Le prompt devient une forme de pensée, pas juste une commande

 Qu'est-ce qu'un prompt ? Techniquement, c'est une instruction en langage naturel adressée à un modèle d'intelligence artificielle (IA). Mais cette définition, exacte, est aussi trompeuse : elle laisse croire qu'écrire un prompt reviendrait à taper une commande dans un terminal, à appuyer sur un bouton avec des mots. Or un prompt, c'est d'abord une question, et une question bien posée est déjà la moitié d'une pensée.

Celui qui demande « écris-moi un texte sur l'innovation » obtient ce qu'il mérite : un texte sur l'innovation, poli, correct, interchangeable. Celui qui demande « imagine comment un horloger du XVIIIe siècle aurait conçu une application de méditation » obtient autre chose : une tension, un décalage, une idée qui n'existait pas avant la question. La différence ne tient pas à la technique, mais à l'esprit qui formule. Le prompt n'est pas un ordre donné à une machine, c'est une pensée qui se rend visible pour mieux se déployer.

Écrire un bon prompt, c'est en effet faire un travail intellectuel exigeant. Il faut savoir ce qu'on cherche, ou du moins reconnaître ce qu'on ne cherche pas. Il faut choisir un angle, poser des contraintes, oser des rapprochements improbables. Il faut, surtout, accepter que la première réponse n'est qu'un point de départ. Le vrai geste créatif commence ensuite : rebondir, contredire, détourner, relancer. Le meilleur usage de l'IA ressemble moins à une commande qu'à une conversation, et une conversation n'est fertile que si l'on a quelque chose à y apporter.

C'est ici que l'IA révèle sa véritable nature : un amplificateur. Elle donne à chacun le pouvoir d'explorer en quelques minutes des dizaines de directions qui auraient exigé des jours de recherche, d'essais et de brouillons. Elle libère du temps, et ce temps est précieux, mais uniquement pour ceux qui savent quoi en faire. Un amplificateur ne crée pas le signal, il grossit celui qu'on lui envoie. Une intention floue donne un résultat flou en plus grand. Une intuition singulière donne un résultat singulier, mais en plus riche.

Il y a par ailleurs une seconde compétence, moins visible que l'art de formuler : celle du goût. L'IA produit sans effort du plausible, et le plausible est exactement ce qui menace la création, car il ressemble à du travail fini. Face à vingt propositions, quelqu'un doit reconnaître celle qui a du relief, écarter les évidences, sentir ce qui sonne juste et ce qui sonne creux. Ce discernement ne se génère pas. Il se construit avec le temps, par la lecture, l'expérience, l'observation du monde et l'exigence envers soi-même. Aucun modèle ne peut le déléguer à votre place.

On entend souvent que l'IA va niveler la créativité, que tout finira par se ressembler. Le risque existe, mais il concerne ceux qui s'en tiennent à la première réponse, ceux qui utilisent l'outil comme un distributeur. Pour les autres, ceux qui questionnent, détournent, croisent les domaines et assument un point de vue, l'IA devient un partenaire d'exploration. Elle élargit leur terrain de jeu sans jamais choisir à leur place où planter le drapeau.

L'histoire des outils est d'ailleurs instructive. La photographie n'a pas tué la peinture, elle l'a obligée à se réinventer. Le synthétiseur n'a pas éliminé les musiciens, il a produit de nouveaux compositeurs. À chaque fois, la technique a démocratisé l'accès, et à chaque fois, ce sont les esprits les plus inventifs qui ont fait la différence. Non parce qu'ils maîtrisaient mieux l'outil, mais parce qu'ils avaient quelque chose à dire, et une façon à eux de le dire.

Il en ira de même avec l'IA. Elle mettra à la portée de tous des capacités autrefois réservées à quelques-uns, et c'est une bonne nouvelle. Mais quand tout le monde dispose du même moteur, l'avantage se déplace ailleurs : vers la curiosité, l'audace, la sensibilité, la capacité à voir ce que les autres n'ont pas vu. Ces qualités ne sont pas dans le modèle, elles sont dans celui qui l'interroge.

Avec ou sans IA, les esprits les plus créatifs feront donc toujours la différence. Parce que la création n'a jamais été une affaire d'outils, mais de regard. Et un regard, aucune machine ne peut le prêter.

