mypipeline

Back in March 2007 Freakonomics author Stephen Dubner wrote this tiny blog post for the NY Times asking a question about why people comment on blogs. Needless to say, it attracted much more than the average comment count for a blog post (4x as many as a successful post according to Dubner’s math.) There are many thoughtful responses to Dubner’s question, but one that got me thinking was this:

Someone should do a study about blog commenters. I think blog commenting is a sociological phenomonen that deserves attention. And the outcome of such a study could be used to dramatically enhance the value of the good blogs. There are the commenters who actually have some facts to add to the subject. There are the commenters who pretend to have some facts to add to the subject. There are the commenters who have only their opinion, and often ill-informed ones at that. There are the commenters who couldn’t care less about the post itself but comment as a form of social interaction. There are the commenters who feel compelled to offer a comment on every else’s comment. And, most common of all, there are the commenters who consider commenting to be a competitive sport and have to have the first and/or last word … I personally would prefer to read only blogs that attract comments from people who have something intelligible (usually facts, but occasionally well-informed opinion) to add to the discussion so that everyone can learn something from the experience of reading the comments..

This puts into perspective a pretty common view, not only of the people who comment frequently on blogs, but of how their intentions are perceived. Which got me wondering about several sources of frustration that users have when browsing blogs - especially popular ones

• Comment streams (also called ‘zones’) are so long that users cannot make sense of the ongoing discussions within them
• There is no easy way to distinguish ‘intelligible’ comments, to borrow the phrase above, from throwaway comments
• There are not enough incentives to provide considered comment using the majority of comment services
• Users find it difficult to both track and respond to ‘multi-threaded’ conversations inline
• Too many websites’ comments area have a legacy of web 1.0 ‘feedback forms’ which were the antithesis of conversation starters

For the past six months we’ve been looking at how we can improve the user experience of social media on the Channel4.com website. By social media, I mean both community features and audience interactions. Whilst I am of course keen to engage with communities around channel4.com’s content, I have been focusing most of my thinking time to the latter group of functionality, and two features in particular - commenting and rating. Two reasons really. One, I am disappointed with the current service of commenting and rating on C4.com enough to want to change it; and two, it’s a tougher problem in my view than setting up a community with a single purpose. Lots of packaged solutions exist out there that can enable any old hack to set up a community around their favourite content. See Kickapps, OneSite, and  Small World Labs for a few front-runners in this field. And granted, there are also vendors who market ’social tools’ as components that you run alongside a traditional CMS, and power the users’ side of the content contract. Pluck and Pringo are the heavyweights in this arena. And then there are specialists, such as Disqus, which offer a SaaS solution primarily aimed at bloggers, and others who want to aggregate their users’ comments and personalise the proposition a little.

At channel4.com we have a tricky situation. It’s not news that we’re in the process of redefining how we produce our programme information pages. There’s a team beavering away right now, trying to nail both a consistent user experience and a well-wrought product that can claim to be the authoritative source for all things around a given C4 programme. It’s no small effort, as the BBC have recently found with the launch of their own /programmes pages.

Of course we’d like to offer our audience the ability to both discuss our content where the content sits, and also be able to pass some judgment over others’ commentary (and our own!). In short - to facilitate a dialogue between the channel’s content producers and its online audience. It was one of the core goals of the project from the inception: engage with the wider web community.

However, we’ve got at least four distinct types of programme, and the kinds of feedback that we expect and receive for each type varies as much as the linear content. I define the four types of comment we most often receive as:

1. Shout Out: user makes a short, typically singular statement that is of a more ‘throwaway’ nature than your average email, say. These comments can most often be found on entertainment programmes such as Skins and Big Brother. What the user expects when commenting in this way is like the experience of a Facebook user’s wall or indeed the Shout Box on Last.fm. These folks aren’t going to be reading others’ comments too often, and certainly don’t need to ‘rate’ each other’s input. However, they are often signed up to at least one social network and as such appreciate the ability to ’stamp’ their comments with something more visual than a handle - either an image or a video.
2. Review: user actively reviews a piece of content. See Film’s user reviews or the 4Car site’s owner reports. These comments are usually more focused, but still unlikely to be dialogue. It’s a case of broadcasting one’s opinion of a non-contentious subject or product, and so rarely warrants either an editorial oversight or indeed the functionality to allow users to respond directly to other users.
3. Debate Opener: user offers a clear opinion, and has taken the time to consider their comment. You might normally find their comments in a message board setting, or on A-list blogs. This can be, as in the case of the TV Show’s commenting system, instantiated by the C4 editorial team, and ‘hosted’ with some degree of mediation. See this page for a good example, albeit brief. Factual programmes, unsurprisingly, tend to garner the most debates. They are currently housed in forums, such as this one for a recent documentary on Muslims in the UK. Though I value message boards for their no-nonsense approach to chat and conversation, I think it’s a bit of a shame that these debates are held ‘off site’ in that people cannot comment at the source of content.
4. Debate Response: user sets their stake in the sand on one side of an argument, and if not on a message board, gets around the lack of a threading system (usually found in forums) by simple stating a ‘responding to X’ at the beginning of their comment.

So how do you design a commenting system that addresses the goals of each type of commenter? I was curious as to whether there were some companies out there that were facilitating conversation and response a little differently, so I thought I’d do a quick polar map of the online conversation landscape for my research. There are two axes - visually-enriched to text-only, and interactive to passive. On the more innovative end were the self-moderating UGC sites, particularly with a video focus. Viddler, YouTube and Seesmic all had replying functionality along with the ability to send a video response. In two of the three you can also rate a fellow commenter, always with a thumbs-up/down idiom. On the more traditional side of things were the papers, broadcasters and most popular blogs (note, for instance, that whilst Flickr encourages commenting on its photo pages, there are no comments on its blog). Somewhere in the middle are most blogs.

The Guardian has just signed up to the Pluck service, and as such benefits from its rating tool which allows users to ‘dig’ a comment. However, that’s as far as the functionality goes. One of the advantages, surely, of being able to rate a user’s comment is to have that data collected and used elsewhere, either collated somewhere as the ‘top rated voice’ of the public somewhere, or simply allowing your profile to aggregate what others’ have thought about your comments. But unfortunately the Guardian’s service doesn’t do either, rendering the ability to rate or slate someone’s view rather meaningless.

There’s another observation that one can’t help but notice occurring naturally on sites with long comment streams without threading: people use the @ symbol to address a comment to another user. However, there’s another problem with the experience at this level, too. Whilst this ad hod emergent, user-generated system works well for the first few ‘responses’, when more data is delivered, the model breaks down - suddenly there’s no way of determining whether @someuser is talking about someuser’s first comment or sixteenth. Try making sense, for instance, of this comment stream on Big Brother.

This is a scenario that Twitter users will be familiar with. One can reply to any tweet from another user, but that reply, when delivered to the stream, is taken out of context - there’s no way to understand what the reply is to. This is acceptable for Twitter chat, because generally the conversation is between friends, and there is a sense of understanding about the context. But not so for casual browsers dipping into a comment stream on, say, the Guardian’s blog posts.

My trawling through various streams looking for ways to improve the user experience of reading comments and being able to send a comment led to a few thoughts about what could work to ease the information overload one feels when faced with a big comment stream. Dunstan Orchard was experiencing similar concerns with his comments stream and took matters into his own hands with this visual guide to navigating comments. Although a good attempt at conveying order to an otherwise chaotic interface, I think it suffers a little from trying to do too much.

One idea that appeals to me is trying to deal with comment replies in a flattened fashion, hiding the thread layers.  I sketched out this early concept to show folks here at C4. The idea is that instead of showing the natural nesting of comment replies, you would instead serve an overlay which aggregated the responses to a comment inline, with the ability to then respond to a respond if needed. Clicking on a reply icon would preload your submission form with a label saying ‘In response to’ or simply @ and the number and author of the comment. In this way, you could quickly scan to see which comments had deserved the most reaction, and it also would mean that wherever your attention was fixed on the comment stream, you could pull in responses inline, without having to skip up and down the page unnecessarily looking for the conversation. It effectively hides a nested experience, and makes the act of replying more accessible.

Hiding the visual complexity of nesting is, for me, a move forward in the evolution of online conversation. Another is to actually use one’s rating service to organise information on the comments stream. As mentioned above, I think there’s an incentive to contribute thoughtful, worthwhile comment if other users’ rating of your comment results in your comment beng selected as the most valuable. Having what I call a ‘Chatter Box’ that shows whose participation in the thread has received the most praise should raise both quality and quantity of submissions.

Being able to define what type of comment a user is making is another way of visualising the data collected from a comments stream. One of the ideas to have come out a brainstorm session recently was a panel of tags ranging from ‘thoughtful’ to ‘revolutionary’ to ‘mainstream’ that users could click to assign a judgment value to a given comment. Granted it’s a bit zany, but it’s tackling the issue of how to use collaborative filtering to let the users decide what bubbles up as ‘good stuff’.

Got any thoughts about how commenting services and ratings online could be improved to facilitate better participation? Let me know.

(By the way, I’m well aware of how basic the commenting service is on this blog, thank you)

No Comments | In: Social Media, UX | tags: channel4, comments, twitter. | #

If you get a moment read 37 Signals’ reponse to Don Norman’s accusation that their arrogance to ‘design for themselves’ is the wrong path. It’s a) a very good riposte to Norman’s cogent but dated arguments, and b) a fantastic jumping off point to other articles, in particular this interview of Steve Jobs (where he makes many of the same arguments for ’selfish’ design).

The quote from Norman’s original article that riled me was this:

“Moreover, we purchase on features, not on their absence, and so the successful business must always face this tradeoff: the very things that customers complain about afterwards are what caused them to [purchase the item in the first place].”

One word, Don: iPod. The original iPod’s almost childish lack of features was one of its most redeeming, er, features. People - not technophiles with CS doctorates but everyday people - learned to love the iPod overnight. And that’s exactly the spirit that 37Signals engages in, and why I’m a fan of Basecamp. Yeah - it’s simple. But that’s the point. Someone used to Confluence or Sharepoint might look at Basecamp and think ‘Gee, this doesn’t have such and such a feature that I use about once a year’. Well, so what? For the 80% of tasks that you might use a project planner for each day, Basecamp will be more intuitive, faster to interact with, less expensive and easier to convince your colleagues to use. I know, because I’ve used both

And there’s the other great reason to not do everything for all people: you actually get stuff out the door. These days, if you keep trying to choke your product with new features, you’ll be left scrambling to pick up the odd straggler who didn’t already sign up for your competitor who put out their product that does a few things really well.

Norman did post an addendum to his article that sounded terrifically similar, actually, to Jason Fried’s rebuttal of his arguments:

As expected, the publication of this note has released a flood of responses, so let me use them as an excuse to clarify my writing.

One correspondent wrote: ” I think you’re somewhat mistaken in your evaluation of 37signals. To them, feature-bloat in web applications is akin to food service and seat reservations for Southwest Airlines. Application simplicity and usability are what the customers need most.”

I do not disagree with the comment: I simply do not believe that arrogance is the solution. Feature-bloat is horrible. 37signals is correct to be annoyed. But the disdain they show for their customers is not just arrogance: it is selfishness. The solution is not ignorance of the needs of your customers. Their approach is both arrogant and selfish.

The solution is to decide which customers represent your core audience, and then to observe them at work, the better to understand their true needs. (Not by asking them, not by questionnaires, not by focus groups). Rapid iterations of prototype and evaluation is the key. The iterative design method of rapid prototyping, test, and iteration (all done within the span of a day or so) is well defined in the Human-Computer Interaction community. It starts with observation and understanding. It then proceeds through rapid prototyping and test, continually refining the project scope and definitions.

The mark of the great designer is the ability to provide what people need without excessive complexity, without feature bloat. Simplicity should never be the goal. Follow the famous Einstein quote: “Everything should be made as simple as possible, but not simpler.” Complex things will require complexity. It is the job of the designer to manage that complexity with skill and grace.

I don’t think fundamentally there’s much difference in their approaches: it seems that Norman’s just not a fan of upstarts such as Hansson and Fried - possibly because they are too boldly confident in their own talents and intuitions, and assumptions that by designing a tool that they would get enjoyment and use out of, they would end up with a successful product. But what’s fundamentally wrong with this kind of approach? It’s certainly served Apple well over the years, if the Jobs interview is anything to go by.

The old Henry Ford quotation came up at least once in the conversation, and it’s one that’s more and more relevant in a world where people don’t necessarily understand how the applications of the future are going to affect them, and whether they think they would actually use or come to need an application in their lives: Ford said, ‘If I’d have asked my customers what they wanted, they would have told me “A faster horse.”‘ For me it also comes down to this: if you give something that’s completely new to a ‘user’, they will immediately buck, complaining that it’s not what they’re used to and that they don’t like it. That’s a recipe for staid design in my book. If we only adopted features that 100% of our users knew how to employ 100% of the time, we would never create new metaphors of interaction and interface, and more importantly, the medium and the mechanism would never become art. So sometimes, we have to be brave enough to strike out on our own.

Regardless, it’s a good thread to become absorbed in - two great thinkers exposing their thoughts for the world.

No Comments | In: UX, Web Apps | tags: 37signals, apple, basecamp, usability. | #

The two things I’ve been obsessing about the most recently are TV shows and data visualisation. And in a moment of synthesis today, I thought that the latter could actually come in quite handy for dealing with some of the former: that new breed of almost Pynchon-esque complex storylines that modern serial dramas are passing along to the audience. I am imagining a visual representation of subplots and characters, the frequency of their appearance, when they are killed off, when certain events in one subplot trigger events in other subplots etc.

I just think this would be a beautiful and intrinsically novel way of exploring (and continuing to explore, season by season), some of the complexities in today’s big storylines - 24, Lost, Heroes etc. I’ve had a quick search on IBM’s ManyEyes project for similar stuff, but it’s all a bit boring and impenetrable. High time for someone to kick arse and develop a truly stimulating lose-yourself-in-this visualisation for a big show - like how the Lost Untold did for the hidden bits of the unfolding story in the first and second series on Channel4.

No Comments | In: Data Visualization, Future Telly, UX | tags: manyeyes, storytelling, TV. | #

A couple months ago Platform4 launched a creative code contest for developers keen to mash up Channel4’s Film4 content with other content from around the web.

We had some great entries but it was Thomas Butterworth with his enhanced cinema search who was crowned the overall winner and awarded £1500 for his mash-up of Google Maps, YouTube movie trailers and Film4 reviews.

The judges were impressed with Thomas’ app’s ease of use, clear utility and how it added clear value to the Film4 reviews that underpinned it. In the words of one judge, it’s a simple idea, well delivered.

Our two runners-up excelled in creating quirky apps that played with ideas of traditional navigation and social interactions.

Daniel Hilton’s group-decision engine for choosing a film to watch and what takeaway to eat on the night impressed us with its kooky nature and attempt to marry usefulness with the social aspects of ‘movie nights’. It also employed a postcode lookup tool, but instead of searching for cinemas near you, found food delivery places nearby.

Andrew Chalkey’s feedreading service with built-in previews and blog reactions took joint second prize. It combined trailers from Apple’s trailer service and the latest blog reviews of the films from Technorati. Both Daniel and Andrew won £250 for their trouble.

We’ve been really impressed by all the submissions received – it’s been a great competition. Hopefully we’ll be able to run similar contests with more Channel4 feeds later in the year. (Maybe even using some RESTful C4 services

No Comments | In: Data Visualization, UX, Web Apps | tags: channel4, mashups, platform4. | #

I spotted this the other day and immediately thought: yes - a perfect use of a web app. Mint.com is part account activity aggregator, part implicit spending behaviour monitor, part deal-finder. In other words, it’s a low-maintenance budgeting software package that works on your behalf to sort out your finances. Pretty cool. Now that I’ve got a wife and kids to look after, I worry more about the old pennies, and I’ve fired up Excel a couple times and tried to work out the sums a bit. But I think I’m not alone in the habit of forgetting about the spreadsheet about a week later not bothering to update any information in it, plummeting into a guilt spiral. Well, it’s never that bad, but you know what I mean.

So when I took the tour of this US-only web app and saw the feature list it appealed instantly: once it knows your account details (yes the security question looms, but heck, if I bank online anyway, what difference is it?) it can categorise your spending by checking what the debit went on, then suggest ways of cutting back on those outlays. In fact, it even emails you if you’re about to go overbudget on a particular category, say ‘Beer and Pizza’. It can also offer you discounts on products that it think could save you money. Having just switched current accounts myself, I wish I’d had this facility ages ago. Anyway, the downside is that it’s US-only for the time being, though a UK launch is planned in ‘early 2009′ according to the corp blog.

No Comments | In: UX, Web Apps | tags: banking, finance. | #