Skip to content

Instantly share code, notes, and snippets.

@BigZaphod
Created April 28, 2022 17:36
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save BigZaphod/6a2e3f1a11a1b337d8fbcec9ae0d9669 to your computer and use it in GitHub Desktop.
Save BigZaphod/6a2e3f1a11a1b337d8fbcec9ae0d9669 to your computer and use it in GitHub Desktop.
  • Identity being tied to the instance you join is very limiting. While a lot of instances have public feeds you could at least browse without logging in, in theory, many don't and require you to have an account. So if you wanted to get that "small town" feel in one small instance, you have to actually make an account and join it to do so. That's cumbersome and feels very unnecessary. I think identity could be sharable between instances in some fashion. Imagine you could join instance X and then click a join button on instance Y that takes your X identity and links it over to Y in much the same way that remote follows work. Perhaps as part of this process you could even optionally take on a new local name on the Y instance if you wanted and appear as if you are a truly local user, but under the hood there'd be a linkage between the two identities creating a kind of web of trust (if that's the right term). Thus you could someday later join instance Z using either your X or Y identity but doing so would link all 3 instance accounts together. You should essentially be able to login to any instance you're a part of using any identity from any of your other linked identities since they're all "you." In your account prefs on each instance would be something like the app list that shows the other instance identities it knows about. Probably there would need to be some kind of key or token exchanges under the hood to facilitate this in a secure way. I'm uncertain if that's a solved or solvable problem, though.

  • I've run across cases where I clicked a link to what I thought would be someone's profile or perhaps a link to a status and ended up on another instance. It's so surprising when this happens. I'm not sure if anything can really be done about this, though, as it is likely a bit of "this is just how it has to work" but it's really really odd. It is especially odd when you end up viewing a person's profile or status on some other instance that has a totally different and customized stylesheet/theme and the site looks completely alien and your eyes cannot find the information you went there to see.

  • I know that a core developer is of the opinion that things like quote-tweets are bad for discourse. I kind of see the argument, but there are a lot of times when it's the right way to give someone the proper credit for a thing. For example, say someone posts a reply that helps you solve a problem. Boosting that reply without context is confusing for your followers. A quoted solution is good in this sort of situation because you can re-set the scene, so to speak, and explain why it was good or useful or whatever. Yes it is also used for evil - however nothing stops screenshots or copy-pasting quoting like old-fashioned twitter, or even just sharing a link directly to the status. In fact, sharing a link to status has all of the same quote tweet downsides but none of the upsides since generally clients don't seem to be able to expand the status inline and you end up navigating away from the site or even the instance just to read it. It's an unnecessary hassle and to non web-savvy folks it's downright confusing - especially if they end up wanting to follow that person and now they're faced with the remote follow flow which is, frankly, a little bit of an obscure thing if you don't have an accurate mental model about how it works. I'm unconvinced that a quoting feature is to blame for negative behaviors. (I also note that nothing would stop a third party client from auto-expanding links to mastodon statuses and showing them inline anyway, would it?)

  • The CW feature is cool but there's kind of annoying usability issues with it, IMO. Whenever someone replies to a CW post their reply is also behind the CW and I have to open it to see it - even if I was in the reply. Likewise if I click on a toot to open the discussion, the entire thing is often hidden behind a CW even though I'm essentially opting in to see the whole thread. (Yes I know there's that little eye icon on the web interface that took me forever to notice that can use to toggle them, but it's another step.) I'm not sure if there's a solution to this but it's really cumbersome to the point where I'm tempted to disable CW entirely. They are also seemingly overused. Sometimes there's potentially clever reasons to use them - for example, hiding the Wordle emoji that some people seem to hate seeing, maybe? But there are also some people who seem to post EVERYTHING with CW all the time - I guess they're treating the CW like a subject line? That makes it super annoying to read their posts since I have to expand them every single time. Again it seemingly just drives me to want to disable CW entirely. I don't know how to solve this sort of thing because it's really more of a social thing and the software can't read your mind to know what really needs to be behind a CW for you vs. what the author thought should have been behind a CW. :/

  • I cannot figure out an easy way to mention people in the web interface that doesn't seem to involve copying and pasting full username@instance paths or whatever. Example case: I received multiple answers from different people about a suggestion to try. I wanted to compose a single toot thanking them by name, but doing so seemed cumbersome. Possibly there's an easier way to do this sort of thing. (This sort of stuff isn't exactly convenient on twitter, either.) If there were a concept of quoted tweets, Mastodon could potentially go a step farther here and implement a way to attach multiple statuses and not just one. In this way you could attach several posts from other people and thank a bunch of of them all at once by directly linking right to their post where they helped you. This makes me think that a possible compromise to a quoted tweet-like feature is to treat it a bit like a CW - implement the feature, but hide the quoted posts behind a disclosure thingy so you have to click to reveal them. (Although unfortunately that'd likely end up being annoying in the same way that CWs tend to get annoying and everyone would want a way to keep them expanded all the time.)