Why the Best Movies Demand a Second Date

On my first viewing of Paul Thomas Anderson’s Phantom Thread, I thought I was witnessing a tragedy. Here was a story of a brilliant, suffocating mid-century couturier, Reynolds Woodcock, and his young muse, Alma, caught in a toxic spiral of codependency. The air in the theatre felt heavy, thick with the tension of a psychological thriller. I left the cinema feeling as though I had just watched a beautifully tailored car crash.

Then, a year later, I watched it again.

Suddenly, the rustle of the silk wasn’t tense; it was a setup. The rigid, immaculate townhouse wasn’t a prison; it was a stage. When Alma scraped her knife across a piece of toast with the force of a chainsaw, violating the sacred silence of Reynolds’ breakfast, I didn’t gasp—I laughed out loud. Knowing how their twisted love story ends flips a switch in the viewer’s brain. It transforms Phantom Thread from a stately period drama into one of the funniest, most romantic black comedies ever put to celluloid.

This realisation isn't unique to film critics; it is a gospel fiercely preached by the comedy community itself. Look closely at who keeps Phantom Thread on a permanent loop, and you will find an army of stand-ups, sitcom writers, and masters of cringe humour. Comedians have long treated the film not as an Oscar-bait drama, but as an elite, 1950s period-accurate episode of Curb Your Enthusiasm. To a comedian, there is nothing funnier than the absolute grandeur of unchecked neuroticism.

But the engine that actually drives this dark comedy isn't Daniel Day-Lewis's tantrums—it is the astonishing, deadpan brilliance of the film's actresses. On a first watch, you are conditioned to look at Day-Lewis, a titan of mythic Method acting, and assume he is the sun around which the film orbits. On a rewatch, you realise he is merely the straight man to a pair of comic assassins: Vicky Krieps and Lesley Manville.

As Cyril, Lesley Manville delivers a performance of weaponised stillness. She doesn't raise her voice; instead, she shuts down her brother's ego with the terrifying, icy precision of a seasoned sitcom matriarch. Her delivery of lines like, "Don't start a fight with me, you'll lose," is so perfectly timed it behaves like a punchline without a laugh track.

Meanwhile, Vicky Krieps’ Alma undergoes the most radical transformation upon repeat viewings. Initially misread as a submissive, fragile victim, a second look reveals her as a master of passive-aggressive psychological warfare. Krieps plays Alma with a subtle, chaotic irony; the tiny, knowing smirks she flashes while watching Reynolds suffer through a buttered-up dinner are the true comedic heartbeat of the film. Krieps and Manville don't just hold their own against cinema's most famously intense actor; they systematically dismantle his dignity for laughs.

This is the magic of the cinematic rewatch, a cultural habit we often dismiss as mere comfort-seeking. We live in an era of content abundance, where streaming algorithms aggressively push us toward the new. To choose to spend two hours with a movie you have already seen is treated as a minor failure of curiosity. But some movies are simply too dense, too layered, and too clever to be understood on a first date.

A first viewing of any great film is burdened by logistics. As viewers, our brains are doing the heavy lifting of narrative management. We are tracking the plot, trying to predict twists, and assessing the stakes. We are asking baseline questions: Will she survive? Is he the villain? What is in that tea?

A second viewing, however, offers the ultimate luxury: liberation from the plot.

When you already know the destination, you can finally enjoy the scenery. You can see the chess match for what it really is. By the time the film reaches its final, pitch-black punchline—where poisoning your partner with wild mushrooms becomes the ultimate, bizarrely sweet romantic compromise—the transformation is complete.

There is a distinct joy in discovering that a movie is smarter than you gave it credit for on night one. Phantom Thread is a masterpiece precisely because its truest, funniest secrets are stitched invisibly into the lining, waiting for a rewatch to be turned inside out. So, skip the "Trending Now" tab tonight. Go back to something you’ve already seen. You might just find a completely different comedy waiting for you.

We Borrowed Singapore's Immigration Policy and Forgot to Borrow the Country That Makes It Work

Every time a minister needs to defend another quota of imported labour, Singapore gets wheeled out. It is Mauritius's wet dream: a rich, disciplined, foreign-worker-dependent city-state that supposedly proves the model works. What nobody mentions in the same breath is that Singaporeans, on the whole, do not want to leave. Mauritians, in striking numbers, do.

That gap is the whole story, and we are not allowed to look at it directly.

