>>1395296That's what I'd like to know.
Despite the immense amount of effort here, and the storage, it's still not even close to most stuff they could capture.
Imagine how much bigger it could be.
With how expensive storage drives are getting these days, yeah it's getting kinda spicy there, those drive prices are surging.
As it is right now, we have small samples of preserved culture between average people in the ancient worlds, scattered across various cultures and eras.
As someone that generally likes reading random old shit like this, I would like to preserve current era stuff for the future as well. Even the shit stuff.
But sometimes it makes me wonder if it is worth the incredible effort in trying to preserve stuff at the highest possible quality when even now we have mere slices of shit from, say, Ancient Egypt, and in terrible quality at the best of times from a long dead language, but we still got some overall message from it, daily life, culture, stuff like that.
Worse yet, I bet if compression was used project-wide, it would be vastly smaller, but obviously far fucking worse to actually distribute and contribute specific chunks of data you care about, etc.
It's a very tricky situation. Very rock and hard place kinda scenario.
Keep the highest quality possible - respects the efforts of those people.
Lower the quality to just barely decent enough to figure out details - far lower sizes but comes at the expense of misrepresenting their efforts.
I've watched Shaun of the Dead at 1fps and sub "100p" on /f/ in under 10MB and it still got something across, but at the same time it lost a LOT of the action, fundamentally changing the media.
Transformer AI is highly efficient at compressing data, but they are shitty at re-creating them due to poorly tagged data and the fact most Transformer AI is probabilistic, not deterministic.
Trying to run a Transformer AI model fully deterministic is extremely expensive, so hallucinations out the wazoo!
