The paradox of modern luxury is that the more precisely it knows us, the less we seem to want it.
AI-driven personalisation flatters our taste so efficiently that desire itself begins to flatten. You open an app and there it is – the jacket you’d half-imagined, or the playlist that mirrors your mood before you’ve named it. The system anticipates, arranges, and completes. It feels frictionless, even generous.
But when everything fits this neatly, what’s left to reach for? Desire once depended on a perceptible gap, the space between wanting and getting. Now that gap has been optimised away. We no longer aspire; we’re simply anticipated.
Behind that easy charm sits a machinery, an industry, of prediction. Every scroll, hover, and hesitation becomes a confession. From these micro-gestures, the algorithm builds a probabilistic portrait: accurate enough to sell to, not to know.
This is predictive luxury – the luxury of convenience. It packages aspiration for the mass-affluent, translating status into data. The product is still expensive, but the experience is engineered for scale: “exclusive” taste delivered by statistical consensus. What once required discernment now arrives pre-approved.
To be clear, this isn’t curation. It’s correlation. Your discernment becomes the weighted average of everyone who clicked before you. Luxury houses once guarded their ateliers; now they guard their datasets. What was once stitched by hand is now inferred by pattern.
The shift sounds harmless until you notice what it removes.
Aspiration (the slow, self-defining kind) relies on uncertainty. We learned our taste through trial, boredom, and even embarrassment. Those edges are gone. There’s no risk in going to the restaurant where the algorithm has all but booked you the table. The algorithm keeps our preferences in a holding pattern, replaying what we’ve already confirmed, always within one standard deviation of safety.
The Predictive Plateau: a system that sells us the most probable choice, not the most interesting one. Left unchecked, it narrows the collective palate. As I argued in Luxury UX: Beyond Veneer, lasting equity comes from structure and restraint, not surface gloss. The real risk for luxury brands isn’t technological obsolescence but aesthetic homogeneity, a market trained to prefer the median.
Prediction is never neutral. Behind every act of personalisation sits a hierarchy of visibility, whom the machine believes is worth showing first. The more data you surrender, the clearer your silhouette in its model; those who resist become statistical ghosts.
There’s a quiet economics to this. By automating inequality, the algorithm devalues any form of wealth it cannot quantify or identify. The ultimate luxury, then, is to disappear from the data entirely, to operate through introductions, word of mouth, and private networks. The truly exclusive product is the one the algorithm cannot find, let alone recommend.
And yet there’s still one lever left: intentionality. The deliberate pause before purchase. The refusal to click “similar items”. The act of finding something the algorithm couldn’t possibly have foreseen. In a world of predictive luxury, this is not passive rebellion but an active aesthetic stance, a luxury of choice by will.
The smartest brands will design for this intentionality, not against it. They’ll reintroduce or retain friction as a feature: the waitlist, the mandatory consultation, the garment that demands to be felt. These are not inefficiencies but signals of depth, proof that the experience values attention over automation.
For all its precision, predictive luxury leaves a vacuum at the top. Once algorithms have colonised the middle (the mass-affluent market chasing “smart” recommendations), genuine exclusivity must move elsewhere. Increasingly, it drifts back to what machines can’t do: interpretation, eccentricity, the unrepeatable judgement of people who know.
That’s where true luxury now lives, in human-centred unpredictability. The ultra-wealthy and the culturally literate aren’t rejecting technology; they’re augmenting it. Data may light the runway, but the finale still belongs to the artisan, the editor, the quietly idiosyncratic expert who can surprise you in ways no model can.
British luxury has long understood this. Our best exports – Savile Row, Bentley Mulliner, McQueen, Hockney, Grayson Perry – thrive on that narrow line between discipline and disobedience. Their genius isn’t efficiency but editing: knowing when to break symmetry, when to leave the imperfect seam that proves a hand was there. The imperfect seam is a brand’s deliberate investment in unscalable production – the final, physical proof of value when all scalable processes have been commoditised. Curation, as I’ve argued before, isn’t collection. It’s the art of choosing what not to automate.
