A few things I'm noticing and feeling friction on WRT Mastodon:
In other similar systems, I've wanted to have some way of classifying other profiles, largely along ideosyncratic axis of my own choosing. A few of these relate to how much content I'm interested in seeing: all, some, little. More have to do with flags or biases I've noted. "Wibbles on about boring shite", such as I may do, f'rex.
Noting people with specifix axes routinely hauled out for grinding is another.
What Mastodon lacks, specifically, is the ability to /withdraw/ from a given thread. The whole concept of an /ephemeral/ nexus of participation of several people, without any formal set-up required, is wonderful. But /not being able to disengage/, shy of, say, muting or blocking everyone still participating, tends to result in a great deal of blunt-force trauma. I've already seen several instances of this, and felt it myself, in 3 weeks.
Thoughts on how to address this would be slick.
3/
@dredmorbius Thaaaat would be cool. Fortunately hasn't happened to me yet but my anxiety-annoyance gets peaked by stuff like that, so yeah, something to remove oneself from a conversation would be awesome.
The ability to categorise authors (sources) /and/ content (topic) for consumption would be useful.
I've yet to really see any site get this right.
Forum-oriented sites (Reddit, HN, Slashdot) tend to have topics which appear, and you and 15 million of your closest friends can all have a nice cozy chat. Um. No, not really.
On Google+, people who were really good at cultivating and managing a healthy discussion, by which I mean Yonatan Zunger, tended to have a series of Salons.
/4
If you're lucky enough to find someone with a) a wide set of b) compelling interests with c) a goodly set of friends and followers who d) mostly behave themselves or e) can be quietly dragged off without too much fuss with f) a well-written intro on occasion and g) a sense of what is and is not a good time, the experience is actually pretty damned good.
Yonatan wasn't the /only/ person to pull this off, entirely, but he was and is among the best I've ever seen. And in particular one trait.
/5
He has an exceptionally good interaction around reasonably-principled disagreement. He and I have some extremely strongly varying points of view on topics, and I've got sharp feedback on points he's often not at liberty to discuss. We've worked out a mutual understanding on both points, and he's someone who can and does hear out sharp disagreement without taking it personally (not a common trait amongst anyone, generally, or Googlers in particular). He shows this to others as well.
6/
What I've wanted to see in several systems, and have now /suggested/ across several systems, with no takers (here's my chance to maintain a perfect record!) is a content model which works as the /intersection/ of Sources, Audience, and Interest.
Basically: there are people I'm interested in seeing pretty much anything they write. Often, that's /not much/, so I /really/ want to know if they post anything at all.
There's another set who are generally interesting.
Others if recommended.
7/
Another, generous set, general discussion.
And there are those who, really, for the most part, aren't interesting. Most of the planet, actually, if only for logistical reasons. Attention rivality is the counterpoint of information virality.
(Cheap and effective high-exclusion filters exist for reasons -- even if that's a cool new club nobody else knows about yet.)
That's sources.
If I'm writing, there's who I think /might/ be interested, or whom I'd limit initial distribution to.
8/
There's another approach to this discussion which asks "just how much platform should any one person have, and how do we decide on that?" Mastodon / OSocial's instance level filtering is a part of that. It's a longer topic with a long background I'd like to get into, but not here, other than acknowledging it.
If we're going to have a means of defining /both/ outbound /and/ inbound content, there's the question of how content are classified. And of what classification scheme to use.
9/
So ... I wrote up a whole thesis at Ello on what channels to create, and how, and why ... And when Google+ created "Collections", I threw the whole damned thing out the window.
I strive to be consistently hypocritical.
I think the Ello classification is reasonably good, but I'd suggest another, and I know the initial response is going to be "get real". But there's a solid case.
https://ello.co/dredmorbius/post/FTpX7LWNFjtOpYGu5xRkHg
https://plus.google.com/104092656004159577193/palette
10/
Use the Library of Congress Classification Outline. Seriously.
* It's an extant system. Not a greenfield
* It's in heavy use. By many organisations
* No one organisation controls it
* IT IS UNENCUMBERED. (Dewey Decimal is not.)
* It is comprehensive, yet reduces to 21 top-level categories
* It's hierarchical
* It has a 120 year history of use, growth, refinement, and change.
* It has a change management process.
In short: it's dealt with error.
http://www.loc.gov/catdir/cpso/lcco/
11/
@dredmorbius this is a gem of an idea and I support this fullheartedly.
@dredmorbius LOC classification is far superior to DDS
@Elizafox I've actually heard arguments both ways. DDS has its defenders.
The killer, though, is that it is subject to copyright and cannot be freely used without licensing. That's an absolute non-starter.
