However, I think better commenting systems are the real problem here, instead of hacking current ones. The "threaded" conversations in this post are very cool, but disqus threaded comments are much more intuitive.
And instead of making just a blockquote, and referring back, github-type line commenting would be awesome.
The model of post then comments is kind of counter intuitive in terms of real discussion.
Inline comments organize comments around the subjects of a post.
That's neat for code. Not so for discussions I think.
Inline comments make it hard to read all the comments at once. And I often care more about that than being able to select which comment I'll read according to it's subject.
The problem with threaded threads is that there's less of a feel of permanence. When everything is linear, then you've got a very real spatial sense of where, say, the second comment is on a page. Depending on how important the blog community is to the blog, linear comments may hurt more than they help.
Actually, the better model comes from forums, where conversations were threaded early on and then switched to linear as forums evolved. Linear threads let you follow the progress of a conversation. Threaded ones lose that same sense of time.
Threaded ones let you follow a main conversation, ignoring digressions and side arguments; you can see the shape of the overall discussion, tracked by evolving subtopic. I have found that I approve of it; "derails" are rarer.
The idea of a linear conversation, though, is that once the topic is set that's what people talk about. If the conversation branches, somebody starts a new thread. That's one of the neat things about forum development: you get to watch what conversations begin, and sometimes your forum becomes something entirely different from what you thought it would be.
You lose some of that with threaded conversations. I've seen a few threaded forums, and on the whole you lose a lot of the people. In a linear thread you're forced to read everybody, which means the community evolves around its specific members. In a threaded one things are less focused. Conversations become more private, in a way, as small little conversations form from the big one, and while that works well on a site like this, there's something lost in the process.
Look at Reddit versus Hacker News as an example of what's lost. On Reddit, the most-known people are mainly the trolls. You get 911was_an_inside_job and Captain_Obvious. Other people don't stay memorable within the community, unless it's within a particular subreddit. When you look at conversations, very often a very bad thread (conspiracy theory or pun thread) will be voted up to the top, which makes community seem either immature or difficult or worthless. It's got some good communities, but they're the close-to-inactive ones, almost always. More people means the community breaks apart.
Hacker News does work, but that's because it's got a built-in audience of people who want to apply to YCombinator - from there it's attracted a like-minded audience. Here, there's no personality among people, or there's very little. I remember whether or not I respect a name, and I often disagree with the same people, but there's still very little personality here. Partly that's because here, as Reddit, discussion revolves mainly around news stories, but partly it's because when you're not linear there's no reason to get to know other people, because you can just skip over their strand of conversation.
So threaded conversations help with some things a lot, not so much with others. Cirqueti is being designed with a linear user forum in mind, because we want a very personal forum to go alongside the polished mainpage. I'm guessing we'll get users on the forum that aren't big on the main site and vice-versa, and I think that's extremely healthy for a site's community. (As the SomethingAwful forum credo goes, "There's a main site?")
-
Okay, too much reply there. I'll end by saying that I've never seen a blog that needs threaded conversation: not enough useful talk goes on in blogs for it to be worth it. A blog is a single-user-centric community: it's not healthy for it to have a cloud of responses like threaded offers. It fractures what minimal community ought to be there. The only blog I've seen with above-decent comments is Ebert's, and it keeps a linear thread, and commenters (and Ebert) get to know each other. That's what's missing out with methods like Disqus's.
I'd be concerned that by making it so easy for users to respond that actual discussion will be stifled because you won't be forced to actually scroll past all the comments to post something.
This is a really slick setup from the highlighting to the interacting with previous comments. Ideas like this really can enhance the effectiveness and/or frequency of commenting.
It reminds me a Jack Slocum's old blog where comments were tagged onto a line of the article. I think combining these two designs would be great.
Fascinating! This is a modified version of the idea that I'm using for my own start-up! Glad to see that the highlighting idea is spreading: I've seen similar ideas used elsewhere before.
The modification that our system uses rather than this one is that it creates the comment box on the left-hand side, rather than scrolling all the way down. The idea is that you then make multiple posts as you read down, rather than one single one that disrupts the reading process, and that they all bundle together again at the end. (Of course, blog commenting is a little different than content critiquing, so perhaps merely quoting once is sufficient here.)
Scalable Inman Flash Replacement (sIFR) is an open source JavaScript and Adobe Flash based technology that enables the replacement of text elements on HTML web pages with Flash equivalents. It was initially developed by Shaun Inman and improved by Mike Davidson and Mark Wubben.
So let me get this straight. In order to make sure everyone sees titles in the same font, they're rendered using flash. So I have to load the buggy flash plugin for a page with all text. I can't select the title. I can't resize the title (nor does it resize with the page zoom). Adblock displays a distracting "Block" above the title (and I'm assuming a large number of people have similar flash-blocking/warning plugins).
So why is this better than an image version of the title with alt text? That seems to be much easier, takes almost no resources, requires neither flash nor javascript, resizes automatically in Firefox (albeit using nearest neighbor, which is still better than not resizing at all).
Finally, is the font used for the title that important?
sIFR degrades gracefully and is accessible. I recommend that you read up on sIFR before recoiling in "no, it's FLASH!" horror. And yes, typography is important in design.
I never said it wasn't accessible or that it doesn't degrade gracefully. I did read the links provided on sIFR.
My points remain. An image would degrade just as gracefully and be just as accessible. And graceful degradation/accessibility seem to be answering the UI question for older browsers. What about current browsers? Since I have flash installed, the title is rendered using flash and now I can't resize/select the title. In this case, the "degraded" version seems like a step up.
Finally, yes, I know that typography is important in design. I was just wondering at what cost.
Yes, the image would work just as well. Now try generating that image on the fly for 1000 product pages, for instance. sIFR draws content from HTML, making it far more flexible. I very easily implemented Trajan Pro product titles for an artist's website for around 400 products, and this would have been prohibitively tedious if I were using images.
As for your resizing concern, sIFR should only be used for headlines. Typically, everyone can read text in that size (body copy may require resizing, but only a fool would run blocks of text through sIFR).
This reminds me of all of those services for annotating the web, but as far as I know none of those have gotten significant traction. Perhaps everyone just prefers referring to things by using just normal language, instead of a special tool to select something explicitly. If StumbleUpon included an annotation feature, maybe it could catch on.
However, I think better commenting systems are the real problem here, instead of hacking current ones. The "threaded" conversations in this post are very cool, but disqus threaded comments are much more intuitive.
And instead of making just a blockquote, and referring back, github-type line commenting would be awesome.
The model of post then comments is kind of counter intuitive in terms of real discussion.