Perhaps less-than-optimally worded. Replace with “Substantially long” - This very discussion thread would qualify.
In a threaded view you get:
Post
Reply
Reply
Sub-Reply
Sub-Reply
Riposte
Sub-Reply
Reply
Sub-Reply
If someone later adds a reply to an early comment in the thread, we get:
Post
Reply
New Sub-Reply
Reply
Sub-Reply
Sub-Reply
Riposte
Sub-Reply
Reply
Sub-Reply
How am I supposed to find that new post if I have to scroll up from the bottom of a 60 comment discussion?
I was responding to your comment that you didn’t understand how indentation would make things more difficult. Indentation is a necessary affordance to make a threaded ordering somewhat functional. The issue, at least for me, is the ordering in the first place.
As an example, every time I receive a message via e-mail about a new notification on Reddit, I merely click the button to take me to the comment, rather than trawl through thousands of comments manually.
Would you elaborate on this?
Do you mean that because the sorting is necessarily applied to each hierarchy individually, it produces problems like a file manager can cause:
You’ve a list of files, and want to sort them by modification date to find some old logs. However, because these logs are in a directory of multiple file types (some irrelevant) they’re grouped by type (which is alphabetical) meaning that although they’re generally chronological, the chance for something new to be at the bottom is non-0.
You’ve a list of music. You want to remove some songs you don’t like, but you can’t order them by rating, because too many have identical ratings. You’ve grouped them by rating consequently, and have ordered them by access date, so that you use when you last accessed them as a determiner, too. However, this means that some seriously old songs are missed even if you rated them highly when you were young.
A “Topic” is a conversation, a volley or round of ideas, each flowing from what came before. As we exchange thoughts, we are affected both consciously and unconsciously by all of the individual comments, even if we are not directly replying to them. Our understanding shifts, we have new information or viewpoints to integrate.
It is fundamentally false to think of a string of direct replies as independent of the rest of the discussion.
Caveat
If a set of replies is truly so divorced from the flow of the rest of the discussion that it benefits from being viewed in isolation, it’s an excellent candidate for being broken out into its own top-level topic.
This is why temporal ordering is important for both reading and composition. For a specific example, our own back-and-forth here would end up as a tree of replies on a platform such as reddit, would it not? However, they don’t really stand alone.
If you were to read only that reply string, without the context of intervening posts, It really does sound nonsensical and disjointed - as well as possibly coming across as more acrimonious than I intended.
If I had to compose using only the previous direct replies, we probably would have stopped engaging after one or two comments, or at worst, started yelling at each other.
Appropriately enough, it was this exact process - watching the discussion as it unfolded, including the parts that I didn’t directly respond to, and thinking about why I act as I do - that has allowed me to articulate and refine on what had previously been a strong visceral reaction. A threaded model would have deprived me of this experience.
I can confidently state that I am unaffected by text which I see whilst scrolling but do not read. To purport otherwise is nonsensical - if the brain does not deliberately comprehend and then commit observed text to memory, it is not stored and consequently unable to affect the reader. Billboards which you pass on the highway but do not read do not affect you.
For me, as long as a discussion about a topic is logically segmented, whether it is arbitrarily designated as a “topic” or “response” has no effect upon me. Think of a filesystem - it is a simple hierarchy of objects, and yet entirely comprehensible and traversable:
A hierarchical conversation should not differ (except that each comment is more frequently multi-line).
Regardless, that’s a rather unsubstantiated opinion, even considering what supposed evidence that you’ve provided:
The example (if I’m correctly identifying the example as the <details>-encompassed section) which you’ve provided doesn’t appear to demonstrate that. However, I understand what you mean, because it’s quite simple to consider a situation in which a threaded response contextually depends upon what it responds to.
However, why do you state this? I ask because all that it appears to demonstrate is that some threads should not be separated into new topics, irrespective of their relevance to the original topic.
I’ve no idea of what you mean by this. It’s too vague.
