CubeSats: a standard that outlived its satellites

In 1999 two professors, Jordi Puig-Suari at Cal Poly and Bob Twiggs at Stanford, proposed a satellite small enough for graduate students to build, launch and operate before they finished their degrees. The unit was a ten-centimetre cube. Cal Poly wrote it down as the CubeSat Design Specification, which today describes anything from one unit, up to 2 kg, to twelve.

The first ones reached orbit in 2003. By the end of 2023 more than 2,300 had been launched, and most of them are already gone.

The objects are designed to disappear

A CubeSat in low Earth orbit is slowly dragged down by the thin upper atmosphere and burns up on re-entry. That is deliberate: NASA’s rules for the small satellites it flies require them to come down within 25 years, and missions that cannot show it have been pulled from launches. A NASA study of predicted lifetimes found low-altitude missions lasting anywhere from five months to two years, with the solar cycle alone shifting the answer by a factor of five.

So the individual satellite is about as far from a durable object as engineering gets. Short-lived by design, cheap by design, replaced rather than repaired.

The specification is what survived

What has lasted is the document. A page of dimensions, mass limits, rail tolerances and the interface to a standard deployer has held for more than twenty-five years, through two decades of change in electronics, launch vehicles and the organisations using it. Universities, companies, space agencies and at least one middle school have built to the same numbers, and a satellite built to them in one country slots into a deployer built to them in another.

Nobody planned this as a standard. It became one because it was published, simple enough to implement from the text alone, and useful to many people with nothing else in common. Those are the same three properties that decide whether a file format will still open in a hundred years.

What a personal record can take from this

Separate the durable layer from the disposable one. The hardware churns; the interface does not. In an archive the same split applies: storage media, services and applications will all be replaced several times, and what has to stay fixed is the description of how the material is organised, so any future system can take it in.

Write the specification for strangers. The CubeSat document works because someone with no contact with Cal Poly can build to it. An archive’s index should pass the same test: readable by a person who has never heard of the company that made it.

Expect replacement, plan for hand-over. A constellation of small satellites is maintained by launching new ones before the old ones fall. A long-lived record is maintained the same way, by copying it forward to new media on a schedule, not by hoping one copy holds.

The honest limit of the analogy

CubeSats show that a standard can outlive every object made to it. They do not show that a standard lasts by itself: this one has an owner that keeps revising it and a community with a reason to keep using it. A record of one person has neither unless somebody arranges it, which is why the hand-over plan matters more than the medium.