In December 1967, an unnamed analyst in the CIA's Support Services Staff sat down and performed a calculation that would quietly anticipate one of the most important economic debates of the information age: when does it make sense to convert physical records into a compact technological format, and when is it cheaper to just keep the paper?

The conclusion was striking in its specificity:

This policy holds that it is to the Agency's advantage, both economically and efficiently, to store the hard copy of records that are to be kept less than 30 years.

Thirty years. That was the breakeven point. Below that threshold, the CIA determined that the cost of microfilming — which ranged wildly depending on the material — simply wasn't worth it. The memo lays out the numbers with bureaucratic precision:

Costs for microfilming as stated in the attached price lists by GSA and our Printing Services Division range from about $8.00 to $80.00 to microfilm two thousand images. The low figure relates to an automatic rotary camera handling index cards of all the same quality or tone, size, color, and thickness of paper without any staples, fasteners, or folded or torn items. The higher cost is for planetary cameras and hand processing of several sizes, typing tones, and varieties of paper in a foot of ordinary files.

That tenfold cost difference between ideal and real-world conditions is a detail that anyone who has managed a digitization project will recognize immediately. The pristine, uniform input is always cheap to process. The messy reality of actual filing cabinets — mixed paper sizes, staples, folds, faded ink — is what destroys your budget. This was true in 1967 with microfilm cameras, and it remains true today with document scanners and OCR software.

What makes this memo genuinely prescient is the underlying framework: don't adopt a new storage technology just because it's more compact. Calculate the total cost of conversion against the cost of simply maintaining what you have. The CIA had already accumulated "some 93,000 reels" of microfilm occupying "about 1,200 cubic feet of Records Center space," plus "another 150 cubic feet of photo miniaturized products (i.e., aperture cards, minicards, microfiche, and so forth)." They understood the allure of compression. They just did the math first.

This exact debate resurfaced in the 1990s and 2000s when organizations faced the question of digitizing paper archives. Many institutions — libraries, hospitals, law firms, government agencies — discovered the same painful truth the CIA had already documented: scanning is expensive, messy documents are exponentially more expensive, and if you don't need the records for more than a few decades, it may be cheaper to rent warehouse space. The National Archives and Records Administration grappled with essentially the same cost calculus when planning its Electronic Records Archives program, which went billions over budget.

Today, with cloud storage costs plummeting and AI-powered OCR improving rapidly, the breakeven point has shifted dramatically. But the principle — that technological compactness has a conversion cost that must be weighed against the humble economics of shelving — remains as relevant as ever. Every company debating whether to migrate legacy systems to the cloud is performing a descendant of this 1967 CIA calculation.