. He would have said that a good engine does not ask you to trust it — it comports itself in such a way that trust comes of itself. A bad engine will leave you on the road after fifty kilometres, no matter how many owner's manuals you have read beforehand. The fundamental distinction is not between trust granted and trust denied. It is between the things that merit trust and the things that ask for trust without meriting it.

Those who built the digital infrastructure of the past twenty years have inverted this order at every turn. They have demanded trust — through ticks on terms and conditions, through corporate value declarations, through branding campaigns, through mission statements painted on the walls of Silicon Valley offices — and have tried to compensate, through an excess of rhetoric, for what was lacking in systemic conduct. The public has, in the end, understood the equation. It has retreated. It is at the small market now, buying tomatoes from Cristești and coffee from the La Cimbali. It talks with the neighbours. It looks warily at the screen of the telephone, on which the platform-infrastructures continue to operate — opaque utilities it tolerates but no longer loves. And at the geopolitical level, the continental blocs are retreating each into its splinternet, preparing for a world in which the single internet will remain a memory that only those beyond forty will know how to tell.

Much patience will be needed, much institutional work, many decades of non-disappointment, for the great digital infrastructure once more to merit the trust which has, today, migrated to the corner of the street. I do not know whether we shall succeed. I hope. But I know that until then, the small markets remain — modest, everyday, indispensable — the places that keep the memory of a world in which the things that merit trust stood side by side, on a counter, under a canvas awning, next to a coffee machine that still works.

And if someone still says, in passing, that the tomato from the corner tastes of coffee, let us not correct him. He speaks a language older and truer than most of the great platforms — and most of the splinternets — know how to speak.

Petru Cojocaru