@dredmorbius @Elizafox For me, it's simple: I've memorized the DDS, I can't remember much of the LOC for the life of me.
Dewey is copyrighted, but unless you are like... a library or large organization, nobody is going to notice or care.
@ocdtrekkie @Elizafox You are asking Mastodon and OSocial instance administrators to assume liability for just such legal attacks.
I'm absolutely not going to demand that.
If you want to front all their legal costs or assume responsibility yourself, and credibly demonstrate that you can do so, then we can talk.
See the MIAA / RIAA wars.
@dredmorbius @ocdtrekkie Using DDS numbers is fair use.
Using the entire system for classifying your stuff is illegal.
@ocdtrekkie @dredmorbius That is to say:
If you reference a DDS number, okay. You're fine.
If you classify your home collection with DDS, you might have a problem. No one is likely to notice though.
@dredmorbius @ocdtrekkie That said I refuse to support DDS through referencing the numbers so I use LOC classification.
Bearing in mind that every country has its own classification and that ISBN is the lingua franca.
@Elizafox @ocdtrekkie ISBN is simply a book-oriented identification scheme (International Standard Book Number, from the earlier SBN). It has a /publisher/ component, but no /content classification/ component independent of that. Otherwise it mostly concerns /formats/ of publications.
https://en.m.wikipedia.org/wiki/International_Standard_Book_Number
@dredmorbius @ocdtrekkie Sure.
But ISBN is searchable in any index worth its salt.
Each country has its own classification so using LOC or even DDS classifications on Mastodon means you are only accessible to US people.
@ocdtrekkie @dredmorbius I mean... the direction of text on the spines of the book isn't even standard. It goes up in French.
@Elizafox @ocdtrekkie Are you aware that in the Large Books collection ... books are often stored /flat/?
@Elizafox @ocdtrekkie "Searchable" != "Classification".
ISBN is an /identification/ schema, grouped by publisher, a serial component, and format modifiers. It does /not/ describe the /content/ of the material identified.
You'll also find works are indexed by title, author, pubication date, subject, LOCCN (catalogue, not classification, number), and a number of other bases.
Subject classification is only /one/ of those.
@dredmorbius @ocdtrekkie ISBN is searchable in any database worth its salt.
Getting other countries to adopt LOC when they all have their own systems... that ship set sail a century ago.
@ocdtrekkie @dredmorbius Example, at my college, I type in the ISBN: 9781473637313
It gives me Thing Explainer by Randall Munroe.
Any database worth its salt can do this.
Giving LOC numbers over the Internet to a non-US audience is like saying "if you're outside the US go fuck yourself."
@Elizafox @ocdtrekkie More specifically, a system built around DDS, with lookups, validations, etc., will probably break.
Worse: it's /trademark/, which doesn't allow for Fair Use at all (though it does noncommercial). There's the Library Hotel case (2003):
https://www.law360.com/articles/443/library-hotel-settles-dewey-decimal-trademark-suit
@dredmorbius @Elizafox Is there any way to look up a book by it's LoC number online? I've run into that once or twice and couldn't find a way...
I ... suggested to a good friend that this be considered in a project they're working on. The answer, which frankly struck me as tremendously /reducing/ my confidence in them and the project was, generally "I don't think it'll be sufficiently complete and besides I've created a few classifications myself", paraphrased.
The US Library of Congress has, and has classified, 164 million works. They're pretty good at this.
https://loc.gov/about/general-information/#year-at-a-glance
12/
As a compromise between my Ello and the LoCCS categories, mapping a set of reasonable general-interest concepts /onto/ the LoCCS would probably be a good start.
What that buys you is the ability then, for either side of a post, sender or receiver, to say:
1. This is what this is about (more on who /gets to/ classify what)
2. This is who I want to send it to (author/distributor)
3. This is who I'm interested in hearing about it from (reader/recipient)
And yes, multiple classes may apply.
13/
@dredmorbius would you recommend the library of congress system for personal filtering of the information one receives, or is it more for integrating into a service for the use of all?
@TheDistinction The critical element here is to have an /agreement/ of classifications.
Unless you've got a classification standard, you cannot have that.
@dredmorbius first step is to add the ability to unsubscribe from threads, preferably with time parameters.
If the best practice is tagging posts according to an agreed system, not everyone will do so.
Everyone seems to What something different out if mastodon; plugins and making the service more modular could satisfy the most.
@TheDistinction Just to emphasise total agreement: *Systems which rely on the voluntary, wilful, and conscious cooperation of the sender alone are doomed to fail.*
Even with the best of intentions, unwitting noncompliance error rates make such protocols unworkable.
