… Which means that businesses are making ‘too much’ money on top to sink into such endeavors, no?
For the vendor (non-)consent thing - Consent-O-Matic provides an appropriate framework.
(Whether such a side would even care about the preference/consent is another matter entirely - I’d suggest a throwaway browser identity and cookie auto delete for a start, anyway.)
Creating rules has a bit of a learning curve the first three or seven times, but I find that more interesting to do than go through a hostile/dark pattern cookie dialog or such the third time.
Hm, maybe the appropriate functionality from CoM could be re-wrapped as a TamperMonkey module…
Web automation for the masses 😱
As for the first points, yes, that may happen, but is it a problem for users who already are part of a ‘better’ experience here than on the for-profit platforms?
I, for one, find much better discourse here than anywhere on reddit, let alone Meta or Twitter.
Also exemplified by me engaging much more here than ever on the others. I do prefer quality over quantity - everyone is invited to join the table, but I don’t see much benefit in luring people there who would ultimately only dilute or be disruptive - ie, not really into the thing that’s happening here.
For the last point, well, legislators can certainly try. While telling people it’s all for their benefit and upholding freedom and democracy and equal opportunity and whatnot. And even keep a straight face.
But they happily give it to Threads, no…?
Yes, I know, I’m being somewhat more provocative here than necessary.
More down to reality, thousands of accounts being registered within seconds, possibly all from the same IP, aren’t ordinary user activity. And quite feasible to filter for.
Heck, you could even ask for the eMail and offer some “or, if you rather wouldn’t, you could…” thing that basically serves as a CAPTCHA.
(Disclaimer: I haven’t read into that referenced article by ninja at all, maybe it already says something related)
For one, it may be possible to filter accounts that were created but actually never used to log on, within a week or two of creation - those could go without much harm done IMO.
And/or, you could message such accounts and ask them for email verification, which would need to be completed before they can interact in any way (posting, commenting, voting). That latter one is quite probably currently not directly supported by the Lemmy software, but could be patched in when the need arises.
How would they ensure this latter thing?
In my current understanding, it’s readily possible today (on Lemmy and related software), what could Meta do to keep this from continuing to work?
I guess this will already have been said, but nonetheless:
I like the feeling of community as it is right now in the Fediverse very much.
Most of me hopes that it will not successfully federate with Meta, ever; or if it “must”, in a way that will be mostly irrelevant to me (communities I wouldn’t subscribe to in the first place, anyway).
I don’t see how that, in turn, would give Meta any control over the parts of the Fediverse that I care about. If they want to join and contribute in good faith, fine. If not, also fine. Why should it change anything for Fediverse “centered” communities?
I never cared about size or majority, but about quality of content and discourse. And I find that in those points, the current Fediverse much outshines anything else I’ve seen (Facebook, Twitter, Reddit, …) in the last decade or so.
Usually those all need to be in the same folder, and you launch unrar with the file with no (if such one exists) or the lowest number (0 of 1) only.