Granted, I really don’t know much about how all this works, but the thought occurred to me that Lemmy - as wonderfully open as it is, and without any kind of ‘disappearing messages’ or other privacy protecting functionality - is basically a smorgasbord for AI scrapers. Or am I (hopefully) wrong about this?
@Fletcher Not only it is a golden mine for scrappers (AI-purposed or whatnot), but even deleted things from fediverse (and, by extension, Lemmy) continue to appear out there (e.g. Google Search), be it through federated instances, be it through direct scrapping.
I feel like a personal example of that: I deleted my Lemmy account. Still, many of my content still linger on Google and other search engines through instances I never saw before.
However, it’s not because fediverse is open: it’s because of how Web (or, at least, Clearnet) works. If someone can access it, it can become available for others to access. When even DRM-protected, pay-walled content still ends up being openly accessible somewhere, it’s no surprise fediverse content can, too. Everything done on Clearnet will end up on many places simultaneously, lasting any deletion: Internet Archive is a common place to find digital ghosts.
While it seems ominous, it is thanks for this very nature that many important and/or useful content can still be accessed (e.g. certain scientific papers and studies that were politically removed by a government, certain old/ancient games that fell into corporate/market oblivion, certain books from long-gone publishers).
To quote Cory Doctorow: “Scraping against the wishes of the scraped is good, actually”. The problem isn’t scrapping, but the intentions behind who use the scraped content, particularly if such a “who” is a corporation (such as Google and Microsoft).
Problem is: to the eyes of a webmaster, well-intentioned scraping isn’t distinguishable from corporate scrapping. They’re all broad GETs (i.e. akin to the “all the things” meme), perhaps differing in scale, distribution and frequency, but broad GETs nonetheless. People have been setting up Anubis (the libre PoW CAPTCHA solution) or CloudFlare (the MitM corporation) to avoid AI-crawling, but they’re also becoming prone to oblivion when, say, their servers ends up disappearing forever one day, taking all their content to the realms of /dev/null: many of which are unique contents, useful contents, gone as no archiving tool (e.g. Internet Archive) could reach them.
IMO, you’re not wrong, but scraping isn’t wrong per se, either.