It was a joke. The blink tag was overused and distracting and eventually drove people away from the websites that were trying to get attention by it (see overuse of ads, see More cowbell skit by SNL)
Imgur has a threaded discussion model and they get around some of the issues by limiting the comment length. I think the limit is 128 characters or 256. New replies pop in dynamically, notifications to replies are sent to the inbox and link back to the discussion and nothing is lost. At 4 or 5 levels deep they load a new page (not my favorite solution).
They have other UX issues on their platform but the threaded discussion works more than it doesn’t. All comments are collapsed initially so when you see a new post you see a clear list of comments below the post and can read view and read easily and then can drill into them to see the discussion (no reloading, no lost place)
Update:
Not suggesting limiting length of content in discussions at all but describing how that option on imgur might have made threading more readable and suggesting having that as an option on a threading model.
@vel, considering the frequent long and intensely useful responses I see on a myriad of Discourse instances, I’m certain that such a solution wouldn’t be worth it merely to add threading. I’d rather have a flat discussion than have the content limited.
I have no idea how your quote of my post got chopped up like that. The contents of the “Details” (labeled “Caveat”) describes a side discussion that may/possibly should be broken out because it is truly tangential to the overall conversation. The quote version seems to reorder things and reverse my meaning.
Okay, I’ll try to be more detailed and address these two together. I’m asserting that an ongoing conversation cannot be logically segmented in a hierarchical manner.
To this point, the back-and-forth between the two of us is comprised of posts 56, 57, 59, 62, 63, 66, 67 & now 71. In a threaded model, these would be their own little tree.
Such a view would, however, be inaccurate. The presentation of my post 62 was influenced by Vel’s 61, and further reflection on it, as well as Piffy’s observation,
have lead me to my current position. There have been other influences, of course, but these are the most direct that I am aware of.
If you note that the arguments I’m advancing have evolved, even in this fairly short time period, you would be correct. I’ve adjusted and refined my ideas as I integrate the various parts of the conversation going forward. If I’d had to work through a threaded interface, the tree-like structure itself would have funneled me into an echo chamber where it was just the two of us, and the cross-pollenation of ideas would never have happened.
If this is how you approach online discussions - speed-running to the next direct response without at least skimming the intervening material to absorb the gist of it - I would respectfully suggest that you are:
Depriving yourself of a more nuanced, broad view of the conversation as it has taken place.
Depriving others in the conversation of a reasoned response that takes into account what has already been said, at least in cases where there are multiple people chiming in at the same time.
I might postulate that this behavior has at least partly been taught to you by threaded discussion models. I suspect that this is how you get Reddit threads that have:
The same comment repeated in multiple locations by different users.
An insightful observation that never gets responded to, and therefore has no effect on the discussion because it’s buried 2 or 3 deep in a thread that didn’t get much love due to its own first post not being flashy enough.
@Sailsman63, you’re making a very good point that I actually tend to make over and over again, when (yes, I’ll use this analogy a lot) discussing filesystems with people - direct hierarchy is rarely enough to demonstrate complex relationships.
I think you’ve provided ample evidence for Discourse to have both a more robust common relationship mechanism and separate extensible view models. Specifically:
When writing a new response, allow a person to choose multiple comments to mark as responded to. This means that a user doesn’t have to tag multiple people in their comment if they’re responding to multiple comments, and ensures that it is objectively communicated what they’re responding to.
This can then be consumed by the user’s choice of view model (flat, threaded, MermaidJS-like relationship diagram for that kind of person who spends his days looking at relational databases).
Flat would merely display multiple avatars in the response indication header.
Threaded would base the hierarchy upon either a comment which the responder designated as “primary” (not an ideal solution, but intuitive).
The Mermaid-like view (Star?) would provide an overview of what topics appear to be of most importance, then allowing the user to select a comment and switch to one of the aforementioned standard views.
Agree?
It depends upon the conversation. In topics like these, all of the context must be considered. However, in a thread about a technical issue which has divulged into discussions between multiple people like this has, requesting a summary or commenting upon a specific part distinct to the rest means that the rest need not be considered.
Everything is a cost-benefit analysis. I don’t have infinite time.
