Rescuing 7,234 Forgotten GIFs From a 1990s Web Archive — A Case Study in Digital Preservation

A tiny, forgotten corner of the early internet — a 1990s library of 7,234 GIF icons — almost vanished because no modern tool could read the way it was built. Saving it required reverse-engineering a navigation technique that most people have never heard of.

A library that refused to be archived

The archive in question is the Ibiblio Icon Browser, a collection of thousands of small icons put together by Gioacchino La Vecchia during the 1990s. For years it sat quietly on a server, a snapshot of the graphic language web designers of that era used every day. Web standards moved on. The server stayed the same. Standard web archives, including the Internet Archive, could only capture a fragment of it. The full collection was quietly slipping out of reach.

Why it was hard: server-side imagemaps

The obstacle was a navigation pattern called a server-side imagemap. On such a page there is no visible list of links. Instead a single large picture — a collage of every icon arranged in a grid — is the interface. When a visitor clicks anywhere on that picture, the browser sends the x and y pixel coordinates of the click back to the server, which decides what page or image to return.

That design was common before it was replaced by the faster client-side imagemap, where the link regions are downloaded with the page and the browser decides locally. But any system that expects real human clicks gets tripped up by an automated archiver, which cannot guess which coordinate belongs to which icon. The links are hidden behind a single opaque image.

How the rescue worked

The fix was to treat the collage like a map. By inspecting the layout of the grid, the rescuers worked out the pixel rectangle assigned to each icon, then wrote a script that requested every image one by one — not by clicking, but by asking directly for each file. With directory listing disabled on the original server, that was the only path to the full set.

What emerged was a browsable replica of the original archive, rebuilt with ordinary links so every one of the 7,234 GIFs is individually reachable and archiveable. The content did not change; the mechanism that was hiding it was replaced.

What this tells us about the early web

Most discussion of web preservation focuses on big sites and large archives. This story is a reminder that the web's long tail is made of thousands of small, oddly built corners — each one held together by a navigation trick, a custom server script, or an outdated standard. When the tools that understood them disappear, the content disappears too. Digital preservation, then, is as much an act of reverse-engineering as it is of copying. For a handful of low-resolution GIFs the effort may have looked small, but the principle it proves is large: the most fragile parts of the internet are often the ones nobody noticed until almost too late.