Mauritius now has some 35,000 migrant workers, over 80 percent of them men, filling construction sites, factories and private kitchens. That is presented as a technocratic necessity, the same vocabulary Singapore uses for its own much larger foreign workforce. But necessity is doing a lot of quiet work in that sentence. It forecloses the question of why local wages in those sectors never rose high enough to make Mauritians want the jobs, and why two decades of "temporary" labour importation never produced the training pipelines that were supposed to make it temporary.

Meanwhile, over half of Mauritius's tertiary-educated population has left the country. In 2023 alone, more than 5,000 young Mauritians moved to Canada. Nurses, engineers, IT graduates: trained at public expense, gone within a few years of qualifying. We import unskilled labour at one end while our own skilled labour drains out the other. It is not a model. It is a leak dressed up as a strategy.

Singapore does not have this problem, and the reason is not culture or size or luck. It is institutions and leadership.

Start with corruption, because it is the least abstract of the four measures. Singapore ranked third out of 180 countries in Transparency International's 2025 Corruption Perceptions Index, with a score of 84. Mauritius ranked 61st, with a score of 48 — down five places and three points in a single year, the lowest score the country has recorded since 2012. Transparency International's own assessment attributed the slide to a weakening of democratic checks and balances. That is not a rounding error. That is two different species of state.

Then there is the currency, which is really a referendum on decades of policy discipline. The Singapore dollar has appreciated against the US dollar for most of the last thirty years, a deliberate outcome of a monetary authority built to protect the purchasing power of Singaporean wages. The Mauritian rupee has done the opposite for just as long, quietly taxing every citizen's savings and imported goods while nobody has to vote on it. A currency that loses value year after year is not neutral. It is a slow, invisible transfer from ordinary households to whoever benefits from a cheaper rupee.

This is also where the real story hides, because the headline numbers are the wrong numbers. What matters is not income in absolute terms but real disposable income: what is actually left in a household's pocket after tax and inflation, measured in something that keeps its value. A Mauritian salary quoted in rupees looks like it has grown handsomely over the past decade. Convert that same salary into a currency that has not been quietly devalued, and the gain shrinks considerably, sometimes to nothing at all. A European retiree or South African executive earning in euros or a currency that holds its value is not just nominally richer than the Mauritian professional sitting next to them. They are structurally richer, because their income was never designed to erode. Two people can hold the same job title and live in entirely different economies simply because of which currency their pay slip is denominated in.

That currency gap is precisely what is now reshaping the housing market, and this is the part policymakers would rather not discuss in the same paragraph as "FDI success story." Mauritius has spent two decades actively marketing residential property to foreigners through the IRS, RES, Smart City and now PDS schemes, each one a deliberate channel for sourcing foreign direct investment through real estate. It has worked, in the narrow sense that it was designed to work: villas and flagship apartments now sell in the hundreds of thousands of US dollars, foreign buyers are queuing for permanent residence at the $375,000 threshold. Meanwhile the country's Residential Property Price Index rose nearly 14 percent in a single year, and Mauritians earning in a depreciating rupee are bidding against buyers whose money was earned, and holds its value, somewhere else entirely.

This is gentrification with a currency mechanism attached. It is not simply that foreigners have more money. It is that the exchange rate manufactures the gap and keeps widening it every year the rupee slides, so that a stable expat salary buys steadily more Mauritian real estate while a stable Mauritian salary buys steadily less of it. Young Mauritian professionals, the same cohort the country is desperately trying to keep from emigrating, now find themselves priced out of their motherland, watching a housing policy explicitly built to attract outsiders' capital do exactly what it was built to do, at their expense.

Meritocracy is harder to score, but not hard to observe. Singapore built a civil service around competitive recruitment and salaries designed to pull talent into government rather than out of it, whatever one thinks of the trade-offs that entails. Mauritius's public sector has spent the same decades associated, fairly or not, with political patronage and post-election reshuffling. Ask any Mauritian graduate which system they would rather build a career inside, and the answer is generally the one they are trying to emigrate to.

Institutional stability is the hardest to reduce to a single number, but recent history has not been kind: a string of scandals implicating state bodies have left many Mauritians with the sense that the rule of law bends depending on who you know. Singapore's reputation for boring, predictable, rule-bound governance is precisely what allows it to run an aggressive labour-import policy without hollowing out its own citizens' loyalty to the state. Predictability is not a personality trait. It is the product that institutions exist to deliver.

