This is a guest post by Nicolas Holzapfel of stealth mode startup Yoomoot. Throughout the summer we’re running guest posts we like – exclusive to TechCrunch Europe – written by people on the tech scene in Europe. If you’d like to contribute get in touch.
Widget developer JS-Kit recently proclaimed the “death of comments”. How so? By way of its innovative comment management system Echo, that’s how. This would-be executioner pulls together disparate comments across the Web about a particular article and places them amid the conventional comments below the article. If it takes off, popular sites like TechCrunch could end up with hundreds if not thousands of additional comments. And therein lies the problem. How many of us can be bothered to read through more than the first few dozen or so comments on an article?
Not more than a handful I’m sure; a handful that won’t be made any bigger by Echo’s invading army of snippets from Twitter, FriendFeed, Facebook, Digg and Delicious. It may be nice for the article’s author to conveniently track every last mention of his or her article, but for the reader it just means hundreds more comments to not-read.
That’s not to say that Echo doesn’t benefit readers at all. Seeing new comments appear without having to refresh the page is handy, but not game-changing. Being able to post images and videos, log-in via OpenID and conveniently share our content across our different networks are all nice touches, though really these are just basics we’ve come to expect everywhere, not the stuff of epic comment mass-murder which Echo trumpets.
In reality an aspiration to be the death of comments is doomed to failure because it ignores the fact that comments, where they become at all numerous, have long been doing just fine at killing themselves by way of drowning in their own popularity.
Lots of comments amounts to an enormous long list of entirely unstructured text. There are no dividers or subheadings, no logical progression of arguments or groupings of opinion and no distinction between unique, intelligent insights and throwaway expressions of approval and opposition. Because nobody can be bothered to read through such a mess before they add their own comment, there isn’t even the structure of a coherent conversation. Instead, there is endless, pointless repetition; conversations emerge, peter out and then re-emerge 50 comments later with new participants who haven’t noticed that the same issues were discussed 50 comments ago. And these are the more productive comment threads. Much more often comments are unreplied-to and unacknowledged: futile, audience-less clamours and lonely questions without a hope of ever being answered.
By massively increasing the volume of comments and taking them from many different social networks, Echo will only exacerbate this problem: completing the transformation of comments into a disjointed stream of mutually-ignoring cries into the void, each destined for a brief flicker of prominence before vanishing without trace under the weight of a thousand tweets.
I don’t say any of this because I dislike comments but because I’m disappointed with how comments are handled. To my mind, the Internet should be the world’s parliament. It should be a massive conversation, a democratizing collective debate which abolishes the distinction between authors and readers – the active opinion-producer and the passive opinion-consumer. Unfortunately that’s not going to happen if all that the readers author is a garbled, unstructured mess that nobody reads.
Some people believe that comments on popular articles will always be like this because many-to-many conversations are impossible. They believe that if we want coherence we must content ourselves with either conversations in small groups (few-to-few) or one-way conversations whereby a throng of admirers hang on the words of an admired expert (one-to-many).
I disagree.
I believe that the Internet offers the potential for coherent many-to-many conversations for the first time in the history of humanity. As MG Siegler points out, today’s “commenting structure [has] been in place basically since blogging began”. What is needed is an evolutionary shift which is suitably adapted to the Internet’s unique potential and pitfalls. We need something that allows massive numbers of comments to be navigated quickly and easily so that coherent mass conversations can emerge.
We don’t need amplified echoes of what already exists.
Comments make the colorful world of social media. Without social media would die as participation would be non-existent.
Yea, author is trying to flamebait. Though real time search of comments by WP/blogger would have been nice.
But commenting in TC is thriving just because of its boss, all hail Arrington!
The author needs to widen his perspective to include blog posts themselves in his logic. There are so many posts out there, I doubt that anyone read past the first paragraph on this one.
The Twitter stream is like a humongous comment thread about a subject in general, specially if you use # topics. It’s also a very good way to centralize responses to your comments about a given subject.
If I click on the box above that asks to be notified of follow-up comments, I’d have to check my mail and that would take a while. However, with Twitter, it’s immediate.
I agree. Comments + status updates are the heart social media. The more the merrier.
Has anyone see what Blerp is doing with comments layered over websites? http://www.blerp.com
Jamie
word!
TC is the only site though where I read ALL comments…
I don’t even know what I am saying right now. I’m already dead.
comments are not broken, people are broken because of their claim to grasp every story in 140 chars, line by line. unstructured wilderness is the nature of the web – if you won’t take the time to dive into a story, discover and discuss it comprehensively in whatever form of presentation then move on and let your facebook stream shape your world view. a single company that moves or copies your verbiage from one silo to the other won’t and shouldn’t solve this issue.