  • I maybe have just figured out a different way to think about a quote-tweet feature. What if we inverted it... so, in other words, instead of treating a quoted tweet like a post from you with an attached status, instead, we boost the original status like normal and attach your commentary as a "context" to the boost. So what I mean is, the only way to "quote toot" then is to actually boost the original toot you're going to quote as the primary action. Your comment on the toot would then be hidden behind a CW-like disclosure that appears after the boosted toot. The idea here is to reverse the incentives a bit. With the twitter way of doing this, it's the person who is doing the quoting that is the "primary" voice, right? So you read that commentary before you read the quoted tweet. They essentially get to control the narrative a bit because the person doing the quoting has "first strike" so to speak. If we reverse this, the original toot is boosted and visible to followers like normal and the only change is a small "here's why I boosted it" attached note you have to opt-in to read. The idea here is that, imagine you wanted to comment on a post from a politician you actually hated. With the twitter model, you get to have your words take priority and essentially override the quoted tweet since they appear first and the quoted tweet appears to come from you. In this reversed model, you're going to be forcing this politician's message (that you hate) upon your followers with your commentary being a small footnote. This should help you think twice about hate-boosting stuff you actually don't like because now your take on it is minimized by comparison. In addition to this, it's already normal for mastodon to prevent a boosted tweet from showing up a bunch of times on your home timeline because it kind of groups them. In this way, it could also group the "quotes" about a boosted post so they become almost like a little thread of related posts. If 20 people you follow all boost-quoted the same post, you as a follower would not see 20 new posts come across your timeline - instead you just see the boosted post once and everyone's commentary is nicely tucked away beneath that boosted post wherever it was when first seen. (There could be advanced options surrounding this behavior - like, always elevate boosts when a follow comments on them vs. not doing that vs. always expanding them or not, etc. but if the default was the first option, then I think it'd disincentivize doing hate-boosts because your "super clever comeback" is going to most likely be hidden way down someone's timeline and they'll never see it anyway.)

  • There are a lot of weird usability issues on the web client. The default behavior to auto scroll to the top all the time makes it hard to keep things where they were or to walk away from the computer and come back later to catch up and nothing is ever in the same place. The notifications tab seems to mix mentions and all the other notifications by default while ALSO having a dedicated mentions view. It makes more sense to disable mentions from the "everything" view, IMO, and leave them in the dedicated tab (I know I can manually do that - and I did). Turning on Slow mode doesn't seem to address some of the scrolling/updating issues because of course I usually want live streaming on the Local/Federated views and when you have the different filtering options applied to the notifications tab, sometimes it'll say there are new items but there really aren't any (because you've disabled them in the filter settings for the tab). Also in slow mode, sometimes clicking to expand the new posts scrolls me up a bit and sometimes it doesn't so it's like a game of "will I lose my place this time or not" every time I click it.

  • Posting GIFs and other silly things like that seems unnecessarily difficult on the web and iOS clients. Some people hate things like reaction images, but IMO they can add a lot of character to interactions. Yes they can get repetitive sometimes, but not everything needs to be so serious does it?

  • The remote follow flow is just weird. It's probably about as simple as it can be given the nature of the web, but it's still weird. (Mostly this is probably a web issue that presumably stems from not being able to share cookies across sites or something like that.)

  • I'm not personally too bothered by the lack of end-to-end encryption in DMs, but it would be a nice thing to see happen. I realize this is probably quite difficult in a federated environment, however. Possibly it would be worthwhile to surface the privacy implications right in the DM interface on the official clients/web in some way such as reiterating that it is not encrypted but it is, at least, non-public while noting that isn't the same thing as truly private? I dunno. I know this is a major concern among some of the people I follow on Twitter. I also realize the irony there, because Twitter itself can of course read all DMs so the privacy implications are pretty similar there but that isn't how it is perceived for better or worse.

  • The post visibility stuff is really neat and I think it has some interesting potential, but it's hard for me to really judge it since I'm on mastodon.social which isn't a themed instance, really. But what I'm thinking about is how on some very focused instances, you could think of the entire instance itself almost like a whole "topic" and to my Twitter-addled mind, it makes sense NOT to post everything to the public feed by default - rather posting just for my followers is the more obvious default. However if I were on a focused theme-oriented instance, it would also make sense to want to post "publicly" from time to time, too! This distinction seems hard to surface, if that makes sense? My Twitter-influenced gut suggests that "followers" should be the default for posts when you're on the Home timeline, and "public" should be the default visibility when you're posting from the Local or Federated timeline views. That sort of thing. However to really "sell" this, I think the whole UI would probably need to be rethought so that the concept of posting a message is more linked to the view you're currently "in." Or something like that. Sorry, this is a bit of a ramble - just that it feels there's something just a little off about how the visible stuff is handled along with how the Local/Federated timelines work. If I was on a sever dedicated to, say, cat pictures then my post about how annoying it is to have to go to the dentist today isn't really meant for the community and is more meant for people who are my "friends" (followers). Whereas if my cat does something particularly derpy and I want to share a pic, then obviously that is exactly why I'm part of the "cat lovers" instance and posts like that should be "public." I'm just not sure how you'd change the UI or UX to capitalize on this or if this is worth trying to address in the first place. (Is it possible to change post visibility after the fact? What if posting to followers was the default and there was a new "boost" button that, when pressed by you the author, "boosted" that post into the public feed potentially after it's already been posted?)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment