We live inside an image culture that has never seriously asked what an image is.
I am not talking about depictions, not what it means, not what it sells — but what kind of thing it is. What mode of existence it occupies. Whether existence is even the right word.
For most of art history, this question was deferred by the physical object. A painting exists because canvas and pigment exist. You can touch it, conserve it, burn it. Its existence is continuous with matter.
But Walter BenjaminWalter Benjamin (1892–1940). German philosopher and cultural critic. Das Kunstwerk im Zeitalter seiner technischen Reproduzierbarkeit, 1935., writing in 1935, identified the moment this anchor began to loosen. Technical reproduction, he argued, detaches the reproduced object from the domain of tradition and substitutes a plurality of copies for a unique existence.
THE AURA WITHERS.
But Benjamin was writing about mechanical reproduction. Photography. Film.
He could not have anticipated what came next.
Because the digital image on the internet does not reproduce a physical original.
A digital image is not a thing you can hold. It is a set of instructions, executed by machines, rendered on screens, dependent at every moment on infrastructure it did not choose and cannot control.
A digital image does not have properties. It performs them, contingently, each time the conditions allow.
Consider what that means. Every time you open a digital image — you are not retrieving it.
You are summoning it.You are creating the conditions under which it can occur. And if any part of that chain fails — the server, the browser, the file format, the operating system — the image does not degrade or fade.
It simply does not appear.
And yet we treat digital images as if they were the most stable things in the world. We share them, screenshot them, assume they will be there tomorrow. We have built an entire visual culture on objects whose mode of existence we have never examined.
This makes every digital image an event rather than an object — not something that persists, but something that occurs.
In 1938, the philosopher Gaston BachelardGaston Bachelard (1884–1962). French philosopher of science. phénoménotechnique introduced in La Formation de l'esprit scientifique, 1938. introduced the concept of phénoménotechnique: the idea that scientific instruments do not reveal phenomena, but produce them. The image does not pre-exist the screen. It comes into being on it, through it, because of it.
Change the browser, change the resolution, change the year — and you change, in some non-trivial sense, the image itself.
And if the image is always an occurrence —
then it cannot be preserved.
It can only be re-staged.
“Memory is not an instrument for exploring the past, but its medium — just as the earth is the medium in which ancient cities lie buried.”— Walter Benjamin, Excavation and Memory, c.1932
When an archived image loads, something phenomenologically strange happens.
You experience it as present. The interface performs immediacy even when it is delivering the past.
Walter Benjamin wrote, in a short essay called Excavation and MemoryBenjamin, W. (c.1932/1994) “Ausgraben und Erinnern” in Denkbilder. Frankfurt am Main: Suhrkamp, p.576. Trans. R. Livingstone in Selected Writings vol.2, Harvard University Press., that memory is not an instrument for exploring the past but its medium — just as the earth is the medium in which ancient cities lie buried. Whoever seeks to approach their own buried past must behave like someone who digs. Must return again and again to the same material. Must not be afraid to scatter it, turn it over, pierce through the layers that had to be removed before the find could surface.
The screen is that earth now. The browser is the dig.
We do not look through the interface at something behind it.
We excavate into it, layer by layer, HTTP response by HTTP response, until something that was buried surfaces.
But for something to be buried, it first has to have been saved.
In the mid-1990s, the web had no institutional memory. Pages appeared and disappeared. Links pointed to addresses that no longer responded. Servers went dark overnight. The average lifespan of a webpage was estimated at forty-four days. An entire culture was evaporating in real time, and no institution was treating it as a loss worth preventing.
But in 1996, a software engineer named Brewster KahleBrewster Kahle (b.1960). American digital librarian. Founded the Internet Archive, San Francisco, 1996. archive.org founded the Internet Archive — a non-profit organisation dedicated to the permanent preservation of digital material. Its operating principle was simple and radical: if nobody saves it, it disappears. And disappearance, in a culture building its entire memory on digital infrastructure, is not a technical inconvenience.
It is a catastrophe.
The Wayback Machine currently holds over one hundred and seventy-five petabytes of data and has archived more than one trillion web pages. It saves approximately one hundred and fifty terabytes of new material every day.
In October 1996, the entire World Wide Web was estimated to be only two point five terabytes in size.
The archive has grown to contain multitudes of what it was originally designed to preserve. And still holds only a fraction of what was lost.
The Wayback Machine works by sending automated programs called crawlers across the web. A crawler moves from link to link, and wherever it lands, it records what the server sends back: the HTML, the images, the stylesheets, the scripts. Its raw technical response. That response is stored, compressed, and indexed — available to replay, on demand, years or decades later.
Each crawl produces what the Archive calls a snapshot — a dated performance of a page as it existed at one specific moment in server time, technically unique and unrepeatable.
If the crawler did not reach the image file: a broken link.
If it reached the image but missed the stylesheet: bare HTML.
If it captured everything in one pass: one relatively complete rendering of one moment that no longer exists.
The snapshot is not the page.
It is the page's ghost — a recording of its response, replayed through a different machine, in a different year, under different conditions.
This is where Wolfgang ErnstWolfgang Ernst (b.1959). German media archaeologist. Professor of Media Theories, Humboldt-Universität zu Berlin. Digital Memory and the Archive, ed. Jussi Parikka, University of Minnesota Press, 2013. becomes essential. Ernst is a German media archaeologist whose work examines how digital storage actually functions — technically, materially, institutionally. He argues that digital preservation is not passive. It is active, contingent, and perpetually at risk. The medium does not simply hold the past. It continuously re-produces it, under conditions the original could never have anticipated.
And here is the paradox he identifies: the more easily something can be copied, the more easily it vanishes entirely.
Ubiquity and fragility are not opposites.
They are the same condition.
Because a digital file has no singular, authoritative location — no place it lives — it is dependent at every moment on being actively maintained, migrated, re-hosted. The moment that maintenance stops, the file does not decay slowly like a photograph. It becomes inaccessible. Instantly. Completely.
And this is where preservation ends
and where circulation begins.
Artist and theorist Hito SteyerlHito Steyerl (b.1966). German artist and writer. “In Defense of the Poor Image”, e-flux journal no.10, November 2009. called the image that survives circulation — not preservation, but movement through the network — a poor image. A copy in motion. Degraded by its own travel.
The poor image is evidence of survival. Its pixelation is a record of how many times it has been copied, rescued from one dying server, relocated to another. The poor image has a life.
Most images do not survive long enough to have one.
But its life is not continuity. It is accumulation. Each reperformance adds a layer. Each re-encoding inscribes a new history onto the file. What looks like degradation is sedimentation. The image becoming, over time, a record of its own survival.