"Accelerated ageing tests on written voxels in borosilicate suggest data lifetimes exceeding 10,000 years."

[bruces@mastodon.social]

"Accelerated ageing tests on written voxels in borosilicate suggest data lifetimes exceeding 10,000 years."

*Well, how "Long Now" of them

https://www.nature.com/articles/s41586-025-10042-w

[zenkat@sfba.social]

@bruces Sure, but what's the lifetime of the decoding apparatus? A cube a glass is just a pretty bauble without a way to decode the information within.

[pieist@ohai.social]

@zenkat @bruces

Extraterrestrial graduate thesis: "A Theory of Microorganic Origin of Terran Embedded Vitreous Nanopatterns"

[rotopenguin@mastodon.social]

@bruces future generations will really appreciate this, as they snap the glass to make arrowheads.

[kolya@social.cologne]

@bruces all I wanna know is if I have to buy the White Album yet again

[mikestok@mstdn.ca]

@bruces @wordshaper

“Long-term preservation of digital information is vital for safeguarding the knowledge of humanity for future generations.”

If it’s digitized it’s crap. The amount of potentially useful physical evidence we discard while digitizing frightens me, and it’s not entirely clear future generations will have electricity, water etc. the way we’re going.

[celeduc@mastodon.social]

@bruces I'm genuinely excited for an archival data format that can keep data intact at least as long as a jar keeps jam.

[pieist@ohai.social]

@zenkat @bruces

And given that data compression ratios approach apparent statistical randomness in a pretty linear way, the chances of someone ever inferring that decoding method from the data itself seem pretty close to nil.

Unless we also leave a lot of Rosetta Stones around that explain it in mathematical and engineering baby talk.

[pieist@ohai.social]

@zenkat @bruces

Of course civilizational collapse isn't explicitly the problem being guarded against here, merely long-term readability.

The "hell, I don't even know if that Zip drive in my e-junk box even works anymore" issue is in a different problem space.

[david_chisnall@infosec.exchange]

@pieist @zenkat @bruces

The Silica encoding model reserves a small rectangle on the side as a calibration space. This lets the, change the laser properties and still decode them later, because a machine-learning system can learn the shape of the voxels.

That said, in the intended use case, all of the data would be encrypted (in part, because there’s no low-level delete and so you ‘delete’ be losing some keys from a tree of keys), so just being able to get the bits off the disk doesn’t help.