And here is the detail that should embarrass anyone still reaching for the Singapore comparison: Mauritian workers have not even been unproductive. Labour productivity has posted real gains across recent years, and the minimum wage has risen sharply in nominal terms since 2018. On paper, the workforce has been holding up its end of the bargain. But productivity gains are supposed to convert into higher real wages, which are supposed to convert into a rising standard of living relative to the cost of the assets people actually need, chiefly housing. 

In Mauritius, that chain keeps breaking at the currency link. Productivity goes up, nominal wages go up, and the rupee's slide combined with a housing market priced partly in hard currency quietly cancels the gain before it reaches anyone's real standard of living. Singapore's productivity gains show up in strengthening purchasing power and rising home ownership among citizens. Mauritius's productivity gains are being absorbed somewhere between the payslip and the property listing, and nobody in government seems especially eager to explain where. A stark paradox has emerged in the workplace: while statistical productivity is up, employee well-being has stalled, and widespread disengagement is now palpable.

Put these four factors together and the borrowed justification collapses. Singapore imports labour and simultaneously keeps building the platform under its own citizens — housing, savings schemes, an education-to-wage pipeline that actually closes the loop, a currency that holds its value, a bureaucracy citizens can trust. Mauritius has adopted the first half of that sentence and left out everything after the comma. We get the labour-market flexibility. We do not get the payoff that made the flexibility sustainable in Singapore in the first place.

The result shows up not in GDP figures, which can look perfectly respectable, but in something GDP cannot capture: whether people want to stay. Mauritius's own emigration figures, alongside international brain-drain indices that repeatedly place the country among the world's worst for retaining graduates, are the real referendum on whether the "Singapore model" is actually being built here. A country whose most capable citizens are voting with their passports is not running Singapore's playbook. It is running its own experiment, and the early results are in.

If policymakers want to keep citing Singapore, they should be made to cite the whole file: the corruption index, the currency chart, the civil-service exam, the housing policy that keeps citizens inside the market rather than pricing them out of it, the decades of institution-building that had already happened before Singapore ever needed a single foreign construction worker. Otherwise, stop invoking a country we have no intention of actually resembling.

Sagesse et raison : une dette occidentale mal assumée

 L'Occident moderne s'est construit sur le culte de la raison pure. Depuis les Lumières, l'élite occidentale a valorisé la logique, la mesure et la maîtrise technique du monde, reléguant au second plan les formes de connaissance intuitives, corporelles ou sacrées. Ce choix a produit un progrès matériel indéniable — mais il a aussi laissé un vide : celui du sens, de la mort, de l'angoisse existentielle, questions auxquelles la raison instrumentale ne répond pas.

De ce vide est né un mouvement inverse : la quête occidentale vers le bouddhisme, le soufisme, l'hindouisme, perçus comme des réservoirs de sagesse que le rationalisme aurait asséchés chez soi. Cette aspiration est sincère. Mais elle reste souvent prisonnière de l'esprit même qu'elle cherche à fuir. On y retrouve un même geste rationaliste : extraire la technique utile — méditation, respiration, posture — et abandonner ce qui l'entoure : cosmologie, communauté, discipline de vie, transmission. C'est ce qu'on appelle l'appropriation spirituelle, moins l'inverse du rationalisme que sa prolongation, appliquée cette fois à un matériau exotique. Le capitalisme de consommation en est le vecteur le plus visible, transformant ces sagesses en produits détachables ; l'individualisme hérité des Lumières et du protestantisme en est la condition de fond.

Ce mécanisme n'est pourtant pas né avec le tourisme spirituel du XXe siècle. Il est bien plus ancien, et l'Occident classique lui-même en offre un exemple frappant : les Fables de La Fontaine. Le fabuliste s'inspire ouvertement d'Ésope, mais une partie de son œuvre puise aussi dans le Pañchatantra, recueil de fables sanskrites parvenu en Europe via des traductions arabes et persanes. Des textes comme La Tortue et les Deux Canards ou Le Rat et l'Éléphant viennent directement de cette source indienne. Or cette dette a longtemps été minorée : l'imaginaire scolaire français a retenu La Fontaine comme héritier d'Ésope, laissant la filiation indienne dans l'ombre. Plus encore, le contenu oriental — souvent lié à une pensée du karma ou de la ruse propre aux sagesses populaires indiennes — a été recadré dans une morale rationaliste et sociale typiquement classique : la fable devient un outil pour instruire la cour et critiquer les mœurs, non pour transmettre une cosmologie.