It might even make sense to have a default classification be assigned to posts, say, "personal", which are set for distribution to selected friends.
If you want something to be /other/ than personal, find the default classification mapping, and assign it. If you are a budding Library Cataloguing Nerd, you can dig into the hierarchy and classify away.
Likewise, someone looking at your profile might find an interest in Personal, but not, oh, just to raise a topic of CW interest, Politics.
14/
In fact, in general, I suspect this might be the best way to address the whole "put Politics behind CW question", where someone feeling that way could elect /out/ of all political discussion /from their own profile/, other than specific exceptions.
If language and geography are considered as categories, you could subscribe or unsubscribe to content within a given language, or concerning a particular area's local events. Again, some fairly common friction points I've seen.
15/
@dredmorbius I seem to recall huge flamewars back in the day over whether any particular thread was 'on topic' for a given group, particularly after wandering through many replies. I imagine we'd get those again with such a classification scheme?
Would be nice though to be able to tag my posts in different categories, even if rough. I am very sorry for anyone who follows me expecting, eg, wuxia who gets Prolog and poetry instead, but that's my head, man, imagine *living in it*.
@natecull I'm getting there right now 😁
@natecull @dredmorbius This has occurred to me as well. On the birdsite, it's not unusual for discussions to change focus. The topic might change, but the participants are getting along anyway, if they share a certain set of common values (much like in "real life" conversations). This might not work with classification schemes.
@stefanieschulte @natecull The classification, if it noted the change, might be useful for drawing others into /that specific/ discussion.
Presuming such a thing would be useful.
Again: post-level granularity is available, /if utilised/.
Which also probably means that:
* New posts default to "personal"
* Resply posts default to "parent classification(s)"
@dredmorbius @natecull Something I often notice on Twitter is that people don't get together based on topics, but rather (as I mentioned before) based on certain values/a certain worldview. For example, I follow people who tweet about education or public libraries and sometimes engage with them, while they sometimes engage with my tweets about economics, even if it's normally not their favorite topic. 1/
@natecull @dredmorbius Participants might engage with each other because they all believe that humans should collaborate more and compete less, or that stock markets and private companies can't solve all problems. I wonder whether this would still work if everybody used a classification scheme. It might change the nature of such a platform considerably. 2/
@stefanieschulte @natecull Hrm. I have no idea how that would play out, but I'd be interested in your thoughts on why / what dynamics might be generated.
Why couldn't a discussion such as that be classified as, say, "politics, economics". Or even "political economics"?
Among, possibly, others.
@dredmorbius @natecull People might disagree about the classification. Some might follow the discussion about libraries only because they are "apolitical" bookworms (is anybody ever truly apolitical?), while others are more interested in the role of public libraries in society.
@stefanieschulte @dredmorbius That level of granularity is roughly what Usenet groups, as opposed to thread topics, accomplished, so yeah.
@stefanieschulte @natecull @dredmorbius
Not just values, but those with identical Myers Briggs types self report higher rates of compatibility. Not sure anything based on that could ever figure its ways into a real communication protocol.
@stefanieschulte @natecull Also, quite often it's the /group/ which is far more significant than the /topic/. Though as the group grows, say, from 40 to 4,000 to 400,000 to 40 million, that might change.
I've recently been contemplating how most of the members of some groups I'm in seem to have nothing in common, other than interest in the group topic, especially if the group is about a geographical area, which leads to misunderstandings and often, it seems to me, silly, disagreements.
While it seems like a good idea for community members to brainstorm virtually, it leads to disputes, as well as cooperation.
I believe it was a librarian or academician who mentioned that there's an MLA or ALA citation standard for Twitter, but not Mastodon. Hell, we could just code the motherlovin' MARC record straight into the damned protocol, if we decided on that.
16/
Classification generally raises a few additional problems[1]:
* Unclassified content.
* Misclassified content.
* Classification disagreements.
For the first, I'm not sure how categorising anything not /otherwise/ classified as "personal" hurts much.
For the second: intentional or disruptive classification could be considered on par with spam. Generally, if someone /does/ misclassify, don't follow, or block/mute.
Notes:
1. Eric Severeid: "The chief cause of problems is solutions."
17/
As for /misclassification/, there are a few dimensions of this.
1. In the case of misleading misclassification, see the previous case. Treat that source as untrusted.
2. There are philosophical disagreements. Though often these are
3. Political disagreements.
I learned a month or two back that the topynym "British Isles" is not universally accepted amongst all residents of, er, the Northwestern European Archipelligo. (On one of Yonatan's salons, as it happened.)
There's the Honey War.
