The Importance of Digitized Archives

The corporate records of Remington Rand and Sperry UNIVAC fill hundreds of boxes at a library in Delaware. A researcher who wants to know how the company decided to buy a computer start-up in 1950 can travel there, request the boxes and read. Most people will not. Until those records are digitized, the history they contain belongs to the few who make the trip.

What digitization does

Scanning a document and making it searchable changes three things. It removes distance: anyone with a connection can read it. It enables search: a name or phrase can be found across millions of pages in seconds rather than across boxes in weeks. And it enables new kinds of questions, because a corpus large enough to search is large enough to analyze statistically.

The effect on scholarship has been measurable. Studies of citation patterns find that digitized collections are cited far more than comparable undigitized ones. History, like everything else, follows the path of least resistance, and digitization paves it.

What it does not do

A scanned image is not a searchable text. Handwriting, old typefaces and damaged pages defeat automatic transcription, and human transcription is slow and expensive. Metadata, the description that makes an item findable, requires expertise. Many digitized collections are therefore scanned but not searchable, or searchable but not well described.

Rights are a second barrier. Material from before the twentieth century is generally in the public domain. Material from the twentieth century often is not, and institutions must either clear rights item by item or restrict access. Newsreels, photographs and corporate records are particularly affected.

Preservation

Paper lasts centuries with little care. Digital files last as long as someone pays to migrate them to new formats and storage. Institutions that digitized collections in the 1990s are already finding early files hard to read. The question of who funds the indefinite maintenance of the digital record has no settled answer.

Why it matters now

Language models trained on digitized text are becoming a common way people encounter history. What is in the digital record shapes what those systems know; what remains in boxes does not exist for them. The choice of what to digitize next is becoming a choice about which past will be remembered.