What if we could send messages into the future, so they couldn't be read until a particular date or event?
That's the promise of time capsule encryption, a set of technologies for securely storing secrets among distributed archives.
Archivists have long understood the need for "dark archives." Some records will not be preserved at all if they cannot be kept secret until they are safe to release.
But some records are too sensitive or valuable to entrust to any one archive. We need a new kind of distributed historical archive to protect the records now being wiped from history — the emails we would otherwise delete, the diaries we would otherwise burn, and the memories we would otherwise take to the grave.
Time capsule encryption splits records or encryption keys among a large group of trusted archives around the world. When the time comes, the record can be recombined and released to the public:
The record can only be released by a majority of archives working together. An attacker who steals, seizes, or destroys a minority of archives does not harm the system.
Short-term privacy is key to long-term cultural depth. Here are some of the people we're focused on:
TCEN, now under development, will allow users to encrypt files or emails up to one year in the future, using distributed keys stored securely at research centers around the world.
What if you want to send a file not just a year into the future, but decades? Or what if you want to split a secret — like your gmail password — and entrust it to your friends and family rather than to remote archives?
For that, you could turn your document into five encrypted “shards.” Each shard would be mathematically impossible to recover any information from on its own; but if any three of them were combined back together your document would pop back out.
We currently have Shard up and running as a prototype for Mac OS X. We’re working with the Harvard Archive and others to figure out how to keep those shards safe for the long haul.
Third-party tools and implementations:
How do you keep a gigabyte of data safe for 50 years, if it's not safe to give out lots of copies?
We're experimenting with an old-fashioned solution: data storage in books. Libraries have lots of experience keeping valuable books safe for hundreds of years. What would it cost to store data as a grid of black and white pixels in custom-printed books?
Our early experiments are encouraging — at high resolutions, data can be printed and bound into books for as little as $20 a gigabyte. We'll be writing up more on that soon.
Here are some of the tools we're looking at: