4 ms·
I think about this a lot recently. I've been using computers since the 80s in one form or another, but so much of "the early stuff" is just lost -- Amiga floppy
by vertis 3y ago
I think about this a lot recently. I've been using computers since the 80s in one form or another, but so much of "the early stuff" is just lost -- Amiga floppy disks that got taken and not returned. Content I did backup is in formats for apps that no longer exist.
Aside from personal examples there is also all of the "right to repair", DRM, type stuff that gets taken down from YouTube, GitHub. There are exotic examples like TornadoCash (for AML rather than the above categories), but there are far more mundane examples of content just disappearing as well.
I've started a personal setup of mirroring the top starred repos and downloading youtube channels for content that I consider important (there is, of course, too much for me to capture it all).
It fits into my thinking about the Long Now. Working in formats that can be opened in 100 or 1000 years. I don't have the hubris to think my own content is particularly remarkable and worthy of persisting for 1000 years, if I find other peoples content interesting though I can help persist that for the future.
For my personal knowledge base I use Obsidian and link everything (including downloaded YouTube videos). I also use the Firefox extension "SingleFile" which allows the easy capture of web pages that you're reading (and supports a direct upload to github, which then publishes to vercel).
It's far from perfect but it's better than some of my early knowledge base efforts where I've come back and clicked on links only to find that they've disappeared (and in some case not even on the Wayback Machine).
- midoridensha 3y ago>It fits into my thinking about the Long Now. Working in formats that can be opened in 100 or 1000 years. This is a real problem for old stuff from decades ago, but much less so now: as long as the data is in some format that uses open-source code to read (i.e., open-source A/V codecs), you just have to make sure the source code is available for it. A lot of data from ages ago is really hard to decipher because it was stored in proprietary formats, and the specifications and code for those are lost or difficult to find. Data stored as PDF or h.264, for instance, probably won't have this problem.