Twitter has been detrimental to comments, I agree.
I certainly agree that people need to consciously take the time to comprehensively explore a subject to do it justice; but I think this process of exploration can be made easier and quicker through a more structured commenting system. ‘unstructured wilderness’ may be the nature of online discussion today but it’s certainly not the nature of the Web in general – Wikipedia is anything but unstructured – and, in my opinion, it won’t be the future nature of online discussion either.
I’m not sure there’s an easy solution to the issue of filtering comments.
The main existing solutions seem to be:
1. Low volumes of comments. Here you read them all, but not really scalable unless you maybe show a random sample each time.
2. Comment ratings. Needs to be crowdsourced to scale, but can be centrally done by moderators for smaller volumes. TC could do this for example by allowing the author to mark “star” comments. Problem with this method is that it favours early comments over newer ones.
3. Apply a social graph. Only show me comments by people in my network (fb/linked in etc). Maybe have a large “good comments” network to which I can add people whenver I see them make an intelligent comment. Requires open id or similar.
But just chucking 1000 comments at me flat isn’t really going to work except for searching/archiving…
1. Agreed
2. I think comment rating is great; I just don’t think it goes far enough. Also I think the favouring of early comments is just a symptom of the underlying problem – too many comments to read through and judge fairly.
3. Thing is, 99.99% of comments on articles I’m interested are from people I’ve never heard of. I think newstreams and comment threads are two very different things and solutions that work for the former won’t work for the latter.
I don’t know why everyone should see everyone else’s comments. Commenting on big sites (1,000+ comments, what?) should be done through channels. Some would be open and some would require invitations. Or maybe call them clubs or neighborhoods or whatever silly shit sticks.
Or alternatively, one option would be to limit the number of comments visible on an article (say 30). Combine this with a rating system which takes the lowest rated out of visiblity once the limit is exceeded to make room for new comments.
I look at comments as the glue between several many t many conversations.
Let’s face it: and important function of commenting is spreading your brand. You are putting in your two cents as a more or less discreet means to tell people where to go.
So on the one side, comments are just another marketing tool, on the other side, they are opportunities for encounters from different world that just touch in that one thread (that sounds more poetic than it was meant to)
For a non douche bag just trying to get back links – comments are just that, what someone thinks on a given subject.
Spam never sounded so poetic
. Seriously I think that kind of thing is perfectly legitimate but I don’t think interlinked blog posts amount to a conversation, and I think real conversations are more enlightening than self-enclosed opinion pieces.
I agree, here are my thoughts on the same subject, written as a how-to:
http://ma.tt/2009/08/kill-your-community/
http://www.sizlocore.blogspot.com/
This is the pure fault geeky thinking. Comments are not only “content”: It’s the personal tone, the flow of ideas, the ping-pong among the people, their opinions and thinking.
BTW, there are many people in the world, and you are not going to sum up all the info they produce, smart or foolish. Nothing happens if you miss a comment or two, and i prefer to read people than the “summery” or rating or any auto system.
True insightful ideas will always show up and gain momentum to be picked up later.
I agree that part of comments’ value is all the intangible ‘human’ qualities you mention and I’ve never come across an automated way of organizing that that I’m happy with.
I disagree that true insightful ideas always ’show up and gain momentum’ though. I think they very often get missed, which is why we came up with a different system.
A thoughtful article, Nicolas, and undoubtedly true. However, your last paragraph hints at your own acceptance of the futility of addressing the problem.
You say we need a way to manage many-to-many conversations, but not how this can be done. The reason it’s impossible is precisely what you hint at: nobody will read the comments of everybody else. There just isn’t time, and no technology can change that.
We can read a few (few to few) or one (one to many), but we can’t all read eevry comment. The ebst hope is a system that braeks comments into sub-conversations that we can break out of the larger stream and follow.
Threaded comments are a first step in this direction. Another would be the ability to filter comments (by author, keyword, etc), so we can extract only those sub-threads we’re interested in.
This way, the many to many mass of comments is broken up into smaller chunks of few to few conversations. Just like real life.
Actually I think many-to-many conversations are entirely possible: via yoomoot.com. The only reason I didn’t say that in the article was because I didn’t want to sound too spammy.
I think you’re absolutely right in emphasizing the importance of breaking larger conversations down. yoomoot encompasses that as part of its strategy, but not in the form of traditional threading or author/keyword filters.