L'exemple est révélateur : garder la forme narrative utile, effacer le fond spirituel, n'est pas un réflexe propre à la mondialisation contemporaine. C'est un geste structurel, profondément enraciné dans la manière dont l'Occident classique a toujours digéré ce qu'il empruntait à l'étranger. L'appropriation spirituelle d'aujourd'hui n'invente donc rien : elle perpétue, avec d'autres outils, une habitude bien plus ancienne.

The Babywalker Economy: Mauritius Is Running Out of Crutches

 For thirty years, Mauritius was Africa's success story — the small island that got rich without the resource curse, the coups, or the chaos. But look past the brochure and a less comfortable story emerges: much of that success was borrowed, not built, and the bill is arriving.

Start with what's easy to verify. Transparency International's Corruption Perceptions Index scores public-sector integrity across more than 180 countries on a 0-to-100 scale, where 100 is very clean and 0 is highly corrupt. Mauritius's score tells a clear story of decay: it peaked at 57 points in 2012, held roughly steady through the early 2020s, then fell to 51 in 2024 and to 48 in 2025 — its lowest score since 2012, and the sharpest single-year drop the country has recorded in over a decade. This isn't just a matter of other countries improving and Mauritius standing still; the underlying score itself is falling, meaning the actual perceived integrity of the public sector is getting worse in absolute terms, not just relative to peers. (Its global rank has slipped accordingly, from the low 40s a decade ago to 61st in 2025, but the score is the more telling number — it isn't affected by how many countries are added to the index each year.)

The miracle ran on crutches, not muscle

Mauritius's post-independence growth was real, but it was substantially manufactured by two mechanisms that had nothing to do with underlying competitiveness. The first was preferential trade access — guaranteed sugar quotas to Europe, then textile quotas under the Multi-Fibre Arrangement, that let Mauritian exporters sell into rich markets without having to compete on price or quality against the rest of the world. The second was currency management: periodic rupee depreciation intended to restore export competitiveness by making Mauritian goods cheaper in dollar terms, without requiring a single factory to become more efficient.

But even this second crutch was weaker than it looked. Mauritius imports most of its raw materials, machinery, fuel, and intermediate inputs — the textile sector, for instance, imports much of its fabric and yarn. When imported inputs make up a large share of production costs, depreciation raises those input costs in local-currency terms at the same time as it's supposed to be making exports cheaper abroad, and the two effects partly cancel out. The competitiveness gain is smaller and shorter-lived than the policy assumes, while the pass-through to domestic inflation is immediate and real. In other words, one of Mauritius's two main levers for staying competitive was less effective than advertised even while it was being used.

Both tools are legitimate parts of any small open economy's toolkit. The trouble is what happens when they become the primary strategy instead of a bridge to something else. A country can look competitive for decades this way while the underlying capacity to compete without the crutch — productivity growth, technological upgrading, workforce skills, efficient logistics, deeper local supply chains — never actually gets built. The trade preference or the weaker currency absorbs the cost of inefficiency instead of forcing anyone to fix it.

Why this produces exactly the pathologies critics complain about

This is the part that connects a trade-policy story to a governance story. Firms that face real competitive pressure have no choice but to cut costs, modernize, and demand efficient, predictable regulation — because inefficiency shows up directly in their margins. Firms operating behind quota walls and a managed currency face no such discipline. They can absorb rent-seeking, weak enforcement, and cronyism indefinitely, because something else is quietly paying for the inefficiency.

Over enough decades, that dynamic doesn't just tolerate a certain kind of elite — it selects for one. It rewards people who are skilled at securing and defending privileged arrangements, not people who are skilled at building competitive enterprises. And once that elite exists, its incentives run in a consistent direction at every level of the system: rent-seeking at the top, in finance and real estate and licensing, where privileged access is worth the most; corner-cutting in the middle, in planning permits and labor inspections and procurement, because enforcement is something that can be quietly relaxed for the right people once competitive pressure isn't there to punish the laxity; and fence-sitting at the political level, where a small number of families have rotated power for decades and no coalition partner has much incentive to prosecute another.

The crutches are being taken away

None of this was fatal as long as the crutches held. But both are failing simultaneously. Preferential trade access has eroded steadily since the WTO liberalized global trade and the Multi-Fibre Arrangement quotas expired. Currency depreciation was never as effective as it looked given the import-heavy cost base, and its returns diminish further once markets and domestic actors price it into their expectations — leaving mostly the inflationary cost without the competitiveness benefit. At the same time, Mauritius's offshore financial sector — one of the newer pillars of the growth model — has drawn its own scrutiny, including a period on the FATF's watchlist over concerns that its tax treaty network functioned as a conduit for capital seeking to avoid taxation elsewhere.

