There’s a street in Bo-Kaap, Cape Town, where the light hits the painted walls at about 7am and turns the whole scene into something you’d struggle to believe is real. I’ve photographed it a dozen times. The shot I actually like — not the technically correct one, but the one that feels true — came from standing in the wrong spot at the wrong angle, reacting to a man walking out of a doorway I hadn’t anticipated.
I thought about that photograph recently while reading about how AI search systems decide which content to surface. The more I looked, the more the overlap seemed worth writing about — not as a metaphor, but as something practically useful for understanding what makes both disciplines work.
What street photography actually requires
Most people who pick up a camera in a city think street photography is about quick reflexes. Find interesting-looking people, photograph them. The shutter speed matters less than the timing, the timing matters less than the position, and the position is everything.
But position isn’t just physical. It’s about having a point of view that’s genuinely yours — something developed through time spent in places, noticing patterns, understanding light at different hours, knowing which kinds of moments repeat and which ones don’t. The photographers I find interesting aren’t the ones with the fastest cameras. They’re the ones who seem to understand something about the places they shoot that most people walk past without registering.
That understanding can’t be faked. You can study other photographers’ work and develop a technical competence that mimics it, but there’s a quality in the best street photography that comes specifically from having been present — from the accumulated experience of standing in rain at 6am waiting for a bus to clear a frame, from learning through failure which instincts to trust.
How this connects to AI search
AI search systems — the kind that power Perplexity, ChatGPT’s web browsing, Google’s AI Overviews — are trying to solve a problem that’s structurally similar to what a picture editor faces when choosing between photographs. Which of these sources actually knows what it’s talking about? Which one has the depth of understanding that separates genuine expertise from assembled information?
The signals they look for aren’t primarily about keywords or metadata. They’re about specificity, internal consistency, the presence of detail that couldn’t have been assembled without actual experience. A piece of writing that contains a genuine insight — something that contradicts conventional wisdom in a way that only makes sense if you’ve been paying close attention — reads differently to a language model than a piece assembled from averaging other sources.
This is where street photography and content strategy start to look like the same problem.
The presence problem
In photography, the presence problem is simple: either you were there or you weren’t, and the photograph usually tells you which. The ones made by photographers who understood the place — who had spent time, who had failed enough times to know what to look for — have a quality that’s difficult to articulate but easy to recognise.
In content, the presence problem is whether you actually know what you’re writing about. Not whether you’ve researched it, but whether you’ve done it, thought about it seriously, changed your mind about it, encountered the edge cases that don’t appear in the obvious sources.
The AI systems that are getting better at evaluating content are getting better specifically at detecting the presence problem. They’re not fooled by correct information assembled without understanding. The pattern of a piece written by someone who has genuinely reckoned with a subject looks different from a piece that covers the same ground without that reckoning — in its structure, in the way it handles qualification, in which things it treats as settled and which things it acknowledges as genuinely uncertain.
Specificity as a signal
The most reliable indicator of genuine understanding — in photography and in writing — is specificity. Not the kind of specificity that comes from looking things up (anyone can add a statistic or name a location) but the kind that comes from having navigated a subject in real conditions.
A street photograph of Hanoi taken by someone who spent three weeks walking the Old Quarter at different hours looks different from one taken by someone on a day trip. Not because one photographer is more talented, but because the accumulated experience produces different choices — about when to be somewhere, about which details matter, about which moments are genuinely unusual versus which ones just look that way to a tourist.
The same principle applies to content. A post about enterprise SEO written by someone who has managed campaigns at that scale — who has dealt with stakeholder politics, with content teams who don’t prioritise search, with crawl budgets on sites with hundreds of thousands of URLs — contains a kind of specificity that signals something to a reader (and increasingly, to AI systems doing evaluation). It’s not that the information is unavailable elsewhere. It’s that the framing, the emphasis, the acknowledgment of what actually causes problems — these things are shaped by experience in a way that’s hard to replicate through research alone.
What this means in practice
For photographers: the work that holds up over time is the work that comes from genuine engagement with a place or subject. The techniques matter, but they’re in service of something — a way of seeing that’s specifically yours. That’s what separates a body of work from a collection of competent images.
For content: the same principle increasingly applies. The content that AI systems are learning to identify as genuinely useful is the content that reflects actual understanding — the kind that has an opinion where the source material is ambiguous, that acknowledges complications, that is specific in ways that only make sense if the author has actually engaged with the subject.
This doesn’t mean every piece of writing needs to be a first-person account of direct experience. It means the content needs to reflect thinking, not just information. It needs to have gone through a mind that has genuinely reckoned with the subject, not just assembled a response to a search query.
The compression problem
There’s something that happens when you compress experience into content. The street photographer who has shot a city for years has an enormous amount of knowledge — about light, about patterns of movement, about the neighbourhoods that produce interesting moments and the ones that look promising but rarely deliver. When that knowledge is compressed into a photograph, most of it doesn’t appear directly. But it shapes every decision.
Content works the same way. A writer who genuinely understands their subject carries a large amount of knowledge that doesn’t appear directly in any given piece. But it shapes the structure, the emphasis, the choice of what to include and what to leave out. It’s present in the quality of the thinking, even when it’s not present in explicit statements of fact.
This is why ‘more information’ isn’t the right frame for thinking about content quality. The photographer who takes five hundred frames of a scene doesn’t necessarily produce better work than one who takes fifty. The compression — the selection, the judgement about what matters — is where the expertise lives.
Why this matters now
The reason this feels worth writing about specifically now is that AI search is changing what gets surfaced. For a long time, search optimisation was primarily about signals that were relatively easy to game — keywords, links, technical structure. Those signals still matter, but the systems that are being built on top of language models are developing a different kind of evaluation capability.
They’re getting better at the presence problem. At detecting whether a piece of content reflects actual understanding or assembled information. At identifying the kind of specificity that comes from genuine experience versus the kind that comes from research.
This is both a challenge and an opportunity. It’s a challenge because the shortcuts that worked before are becoming less reliable. It’s an opportunity because it rewards the kind of content that good writers and thinkers have always wanted to produce — content that reflects genuine engagement with a subject, that has something specific to say, that is useful because of what it understands rather than just what it contains.
The street photographer analogy holds here too. The shift toward better evaluation of genuine understanding is, in a sense, a return to craft. Not a rejection of strategy, but a recognition that strategy in service of nothing is self-defeating. The principle is the same as documentary photography: you can fake the surface, but you can’t fake having been present. If you want to take those same principles into AI search, I’ve written a practitioner’s guide to LLM SEO that covers what actually moves the needle.
No comment yet, add your voice below!