I remember http://debatewise.com/ from seedcamp 07 – I always wanted to see that concept as a comment system.
Instead of everyone just adding a fresh comment (whether is has been said already or not) they could re-enforce a point of view. That would condense comments into a discussion rather than a set of sequential text strings…
I like Debatewise too, but I don’t think discussion-via-wikis are the way forward for comments. I think comment posters appreciate being able to speak in their own undiluted voices and receive clear recognition for what they’ve said. I think online discussions need more wiki elements, but not too much. It’s a balance we’ve tried to get right in yoomoot.
Hopefully this hasn’t been discussed already in the comments above…
I actually disagree comments are broken. Sure a popular article on TC.com (a clearly very popular site) may get 300 comments but actually such articles are the exception, and they tend to mean there is a contentious (interesting) issue at hand generating lots of opinion. If I’m interested in the subject I will often read all of the comments. Personally I’ve gotten quite good at filtering out (scanning) the comments – it doesn’t take long.
Improvements have been made, partial threading I’ve noticed on TC where you reply to a specific comment and it posts higher up, but I kind of like the unstructured chronological style we’ve been accustomed to.
One model is Slashdot which for a long time has a rating system on comments which works well. I’m sure there’s many others.
Interesting article.
I’m glad there are some people like you! However I suspect even dedicated comment-readers like you would prefer for comment threads to be as coherently structured as, say, a Wikipedia article. Also, it’s true that articles with huge numbers of comments are unusual but I think this is partly because most people feel comments are a waste of time. I think if commenting systems were better they’d be a lot more comments.
I like the comment-improving features you mention but I think more could be done.
Obviousslllyy not read all of the other comments, too many
But no, seriously, I do genuinely read most of the comments – I scan for the interesting ones then read. I find 20% of my reading is spent on the article, and the other 80% reading comments.
This is on digg, HN, TC etc etc.
You’re right. 127 Re-Tweets and 57 Likes on FriendFeed aren’t comments. An ego boost for the author maybe but they offer no insight or help to spark discussion
Which is what I said here two weeks ago.
I much prefer the way Disqus separates ‘Comments’ from ‘Reactions’.
I am surprised no startup evolved yet, providing some ‘essence-engine’ which accumulates the essence of a flow of comments on a specific article. (inveztors give me buzz)
if you checkout the huffington post for example; I stopped reading comments, its so demotivating to even start reading comments on +2500 comments based article.
I was doing some research last year into tag clouds and saw that Priceline UK was filtering keywords from the consumer reviews to create a sort of “tag” cloud (a feature they called PillowTalk). It seemed like a useful application for tag clouds and could be applicable to comments and reviews in order to generate “sentiment”. However, it’s now gone from the site — wonder if it simply wasn’t utilized by users? Some way of indicating sentiment is absolutely the next step after aggregating — because the volume is simply overwhelming.
Yes we were surprised too; that’s why we started yoomoot!
The development of features such as Echo and other similar services that aggregate your conversations in one place reflect the growth of the content aggregation trend. After conversations and content spread and pollinate, the need to bring them back together and see the whole story becomes essential.
Real-time commenting feature on your blog or website means being able to aggregate conversations in different social networks about your posts as they occur, making it a more dynamic narrative. Aggregating conversations in your site adds great value in my opinion. Sure, there can and should be better ways of grouping/filtering comments, and hopefully we’ll be seeing those developments. But whereas conversations are spreading outward from your site — thereby decreasing to a certain extent the relevance of your site — pulling all these “ripples of conversations” to flow back puts your blog / company / product / brand back in the centre of conversations without disrupting the whole social media narrative.
The comment is dead. Long live the comment!
Comments are a real problem. M. Thomas and M. Torrango make very good points. We put up with a lot in the conversation of the internet, principally because we can skippily read through more non-sense words than we’d ever have the patience for in a spoken forum. But it is still possible to saturate even this sense, with the risk that we miss good stuff. Has anyone ever scrolled back over YouTube comments?
It’s an impasse between the dominant liberal/libertarian values of the internet, which values free speech above all else, and the informational realities of our current attention economy.
Ways of not-reading are essential. Hopefully this will differ from the not-listening implicit in real life social class systems, in that each individual can choose his/her filters, rather than being monopolistically herded.
We need systems that encourage users’ internal controls: does this add? If not, why not remain silent? We might call it “conscience.”
And we need social structures that encourage users to develop their consciences. Fear of exile (a “never read comments from this user again” button? sock puppet filtering?) to encourage the pause for reflection, the development of constructive online personae (a “this person is great, give me all their stuff in future” button).