Strip away the trade preferences and the currency trick at the same moment, and what's left is whatever institutional capacity the country actually built during the years it didn't need to. On the evidence of the score's decline, that capacity was thin. It was never load-bearing, because it was never load-tested.

Seychelles and Cabo Verde started from similar positions and chose, over the same twenty-year window, to build enforcement capacity and diversify their economies. Mauritius chose, largely, to protect the coalition arrangements and preferential deals that had worked before. The falling score is what that choice looks like once you can no longer put it off.

The Tax Haven Double Standard: Why Mauritius Gets Blacklisted While London, Geneva and Luxembourg Get a Pass

Every serious ranking of global financial secrecy tells the same inconvenient story: the world's biggest enablers of tax avoidance and financial opacity are not small tropical islands. They are the United States, Switzerland, Singapore, Luxembourg and the United Kingdom's own financial network. On the Tax Justice Network's Financial Secrecy Index, these are consistently among the five biggest suppliers of financial secrecy on Earth. Mauritius, by contrast, sits far down the list — around the 54th spot globally, well behind Egypt, Nigeria and Kenya, let alone the traditional secrecy giants.

And yet it is Mauritius that has spent the last several years fighting to shed the "tax haven" label — lobbying the EU, negotiating with the FATF, and rebranding itself at every opportunity as an "international financial centre." Switzerland and Luxembourg, meanwhile, have worn the label for decades with something close to indifference. London barely bothers to argue.

This is not a story about who is more guilty. It is a story about who gets to write the rules — and who has to live by them.

The numbers don't support the reputation

By the technical measures researchers actually use — effective corporate tax rates, banking secrecy, beneficial ownership opacity, treaty shopping facilitation — Mauritius is not meaningfully worse than the jurisdictions that face far less scrutiny. If anything, the index data suggests the opposite: Luxembourg and Switzerland have ranked among the world's very top secrecy jurisdictions for years, described by researchers as jurisdictions that "sell secrecy services at scale" rather than places accidentally caught up in wrongdoing. The UK's own offshore network — the City of London plus Jersey, the Cayman Islands and the British Virgin Islands — has repeatedly topped the same rankings.

Mauritius was formally grey-listed by the EU between 2019 and 2021 and flagged by the FATF over anti-money-laundering gaps. It fought hard, made reforms, and was eventually removed from both lists. Switzerland and Luxembourg have never faced comparable blacklisting, despite scoring as high or higher on secrecy metrics throughout.

Why the discomfort is so unevenly distributed

Who writes the rules, and who gets judged by them. The EU, OECD and FATF are the bodies that produce the blacklists — and they are dominated by the very countries whose financial sectors would qualify for those same lists. Luxembourg is an EU member state; it is never going to blacklist itself. The UK helped design the modern anti-money-laundering architecture. Mauritius, by contrast, is a rule-taker. It gets evaluated by a system it had no hand in building.

A double standard with an uncomfortable undertone. Researchers who study these rankings have pointed out something worth saying plainly: Global North financial centres tend to get gentler language — "competitive tax regime," "wealth management hub," "financial centre" — while smaller, often non-white-majority jurisdictions doing structurally similar things get "tax haven" and "high-risk jurisdiction." The practices are comparable. The vocabulary, and the consequences, are not.

Economic dependency changes the stakes. Switzerland and Luxembourg have deep, diversified economies where financial services is one pillar among several, backed by strong currencies and, in Luxembourg's case, full EU market access. For Mauritius, offshore finance is a much larger share of GDP and state revenue. A blacklisting is an inconvenience for Zurich. It is closer to an economic emergency for Port Louis.

Historical inertia protects the incumbents. Swiss banking secrecy and Luxembourg's holding-company regime predate the modern scrutiny regime by decades. By the time the FATF, the EU and the post-2008 G20 crackdowns arrived, these jurisdictions were already too embedded in global finance to sideline without disrupting the system itself. Mauritius built its offshore sector mainly from the 1990s onward — late enough to be judged by a rulebook the older players never had to answer to when they were establishing themselves.

