Cauliflowers, CWs, and Unlisted Toots
Looking at the draft response to Zignani et al, I'm surprised at the description of unlisted toots, my preferred mode. Apparently:
"This post privacy setting is specifically intended to allow circulation of a post without consenting to release of a post outside of Mastodon itself"
That is NOT my usage.
I make use of Unlisted specifically to ENABLE external access whilst minimising intrusiveness on Mastodon.
https://www.sunclipse.org/wp-content/downloads/2020/01/open-letter.html
#zignani #scraping
1/
I still remember when Usenet clients included a long admonishment about just what would happen when you hit "Send", and that you were going to be making demands on many people, in public. Netiquette evolved for reasons:
https://livinginternet.com/i/ia_nq_send.htm
It also scaled poorly. Usenet was breaking by 1992, though it survived, somewhat for most of another decade. *In* 1992, it was far smaller than Mastodon is today.
But maybe Mastodon could use a similar warning. As an "advise but don't blame".
11/
@dredmorbius Usenet had an X-No-Archive header that might or might not be respected by the various news servers. DejaGoogle did/does respect it, and so its archives are incomplete. Mastodon might profit from a similar mechanism.
@bthylafh So at least response is by primary or secondary interest, rather than just Random Internet Person.
At the same time, the toot _is_ at least in theory visible to others who want to go looking for it. Though not, again, by #hashtag, which again, I feel is a Really Poor Design Decision regarding Mastodon.
- Unlisted avoids global incidental viewing
- Hashtags assist specific interest in a given topic.
So I'll toggle visiblitiy _within_ a thread where I'm using hashtags in some...
2/
@bthylafh ... toots but not all of a thread.
(That's the case in this thread as well, though I've made a point of *not* making several toots with arbitrary #hashtags public as part of the point of noting that *those toots don't appear in #hashtag search. As with this particular toot. Even though it's otherwise viewable on the Public Web.)
3/
@dredmorbius @bthylafh I'm building something *sort of* like that. Dark forest inspired, but the default privacy level is acquaintance-of-acquaintance-of-acquaintance. Fairly broad, but always with social ties.
@dredmorbius I'm also kind of amazed at how "Mastodon's privacy levels are really confusing" is a Known Thing on Mastodon, and yet people are blaming this collection of PhDs and/or grad students for misunderstanding it, to the point of "well if you'd just read the CODE and the LONG GITHUB THREADS, you'd UNDERSTAND".
@dredmorbius Mastodon's privacy model is terrible, and you can neither blame the users nor the researchers for misunderstanding it.
@varx This reminds me of a long, long ago commercial X11 service call I was on (as the customer). Support engineer was exceptionally helpful and patient. I eventually asked what their background was. "Neurology".
It really *shouldn't* require a brain surgeon to use a system and ... still get it wrong.
(Though there are many PhDs and grad students with incentives to get things wrong, still, they're hi-side cognitive outliers.)
@varx @dredmorbius In Europe the GDPR takes precedence and they have no excuse for not understanding the most basic parts of the GDPR, masto's model has nothing to do with it. They had a section where they "considered" the ethics of their activities, demonstrating erroneous or malicious decision-making. They were reckless, unprofessional, and it could seriously cost their institution if someone sues. I hope someone does, this lesson needs learning.
@cathal I'm less familiar with GDPR and academic ethics than I should be, but then, that also seems to be a not-uncommon situation. The law and understanding are still being established, and new variants (e.g., California's recent privacy act) are emerging.
That said, there are numerous elements of the practices and methods of Zignani, most especially of publishing source toots w/ user-specific URLs, utterly voiding the "anonymisation" achieved.
@varx @dredmorbius Their misunderstanding is not about how Mastodon operates. Research ethics has guidelines on how anonymity should be preserved, and they failed in that.
(They also misunderstood Mastodon, but that only means their results are wrong.)
@jonne I see:
- Research methodology.
- Research ethics.
- Anonymity preservation.
- Mastodon's intent signalling.
- User expectations.
- Broader implications for Web scraping.
- Broader implications for surveillance aggregation.
- Broader implications of _unknown_ scraping, aggregating, linking, and inference activities.
All being in play here.
(Probably a few other elements as well.)
@varx Third-order contact is _still_ very broad.
Graph-adjacency or distance is attractive to programmers because it is accessible computationally. It maps poorly to _either_ trust _or_ utility functions, where trust is limited and defined scope, and utility is liberal and extended scope.
There's a mix of user-defined circles and collectively-defined communities which seems possibly more useful.
Or of some kind of time-limited posting, though that's not computationally amenable.
@dredmorbius I agree. While the prototype client needs to be relatively simple, the protocol is more flexible and would support clients with arbitrary visibility rules (with the caveat that they would have to select from people within 3 degrees of separation or so.) I just can't invest the time and energy to make a better default in the prototype client for now.
@varx Build your Thing.
But consider (and listen for / participate in discussions of) ideas of just what it is that entails and defines our social contexts, discussions, communications, and record-exchanges.
I've been kicking ideas around for a few years, find few doing similarly; Marshall Poe's "A History of Communications" approaches.
It's complicated: trust, relations, change, signals, records, meaning, context, legibility, durability, verification, identity/identifiers, credentials....
@varx Much of what's been done in computer contexts is more defined by what's programmatically and/or computationally tractable than what's socially / historically appropriate.
Maybe the medium will end up shaping the method. That often does happen. But I expect some give on both sides.
@dredmorbius Do you think there are existing systems that have gotten this right?
(For some value of "right", given that different people, different communities, and different spaces need different methods...)
@bthylafh An explicit intent-signalling method, rather than the implicit ones of toot distribution (at least as compares Public / Unlisted) might be an improvement.
I'll note another partial advantage of "unlisted" over "public" and one I have in mind: that a less-wide-but-public distribution tends to favour a higher quality of discussion. My toots are being seen either by those following me, or through the effective "I've vetted this" (not to be confused with "endorsed") of a boost.
1/