Pros and Cons of the Canvas Discussion Redesign

[benjamin_rodrig]

Hi Instructure folks. 

The Discussion redesign has some good points to it, such as mentions, quoting, and reporting. Those are contemporary features that people can enjoy in an online course discussion. 

However it looks like threading is limited to just three levels before the nesting stops and replies are stacked in chronological order. 

This is disorienting for users because people aren't used to their forums changing modalities based on thread depth. It's kind of like Reddit until it reaches three levels and then it becomes YouTube comments. 

Additionally the sidebar truncates after 4 replies, and you cannot expand threads within those to see what your students said in response to a post. That makes it easy to miss comments, or lead to frustration, or fatigue in trying to explore the interface. 

Training is required to enable people to use this. It's not "intuitive"; not easy to understand off the bat. Especially with the way the model changes literally mid conversation. People are used to social media for talking online, where the user experience is consistent no matter how long the conversation has been going on for. 

It would be awesome if the threading was maintained from the classic discussions, while incorporating some of those more modern features like mentions, quoting, and reporting that have definitely already added value. 

Question: Does Instructure plan to bring threading beyond three levels back? 

  

[mwolfenstein]

Can I just start by saying that I appreciate how @benjamin_rodrig is starting with recognizing that there are some good features in the redesign? I get that there are some non-trivial issues, but there are also some things that Instructure absolutely got right.

With that said, given the fact that (among other things) the non-threaded discussions feature hadn't been implemented prior to August, it does seem evident that the redesign was enforced prematurely and I think it would reassure users to get some recognition of that fact from Instructure.

However, I've been using the redesign for over a year at this point, and based on my own use I think that some of the backlash we're seeing is either tied to fairly unique use cases (e.g. extremely large classes with hundreds of students), or resistance to UX changes which is just part of the process of rolling out any new design. With that in mind, there are a few things that I wanted to roll up in one place as issues that are not just about user resistance to change or unusual use cases. In no particular order I'm seeing the following issues that are in need of improvement:

• The changing presentation on threading that @benjamin_rodrig highlighted in this post. The fact that it changes is 100% problematic, and users need to be able to see posts and replies in their full context which is getting lost three levels down. I know Instructure did analysis on this question and made a choice based on broad usage patterns, but the fact that it keeps coming up indicates that enough users do have discussions going deeper such that it wasn't the right design choice. Quantitative data can only guide you so far, and using only quantitative data to drive design decisions can take you in the wrong direction.
• The ability to have threaded posts expanded by default. Users are asking for the ability to enforce this from the teacher role in a course which is potentially problematic, but at least giving users the ability to set it as a user preference for viewing would be extremely helpful. Faculty could then at least guide students to select a setting in their course rather than telling them to always click a button every time they come to a post.
• Pagination is not a good solution and the UI element is small. If all replies can't be loaded on a page in order for the page to be able to load well, then a very visible load more replies button below the last loaded post would be a much better solution. If that's not a viable fix, then prominently displaying the number of replies on the page/number of total replies would at least signal to users that there is more to see.
• Viewing Unread: I don't know how this is the solution for viewing unread that was arrived at, but it's genuinely a weird experience. Search has the same issue. The fact that you can be in threaded and then when you select Unread it forces a switch to Split Screen is awkward and it's not what the user wants when they're trying to view unread messages if they're viewing in threaded. They still want to see that context. Also, why does the drop down menu have an unclickable "View" in it? This just seems like a feature that wasn't fully developed and is another good example of why enforcing the redesign was premature.
• It's awesome that we can now pull up the discussion inside of SpeedGrader, but as a number of people have commented it's not awesome that it's the only way to see the context of reply posts inside of SpeedGrader. In terms of the grading workflow you want to be able to first understand what the context of posts and replies was before making your choice of whether you are engaging with the discussion, and pulling up the full discussion is not a good substitute for being able to see just the parent comment and thread where a student was replying.
• Bring back sorting by likes. Why have likes at all if you can't sort by them? I know, there are some other use cases but from the folks I've spoken to this is the one they're most concerned about.
• Add a feature for pinning a reply or replies. This is a very old feature for online forums, and it's especially useful for the educational context where a teacher may want to pin a sample reply or pin an important reply that they want to push engagement with. I'm not sure why pinned replies weren't part of the original spec for the redesign, but they really need to be added.
• Images taking up space in user storage was a problem in the old discussions and as far as I know it's still a problem in the redesign. Please make images consistent with file attachments where they don't take up space in the user storage when they're added in a discussion.

There's one more thing that I suspect is more contentious, but personally if I'm quoting a reply, I basically want to select part of a reply to quote that has the appropriate tag on it. Quoting the full reply has a lot less utility because otherwise I'm already looking at this thing threaded. Honestly, I'd like to see the block quote style integrated into the RCE toolbar instead of buried in the menu under Format>Formats>Block, but obviously that's not a Discussions feature issue.

Finally, I really think that users should be able to sort by post date or recent activity. I know there was a ton of push back on sorting by recent activity and I was one of those voices because it's no replacement for sorting by post date, but it seemed like consensus was that both are good.

I might be missing some things, but these are some of the top issues that I and others seem to be encountering with the redesign. Unlike some other folks, I'm not here to say that I hate it or to ask for the old discussions back. There's a lot that's really good here, but there are also some necessary changes and additions that have to happen for the redesign to meet some of the most common use cases for discussions.

[lshulman]

I just noticed this week, for first time, that threads no longer indent beyond three levels. And I too don't like this. This was not happening last fall.

If there is an ongoing, deeply involved discussion with replies to replies to replies to replies (etc...), but without indents, the sense of who is replying to whom gets confusing. 

I do notice by the forth level reply, clicking "reply" does automatically add @Name of the person being replied to, but that can become misleading as to which post is being replied to. 

I have not noticed the date order of these deeper level replies. Chronological order would be LESS helpful than directly below the lower level post it is a reply to.

(Note: I am viewing the threads "in line")

PLEASE RESTORE FULL THREADING AT DEEPER LEVELS.