The India factor. Mauritius's offshore model was built substantially around treaty-based investment routing into India, formalised through a double taxation avoidance agreement. India has periodically renegotiated or threatened to renegotiate that treaty specifically because of "treaty shopping" concerns — a pressure Switzerland and Luxembourg, with more diversified and less dependency-exposed client bases, simply don't face in the same way.

Two things can be true at once

The Mauritian offshore sector's objection to the "tax haven" label is not baseless. The double standard is real, measurable, and documented by the same indices that would justify the label in the first place. Mauritius is being judged more harshly than jurisdictions that objectively do more to enable global financial secrecy.

But that does not make Mauritius innocent of the underlying practice. Resisting the label also serves to protect the very structures — low-tax treaty routing, limited beneficial-ownership transparency, an economy heavily reliant on offshore capital — that invited the label in the first place. The uncomfortable truth is not that Mauritius is falsely accused. It's that everyone named in this article is running some version of the same playbook. The only difference is who is powerful enough to have the label not stick.

Sources: Tax Justice Network Financial Secrecy Index (2025–2026 rolling update); Ecofin Agency; U4 Anti-Corruption Resource Centre.

Un « rule of law » faible engendre sa propre transgression

L'État de droit de tradition romano-germanique désigne un principe constitutionnel : l'existence de lois, d'institutions et d'une séparation des pouvoirs qui encadrent l'exercice de l'autorité. Le rule of law, notion anglo-saxonne, va plus loin : il exige que la loi soit non seulement écrite, mais appliquée de façon prévisible, uniforme et universelle. On peut avoir un État de droit sur le papier tout en souffrant d'un rule of law défaillant dans la pratique — c'est précisément le cas lorsque l'application du droit devient sélective au point de perdre sa légitimité morale aux yeux de ceux qu'elle vise.

Quatre illustrations mauriciennes convergent vers ce constat.

• L'opération «coup de poing» contre les marchands «ambulants», menée sans préavis après des années de tolérance tacite, ne restaure pas l'autorité de la loi : elle la fragilise, car une descente soudaine ressemble à un règlement de comptes plutôt qu'à l'application d'une règle constante.

• La tolérance envers les contrefaçons (fake goods) obéit à la même logique, minant la propriété intellectuelle et décourageant l'innovation de demain.

• L'occupation sauvage des plages publiques — par l'installation permanente et commerciale de transats privés, de structures de restauration illégales ou la privatisation de fait d'espaces communs — montre comment l'inaction publique laisse le domaine public être confisqué par des intérêts privés au détriment des citoyens.

• Enfin, l'impunité chronique entourant les nuisances sonores — qu'il s'agisse de véhicules modifiés vrombissant sur les routes ou de haut-parleurs diffusant de la musique et des chants religieux à un volume excessif lors d'événements publics, commerciaux ou privés — illustre parfaitement cette faillite de la prévisibilité. Les lois contre la pollution sonore existent, mais leur non-application quasi systématique crée un sentiment d'anarchie quotidienne, altérant gravement la santé publique, le sommeil des citoyens et le vivre-ensemble.

• Dans ces quatre cas, la solution n'est pas l'indulgence perpétuelle ni la répression brutale, mais la cohérence : une règle annoncée, applicable partout, assortie d'un délai raisonnable et de voies de formalisation ou de mise en conformité claires. Comment procéder, concrètement ?

Les marchands «ambulants» : une méthode en trois étapes

• L'annonce publique : Une annonce nationale fixe une date unique d'entrée en vigueur — par exemple six mois — communiquée à travers le pays simultanément, sans exception géographique, pour éviter l'impression d'un ciblage arbitraire.

• Le guichet de formalisation : Pendant ce délai, l'État procède au recensement des vendeurs, à l'attribution de permis temporaires et à l'identification d'emplacements autorisés ou de zones de marché réaménagées.

• L'application stricte : À l'échéance, une application ferme et uniforme est déployée sur tout le territoire, sans nouvelle tolérance ponctuelle qui relancerait immédiatement le cycle de méfiance. La fermeté n'a de sens que si elle survient après que l'État a tenu sa part du contrat.

La contrefaçon : responsabilisation et traçabilité

La même logique s'applique mais avec des leviers différents. Une échéance est fixée pour la mise en conformité des importateurs et revendeurs, accompagnée d'une campagne de sensibilisation sur les risques juridiques et sanitaires. Durant cette période transitoire, les douanes et les autorités de la concurrence publient des lignes directrices claires : catégories de produits ciblées en priorité, seuils de tolérance zéro et sanctions graduées selon la récidive. Passé le délai, les contrôles s'intensifient de façon prévisible et documentée, avec des saisies rendues publiques pour asseoir la crédibilité de la mesure.

Les plages publiques : la restitution du domaine commun

Pour libérer le littoral des occupations abusives sans léser brutalement les petits opérateurs économiques, une transition ordonnée est requise :

• Le moratoire et la délimitation (Mois 1) : Les autorités (comme la Beach Authority) publient une cartographie stricte des plages, délimitant les zones 100 % publiques et les quelques espaces restreints dédiés aux activités commerciales autorisées. Un moratoire de six mois est accordé.

• La mise en conformité et l'octroi de concessions (Mois 2 à 5) : Les opérateurs illégaux (loueurs de transats, food-trucks, marchands) doivent postuler à des concessions officielles et limitées, soumises à un cahier des charges esthétique et environnemental. Les structures fixes permanentes et non autorisées reçoivent l'ordre d'être démontées par leurs propriétaires.

• La libération et le contrôle (Mois 6) : Passé le délai, les structures illégales restantes sont démantelées par l'État aux frais des contrevenants. Des patrouilles régulières et l'imposition d'amendes dissuasives garantissent que le domaine public reste accessible à tous les citoyens, de manière uniforme sur toutes les plages de l'île.

La pollution sonore : l'alignement sur les standards mondiaux

Pour la pollution sonore (véhicules et haut-parleurs), la même rigueur séquentielle s'impose, adossée aux meilleures pratiques de l'Organisation Mondiale de la Santé (OMS) et de l'Union Européenne. Les seuils de bruit en décibels (dBA) doivent être clairement définis par zone et par période :

• Zones Résidentielles : Limitation à 55 dBA le jour (06h00 – 22h00) et 45 dBA la nuit (22h00 – 06h00) pour préserver le sommeil (ce qui garantit un maximum de 30 dBA à l'intérieur des habitations).

• Zones Commerciales et Centres-villes : Limitation à 65 dBA le jour et 55 dBA la nuit pour concilier dynamisme économique et tranquillité des riverains.

• Zones Sensibles (Hôpitaux, Écoles, Lieux de Culte) : Un seuil strict de 45 dBA le jour et 35 à 40 dBA la nuit pour permettre l'apprentissage et le repos.

L'application de ces seuils suit une séquence rigoureuse :

• L'Annonce et l'Équipement (Mois 1) : L'État annonce un moratoire de six mois. Parallèlement, il dote la force policière et les centres de contrôle technique (fitness centres) de sonomètres homologués. Pour les véhicules, la norme de contrôle se base sur le bruit de crête (LAmax) : aucun véhicule ne doit dépasser 80 à 85 dBA lors d'une accélération à l'arrêt (mesuré à 50 cm du pot d'échappement selon les protocoles ISO). Pour les haut-parleurs, l'augmentation du bruit de fond du quartier (l'émergence) ne doit pas dépasser 3 à 5 dBA la nuit.

• La Transition et les Tests Gratuits (Mois 2 à 5) : Les propriétaires de véhicules et exploitants de haut-parleurs sont invités à se mettre aux normes. Les fitness centres offrent des tests de décibels gratuits pour permettre aux automobilistes et motards de vérifier la conformité de leurs pots d'échappement sans être pénalisés. Les municipalités instaurent des protocoles clairs pour l'obtention de permis temporaires lors d'événements spéciaux, avec horaires et volumes rigoureusement plafonnés.

• L'Application Uniforme (Mois 6) : À l'échéance, les contrôles routiers et les interventions de proximité deviennent systématiques et prévisibles. Les sanctions — amendes progressives, saisie immédiate du matériel sonore ou immobilisation du véhicule modifié — sont appliquées de manière automatique, sans distinction ni passe-droit.

Conclusion

Dans les quatre cas, le principe reste identique : un délai n'est légitime que s'il est suivi d'une application sans exception, et une application n'est légitime que si elle a été précédée d'un délai. C'est cette séquence — annonce, transition, rigueur constante — qui distingue une politique de rule of law d'une simple démonstration de force. Le rule of law ne se mesure pas à la dureté d'une opération ponctuelle, mais à la prévisibilité de son application. C'est cette constance, plus que la surprise, qui fonde l'autorité véritable de la loi — et la confiance nécessaire dans une société.