Vint Cerf Warns Against 'Digital Dark Age'

An anonymous reader writes: Vint Cerf, speaking at the American Association for the Advancement of Science, said we need better methods for preserving everything we do on computers. It's not just about finding better storage media — it's about recording all the aspects of modern software and operating systems so future generations can figure out how it all worked. Cerf says, "The solution is to take an X-ray snapshot of the content and the application and the operating system together, with a description of the machine that it runs on, and preserve that for long periods of time. And that digital snapshot will recreate the past in the future." Cerf is also pushing for better data preservation standards: "The key here is when you move those bits from one place to another, that you still know how to unpack them to correctly interpret the different parts. That is all achievable if we standardize the descriptions."

Comments

Our local time capsule...

[bazmail]

I remember our Mayor presided over the opening of a 25 year old time capsule put there by the local schools. Inside was a lazer disc. When he asked to view the contents of it, nobody could find a device to play it. Vint is right. And its not just a DRM thing, its a lack of standards thing too.

Re:Our local time capsule...

[solios]

Part of my old job (in a museum Exhibits department) was upgrading interactives and videos from the 80s and 90s to modern equipment - that included "transferring" laser discs the old fashioned way - plugging one of the still-working players from the floor directly into the capture hardware.

The thing is, I was transferring LD to DVD, which is actually a step *down* in quality. Kind of but not quite like how VHS is a step down from Beta (which I also dealt with).

The great thing about standards is there's so many of them!

Re:That is what VM's are for

[scsirob]

I actually bought a license for VMWare 1.0 when it came out, dates back to 1/7/1999. My V2.0 license is from 2000. I also still have the executables from back then. Hmmm. Wonder if I can get them to run on today's Linux kernels.

Which actually shows a prime issue with modern data and executables. There are a growing number of external dependencies. Stuff that is only accessible if DRM keys are available online, or when license activation servers are up and running. I have some stuff that only runs on Windows-XP. Even though I have a valid license, I may not be able to re-install it in 2025 or beyond, even in a backward compatible VM environment.

Re:only ancient encryption not breakable by fast c

[Dutch Gun]

Any and all other methods of encryption are subject to at least brute-force attack, which means they can be broken almost instantly, given sufficient computing power.

I think you may be underestimating how much computing power is required to brute-force attack modern encryption, especially when using a sufficiently long key. At the moment, we're talking about a modern PC operating until the heat death of the universe timeframe to break the encryption of a 2048-bit SSL certificate. Many of the early encryption schemes were broken because of flaws in the algorithms which allowed massive shortcuts to be taken or were weakened with very short keys (remember 48-bit keys?). Remember, with every bit added to the key, we double the inherent strength of the encryption, and cryptologists have gotten much, much better at creating incredibly secure algorithms as well.

It really isn't just a matter of waiting for hardware to catch up. Even with exponential speed increases in computing power (which isn't happening anymore, btw), in 30 years, we'll still be nowhere close to breaking today's state of the art encryption unless breakthroughs have been made that allow us to shorten the compute time via a weakness in the algorithms. It would take an unbelievable leap in computational efficiency (say, quantum computing) before we can even dream of brute forcing keywords of today's most secure algorithms, even within our lifetimes.