In other words, apply the whip of social inhibition, and the carrot of increased attention for comments of value.
As for “Echo”, I guess might be useful as a bucket of data from which to mine, but they’re going to have to work out how to build a social system on top of it, because current artificial intelligence/semantic web (this is a comment! yeah I knew that) just can’t and won’t add value; the opposite in fact.
Nicolas, please consider some of the additional points:
1. One should ad to the conversation Disqus commenting system arguably more influential platform compared to J-KIT that just introduced “echo: functionality this week, many say in a competitive response to J-KIT. The race between commenting platforms is about the geekiest of functions not about the “parliamentary culture”. Human conversation craves simplicity of a dialog, not the hotness of the next gadget.
2. And speaking of “a throwaway expressions of approval and opposition” one should mention Twitter as one of the culprits and it’s detrimental influence on traditional comments.
3. The idiotic likes and dislikes. They all point to the style that departs from a coherent conversation towards the supremacy of emotional reactions. The importance of how I feel as opposed of what I think. For an author the thinking part is so much more valuable yet the morons still persist with “LOL, Amazing!!!!!”
4. Anonymity is at the root of the eroding conversation. You can’t have a dialog with handle, especially a deceptive handle. This is huge problem; people need to be taught away from the retched anonymous culture.
5. As long as the meta-blogs don’t lead in this, little will change and the comments will continue to erode. Alas the big blogs don’t care about conversation. They hideous “likes”, the “echo” and the rest of the visual garbage are the vanity mentions that translate into clicks.
Lots of interesting points there
1. I agree. Echo makes sense from that perspective.
2. I actually think it’s equally possible to argue that Twitter has improved comments by teaching people to write concisely. I suppose it’s possible that Twitter has encouraged inappropriately tweet-like comments but I can’t say I’ve noticed a change from pre-Twitter days myself.
3. I agree, more or less. I think ‘likes’ make sense on Facebook, which was never about in-depth discussions, but they have no place in real debates since liking something makes no difference to whether something is true or not.
4. Anonymity kills social inhibitors, but that encourages openness as well as bad behaviour, so I’m not sure complete abolition of anonymity would be entirely good. It’s worth noting that even if a user’s handle isn’t their real name, they can still develop a good reputation within an online community and hence have a reason not to risk jeopardizing that reputation with bad behaviour.
5. I’ve no idea what goes on behind the scenes but personally I believe the meta-blogs simply aren’t sure of how to resolve the problems with comments.
2. Its like saying that we publish so many books today, we are getting better at publishing. Everyone understands that book spam is a huge problem for reading.
4. big subject, not suitable for a comments but I am with Paul Carr on this one.
I claim prior art in comment ranking algorithm.
I’ve always been a fan of Stack Overflow’s* commenting system. Lots of productive conversations happen there, though I don’t know how well it could work outside the niche it has carved for itself.
*http://stackoverflow.com/
I think Stack Overflow is a fantastic improvement on older Q&A forums and I think their comment system works great for them. However, I think you’re right to suspect it wouldn’t work well outside of its niche.
From Stack Overflow’s FAQ:
“Avoid asking questions that are subjective, argumentative, or require extended discussion. This is not a discussion board, this is a place for questions that can be answered!”
This is just as well because “extended discussions” aren’t even possible on Stack Overflow. The voting system means that comments are not in chronological order and reply-threading is not possible. If you can’t tell who’s replying to what you can’t have a coherent conversation.
While root comments are ordered by votes, replies to those appear to be ordered chronologically. Example:
http://stackoverflow.com/questions/1246058/senior-programming-guru-who-cant-program-should-i-find-a-different-career
Am I seeing things? It might just be a fluke, but it does look like the votes aren’t deciding the order.
You’re right. I made a lazy assumption and got it wrong. So extended discussions are possible via Stack Overflow comments after all, though of course they’d suffer the same problems encountered on other sites.
Didn’t Backtype start doing this ages ago? Or at least, on my blog comments are pulled in from Twitter, Digg, etc and placed into the Post’s comment stream…
I gave up reading comments on this article after Matt’s.
Comments on article are author’s responsibility. The author is the one who decides whether he want people to stay focused or take his article out of context. If you are not monitoring for repetition, unnecessary sarcasm, bullying, out-of-place sentiments, than you are yourself ruining what you wrote.
My take.
And the trouble with tech industry is that they always try to find technical solution to human problems. Not everything can be solved by technical solutions and I believe commenting is one such area (though there can certainly be technical improvements).