18/
Seriously, a border dispute between Iowa and Missouri.
It turns out that the territorial evolution of the United States is vastly more complex then I'd ever considered. Fun trivia question: what /current state/ has lost or ceded more territory than any other? (And no, not as a result of climate change.)
And that's a /relatively/ calm and sober example.
https://en.m.wikipedia.org/wiki/Honey_War
https://en.m.wikipedia.org/wiki/Territorial_evolution_of_the_United_States
19/
I have an ongoing discussion with a friend about language, which seems to spend times in the various corners of "use precise terminology", "the corruption of proper English", and "avoide jargon". These being my friend's hobby horses.
They're all the same argument, argued from different sides. I haven't had the heart to pull out that response yet....
Ultimately, on disagreements: the classifications applied should be accurate enough to be useful, flexible enough to avoid gridlock.
20/
Another point is that a classification would apply at post-level granularity. If a discussion shifts (and /good/ discussions /do/ in fact drift naturally, though /not/ in a derailing fashion), then yes, take not of that. This, incidentally, is a point of flexibility Usenet lacked, as topic and granularity were /defined/ by the forum.
More generally, in /most/ instances, a /rough/ classification within high-level buckets /is good enough/. I don't expect much use of lower-level levels.
21/
Where highly-detailed classification /could/ become useful is if Mastodon / OSocial's localised and networked structures are used to create distriubtion hierarchies, with selected content bubbling up for broader discussion. In that case, some level of agreement on finer-detailed classification of specific widely-read threads (think of a Fediverse with 100s of millions or billions of users) could be quite useful.
(And political battles might also be much hotter.)
22/
@dredmorbius Right, that's where Usenet's group hierarchy meant admins could choose which subjects they wanted to host. That seems important for the Federation too.
'Content warnings' seem a quick but poor substitute for proper category tagging.
Also I imagine post *types* as well as categories may be required eg (like Facebook) meetings/appoinents.
Would need some easy way of picking a category in the post UI though.
@natecull /Types/ gets into a whole 'nother world of stuff:
* Events.
* Requests.
* Workflows.
...
Something to think about.
And yes, making the category picker /lightweight/ from a UI/UX is absolutely key.
@dredmorbius Yep. It seems like it would be a shame to waste all this infrastructure on just ephemeral chat when we could also use it for serious organising.
Some way of pinning posts into wiki- or page-like structures, so they don't get lost in the flow, I think is also important. Like, use them like index cards for solo or group planning. They're about the right size.
@natecull The ability to capture and coalesce discussions, particularly a few of my tootstorms, readily into more polished work would be excellent.
*And the underlying structure is there.* Mostly this only wants for a client-based support to be able to put the pieces together.
Hell, I could see Mutt as a Mastodon / GNU Social client. Maybe even auto-tootstorming by breaking a long submission into line-based toots, with sequence numbering and references built in.
@natecull @dredmorbius I was reading where someone had made a server-based tiddlywiki implementation that is more along the lines of a building block for what you are wanting. Also, some of the indieweb reader stuff applies.
@gcupc @dredmorbius I've used Tiddlywiki and it's really not!
More generally: I am not looking for any 'solution' to managing shared social data which requires taking some types of data (but not others) offline to a completely different silo and framework. Does that make sense?
I want data (like meeting dates and meeting notes) that one can create, distribute and access *as part of the same platform* as the conversation.
@dredmorbius @gcupc The reason for this is that a social platform manages a WHOLE lot of otherwise 'invisible' concerns around getting people together: identification, authentication, distribution, backup, support, a system administration culture, group norms, etc, etc.
The current situation where we have, eg: a website, a blog, a forum, a githbub, an IRC, etc, each with their own syntax, culture, logins... is the *problem to be solved*, not the solution.
@dredmorbius Also, filtering the personal/local/federated timeline by category would be a very useful use case now it's too fast to find interesting new people
@natecull Hell yeah.
So, those are my thoughts on user and post labeling and controls.
23/end/
@dredmorbius have they made a way to pick a language, yet?
Can't quite read Japanese (or French, or Portuguese!), yet.
@vmcampos No, though there's discussion in Github.
@dredmorbius it would be so helpful. But, I won't look a gift mastadon in the mouth!
G+ has about three features in particular which work quite well, shared somewhat with Ello:
1. If you participate in a discussion, you automatically subscribe to (most) updates to it. Ello has an explicit "subscribe" feature. This keeps discussions alove.
2. If you no longer care whether a discussion lives or dies, you can mute or unsubscribe it. (I'd love to see "mute the fuck up for a day" as another option...)
3. Notifications, which draw you back into topics, as Mastodon has.
2/