The Twitter-to-Mastodon migration is like going from beta testing the Fediverse to production. Just like a public beta always turns up issues that were missed during development, when going to production you suddenly have a *huge* pool of new users who are going to use the system in ways you didn’t anticipate and haven’t already accustomed themselves to its quirks.

And that turns up a lot more things you need to fix!

Some thoughts on features/user experience for Mastodon and other Fediverse software, based on usage and discussions I’ve seen lately:

1. Missing replies aren’t just an inconvenience, they’re a big problem. Instances really do need to reach out and check for additional replies when someone views a post. I’m not sure how to balance the extra network traffic. Maybe just have a manual “check for more replies” button.

2. Quoting is better than screenshotting. I can read quotes on any size screen. So can screen readers.

3. Lack of quoting hasn’t prevented flame wars or dogpiling, and it there’s no indication it reduced them either. If you don’t want to embed an entire post, at least generate a preview like you would to a website with suitable metadata. And let any third-party clients know they can fetch the message themselves and not hand it off to the web browser.

4. If you really want to keep some friction in the quoting process, don’t add a button, but add the preview/embed on display.

5. Link previews should be generated and displayed during composition, without interrupting typing. Whether the preview gets federated along with the post or re-generated at the destination is another debate.

6. User discovery on third party clients needs work, and autocompletion really needs to be part of the composition UI.

7. Remote interactions on posts that aren’t in the app *really* need work.

8. Basic interactions (profile, follow, like, boost, reply) should Just Work(tm) between different federated software, even if they don’t recognize all the same post types or display them nicely. You can always fall back to displaying a link to the source, like Mastodon does with Article types.

9. Mastodon ought to at least *try* to display Articles as long as the formatting isn’t too complex or the length too long.

10. Mastodon’s “Your admin can read your DMs” notice should make it clear that *most* messaging software has this issue, not just Mastodon.

11. Federated hashtag searching is also more important than the inconvenience I used to think it was.

12. I’ve seen several mentions of the need for local-only posts (which some platforms have) and mutual-followers-only posts, and I totally agree with both.

13. (Added 1/23/23) I want to be able to bookmark profiles, so I can mark people/groups that I want to occasionally interact with, but don’t want to follow all the time – but when I do want to look them up or mention them, I can be sure I got the name right.

While I’m griping about Instagram, why the heck are the detailed notification preferences split between the app and the system notification UI?

That’s terrible design.

Well, if it’s intended for usability, anyway.

If your goal is to make people see more notifications, though… 🙄

Yeah.

IMO there are two sensible ways to handle granular push notification preferences:

  1. Use the system’s per-app settings for all of it. (Tusky does this, even putting your per-account preferences in the system UI.)
  2. Use the app’s settings for all of it, and let the system just be an on/off toggle for what you’ve chosen in the app (like it was before Android even had UI for it).

Either way, everything’s in the same spot so you know you haven’t missed anything you want to turn off. Or anything you want to turn on, for that matter.

Sometimes it takes longer to automate something than it would to just repeat it yourself. Calvin designing a robot to clean his room, for instance. The method of estimating how long it takes to do the thing, how many times you have to do the thing, and then how long it would take to automate doing the thing, is a pretty good guideline.

But there are other factors: Like, can you include it in a checklist? If not, what are the chances that you’ll forget to do the thing? And what happens if you forget? What if you might hand things over to someone else and three people down the line, the fact that you need to do the thing doesn’t get passed along?

Or what if you have a situation like Desmond at the Dharma Initiative numbers station, and they know the step is “required,” but don’t know why? (Not that you’re likely to have quite so severe a failure mode!)

Anyway, today I automated some post-processing on a site that I hardly ever change. Not because it’s a pain to do the post-processing. Not because it takes a long time. But simply because if I don’t build it into the process, the next time I change something a year down the line I’ll probably have forgotten that I need to do the post-processing!

Twitter is suited for short statements and back-and-forth conversation.

It’s terrible for anything long-form.

Long Twitter threads* and images filled with text remind me of the old tech support days when users would paste screen shots of error messages into Microsoft Word documents and email me the document. It was a terrible tool for the job, but it was the one they knew.

Once you get past two or three tweets (doesn’t matter whether they’re 140 characters or 280, it’s the structure that matters), your ideas will hang together better and be better understood if you write an actual article somewhere. Sadly, Twitter has trained people to stay in Twitter instead of going outside to read the %#$ article**, because you won’t be able to get back to where you were in your timeline, and besides, that’s just too long to read right now.

And that would require you to have, like a blog or something, and what sort of weirdo has one of those? 🙄

So people use what they know, and we get screenshots of long paragraphs that are awful for accessibility. And we get 40-tweet threads that people only see fragments of and take bits out of context. And they’ll reply to tweet #5 complaining about something that’s addressed in tweet #12, but they didn’t see it, because that was hidden behind the “read more” link, and how long does this thread go, anyway? (Scroll bars solved this problem decades ago.) And we get links to articles that people don’t read, but they reply to them anyway — or rather they reply to what they assume was in them.

Which I suppose is what we had in the old days, I mean “nobody reads the articles” was a joke on Slashdot 20 years ago. But it’s still frustrating.

Update: I realized I don’t see this so much on Mastodon. I wonder if that’s one of the ways the culture is different, or if I just happen to not be following anyone who writes/boosts long threads on a regular basis, or if 500-character posts give people enough room to breathe that they don’t feel like they’re already writing a long chain, so why worry about keeping the number of posts down, what’s the difference between 10 tweets and 15?

Update March 2024: Apparently it was a posting culture thing, because it’s common on the Fediverse now too. At least on platforms like Mastodon that maintain a smallish size limit.

*To clarify, I’m talking about long threads that are effectively one piece of writing, not a series of “oh, and another thing” follow-ups, live-tweeting as things come up, actual conversations, etc.

**This part is true of Facebook as well.