The challenge for brands now is to build value not through correlation but through judgement. To shift from efficiency to experience, from prediction to anti-prediction. Their next digital frontier isn’t better personalisation; it’s deliberate unpredictability, the algorithm that refuses to close the loop. Designing such friction isn’t romantic contrarianism; it’s the only sustainable strategy for generating new forms of scarcity, and with them, price elasticity.
Because in an economy obsessed with knowing what comes next, the rarest thing a brand can offer is the pleasure of not knowing, of being surprised, seen, and momentarily off-script. That’s the new exclusivity. That’s predictive luxury, undone.
Acknowledgements: This piece was partly inspired by Antonia Hock’s recent post on invisibility and the next era of ultra-luxury.
AI: This piece was refined with AI, for the image prompt, tags, excerpt, and a little sub-editing. The ideas, references, and rhythm are mine. You can still see my hand.
John Gibbard 
Too Many Podcasts?
Or, why skipping an episode feels like abandoning a friend.
There’s a particular guilt that comes from skipping ahead in a podcast series. Not the comedian-chats-to-comedian ones, or Desert Island Discs, those you can binge or bin at will. I’m talking about the recurring ones. The talky ones. The ones hosted by people you like, or worse, people you know. Miss a week and you don’t just lose the thread, you lose the right to laugh. The callbacks make no sense. The in-jokes have moved on. You’re no longer in on it.
I’m aware this sounds neurotic. But I’ve stopped listening to several podcasts not because they got worse, but because I missed two episodes and couldn’t face the trauma of catching up. I know I could jump in. I know no one cares. But somehow, I do. It’s the same part of me that keeps unread issues of The Spectator in a stack, muttering, “I’ll start again from the first issue.” That all came about when Jeremy Clarke got ill and I couldn’t bear reading his brilliant column out of sequence, inevitably posthumously.
The problem therefore, I think, is narrative continuity without narrative urgency. Podcasts, like newsletters or Jeremy’s Low Life column, have become serialised companionships. Their UX rewards loyalty, but punishes lapsed affection. It’s a structure built for the always-on, and it assumes you never really leave.
And the volume. The sheer, relentless sprawl of it. Everyone has a podcast now. Kind, intelligent friends. Former colleagues. Distant people I admire. I say this with genuine affection and no small dose of complicity, I write a blog read by literally tens, so I’m not throwing stones from the hilltop. But podcasting’s democratisation has created a landscape where the bar to entry is nil and the bar to quality is… unacknowledged.
This isn’t a snobby defence of old gatekeepers. The best podcasts out there are often the weirder, niche ones. The ones that would never make it past a commissioner’s desk. But that doesn’t mean the friction was all bad. A copywriter at my former agency once said, “Don’t waste the reader’s time.” With podcasts, the time-wasting is part of the premise.
There’s also the question of emotional design. If podcasts are a medium of intimacy, why are the interfaces and audio frequently so transactional? There’s no gentle onboarding for returners. No “here’s what you missed.” No warm “start here.” Just a reverse-chronological list and an assumption that you’ve kept up.
Imagine if books worked like that. Chapter 17 opens with “As we were saying…” and you’re left frantically flipping back (actually, come to think of it, that’s the exact reason why I really started to hate Thursday Murder Club). Or if Netflix removed season recaps because you should’ve been paying attention. It’s not hostile, exactly. Just… indifferent.
So what would better design look like? Perhaps:
Small things. But they matter. Because as much as podcasts masquerade as friends chatting in your ears, they’re still products. And products that ignore re-entry, or punish time away, eventually lose people, not to rage, but to fatigue.
I don’t think we need fewer podcasts. That would be like saying we need fewer books. But we do need better affordances for how people actually consume them: messily, sporadically, guiltily.
We don’t stop listening because we’re bored. We stop because the emotional lift of rejoining feels heavier than just starting something new.
And if you’re wondering whether I’ll catch up on that podcast you recommended last month, the answer is no. I fell behind. And now I can’t remember when his dog died.
AI: This piece was written by me, I did use ChatGPT to sub-edit, help shape the structure, and keep the tone aligned with my voice. The experiences, perspectives, and final edits are mine. AI also produced the tag list, excerpts and image that accompanies it.