Check this out!
http://www.fullsoftwaresfree.blogspot.com/
Maybe it’s not that comments are dead, or a bad way to manage the conversation. Maybe it’s that the tech community gets bored with things way too fast.
I mean, seriously, just because some new technology hasn’t come out to allow people to add their opinion to someone else’s online doesn’t mean comments, as a feature of blogs, is utterly broken.
For smaller blogs comments shown by echo by pulling from social sites doesn’t really a problem as mentioned for a popular blog like tech crunch imagine post filled by all comments mentioned in social sites? can echo be useful needs to rethink.
Text is not unstructured:
Information = Data in context
Context = Organized data
Learning = Self Organization of data
Learning creates context. Learning in a given social environment creates shared contexts, language for example.
If you have a learning system, it’s relative easy to see if a comment makes sense in a given context for that system, if it knows that context. But to be really helpful it has to know the users contexts and therefor preferences.
Only if you try to parse text in nouns, verbs ….. becomes text unstructured, since we use a lot of double meanings which become only clear in context.
That’s also why many-to-many conversations are so hard, since we encounter a lot of ever so slightly (friends) or really different contexts.
So we tend to group comments the way what we know, tendency reinforcement (vast majority of people). People actually feel good if they get/read a lot of comments which reinforce their view on any given context. Only if you want to learn more do they get in the way, but leaning requires effort (don’t expect everybody to like that, all the time).
In other words if you want to feel good you read a few comments which reinforce your view, if you want to learn more you read the few well thought out comments which add to your context.
But in either case to much is simple to much to process, to gain Information you have to process the data in front of you, it’s not effortless. Therefor tiring and time constraint.
echo does not present a problem that cannot be logcially and easily handled.
echo should work in tandem with traditional comment option. the UI could handle this in different ways such as a few tabs beneath the post, using a sidebar or even a popup.
echo should not replace your own comment system.
every blogger should want to have control over comments. this is not very different than the topic of owning control of your short tracker urls and using your own domain etc. comments are important enough to control in the same way.
in addition to trackbacks/pingbacks, echo can co-exist. just dont disable your own comment system, even if moderated.
For anyone who still believes in Comments — as I certainly do — the domain name Comments.com is available for development, co-development, as well as looking to acquire technology assets that can power a federated comment platform that ties into the Identity.net federated identity schema and Epik.com’s fast growing semantic media network.
Comments have long been a way for publishers to increase their CPM revenue, as well as a way for the “community” to feel connected.
Having used Echo, it actually does a pretty good job as a comment aggregation tool, especially if you expand your idea of what a comment is to include all the social media outlets.
Yeah, from what I remember… Echo sucked.
Great article. I always appreciate insights that help crystallize what I’m thinking through.
A few thoughts:
1. Most commenters don’t really care about conversations (how often do you check for replies to your comments?); it’s just about giving voice to your opinion.
2. Echo will give lower traffic niche articles a longer shelf life.
3. Echo isn’t the be-all, end-all. But it’s ahead of the crowd in rethinking solutions.
Usenet just never goes away
And penning that list of complaints didn’t inspire a single idea for a solution?
I would like to think that Google Wave could be a comment engine. If they packaged Wave into a “Connect” would that be the engine you are looking for? Editable, searchable, traceable, threaded comments?
Wave could also potentially be a Twitter where you could actually follow a conversation. It could be the ultimate cloud public forum.
Hope Google has big plans for Wave.
“And penning that list of complaints didn’t inspire a single idea for a solution?”
Not at all; the solution is yoomoot.com. I didn’t want to mention it in the article because I didn’t want it to appear spammy.
is anybody reading this?
Yah i just started using Echo on my site for comments, works great, alot better then conventional commenting systems.
Thought the ‘Spectrum’ app that the BBC did for one of their more controversial series was an interesting way of exploring a comments archive. Take a look on this link, you need to click the Spetrum button to launch it http://www.bbc.co.uk/white/
I certainly think it’s interesting, but I don’t think it works very well at all. I first looked at the comments in the “happiness” section and got a comment about how weak UK politicians are compared to Australian ones. It was apparently classed as “happy” because it included the word “good”! Most of the comments I saw seemed to similarly miscategorised.
Also there is a lot of clicking required in ordered to view comments, so it’s impossible to quickly scan through.
Nicolas Holzapfel has been hoisted on his own petard.
Somebody has to say it…
‘no comment.’
Most of them don’t have to be displayed at 100% but at the size i decide in the embed codeAnd how google treat the associated flashvars ? ,