The Lesson of the Ice Core: Reading Time in Layers

We often think of digital preservation as a problem of forward momentum—pushing data formats, storage media, and access protocols into an uncertain future. It’s a race against obsolescence. But I’ve found a more useful, and humbling, analogy in a field that looks backward with astonishing clarity: paleoclimatology. Specifically, in the practice of reading ice cores.

In Greenland and Antarctica, scientists drill deep into glacial ice, extracting cylinders of frozen time. Each layer is a year’s snowfall, compressed under the weight of centuries. Within these layers are not just ice, but tiny bubbles of ancient atmosphere, dust from distant volcanoes, traces of pollen and pollutants. The core is a non-linear ledger, a physical archive where data is stored not as discrete files, but as a continuous, stratified context. The depth is the timestamp; the surrounding layers are the metadata.

The Stratigraphy of the Server

Our digital archives are rarely built like this. We tend to snapshot a website, a dataset, or a hard drive, and treat that capture as a singular, isolated event—a digital specimen in a jar. We preserve the ‘what’ but often lose the ‘when’ and ‘next-to-what.’ The ice core suggests a different approach: what if we designed systems that intentionally preserved the layering of information? Not just version control, but the ambient digital atmosphere of a particular moment.

Imagine archiving a city’s open data portal not as a series of discrete, scrubbed CSV dumps, but as a living record that includes the broken links from that month, the deprecated API warnings, the forum posts where citizens reported anomalies, and even the prevailing web design aesthetic. These are the dust and air bubbles of our digital climate. They tell a story about the state of the technology, the priorities of the institution, and the community’s interaction with the data that the pristine dataset alone cannot.

This is a challenge to the instinct to clean and normalize. The ice core is meticulously clean, but it is not normalized. A layer with exceptional volcanic ash is not smoothed over to match the cleaner years below and above; its anomaly is the data. Our ‘messy datasets’ are often the volcanic layers, full of rich, contextual signals we risk scrubbing away in our rush for machine readability.

Applying the ice core logic means prioritizing continuity and context over convenience. It asks archivists to think like geologists, considering deposition rates, stratification, and the environmental conditions of preservation. It means designing protocols where every capture inherently references the one before and after, creating a true timeline, not a scatter plot of points. The goal is not just to save bits, but to preserve a readable cross-section of digital time, where anyone, in some future ‘now,’ can drill down and understand not just the content, but the epoch it came from. They could, like a climatologist, trace a sudden spike in misinformation or a revolution in data formatting like we trace lead or isotopes in the ice.

The lesson isn’t about cold storage. It’s about seeing our archives not as cabinets of curiosities, but as landscapes. Time isn’t a series of ticks; it’s a accumulating, pressing, ever-growing stratum. Our job is to ensure the layers remain distinct, contiguous, and richly, meaningfully dirty.

Notes & further reading

A few pages I came back to while writing this: