"ux" in Weblog

After seeing Sources of Power referenced by two of my favourite books, Don’t Make Me Think and Blink, and having it recommended to me by a chap at work I finally got around to reading it. In the book Gary Klein studies the decision making process exhibited by people under time pressure: firemen, tank commanders, chess players and the like. Conventional wisdom tells us that people weigh their choices and then make a rational decision based on the result. What Klein and his researchers found though is that, under pressure, people subconciously recognise patterns from their experience and use them to reach a decision without any consideration for the alternatives - there just isn’t time. If that option doesn’t work out then they’ll jump to the next most reasonable one. This they called the RPD model - Recognition Primed Decision-making Model.

Reading the details of Klein’s RPD model really reminded me of the way I use the internet. I am an impatient and impulsive user and I have also seen this behaviour exhibited by users during testing. Not all users though… From what I’ve seen, experienced users, in particular, tend to be impulsive.

I suspect that this is because as we use the internet we build these usage patterns into our subconcious. The more we use it the more we rely on these conventions. And, in certain circumstances anyway, we are under self-imposed time pressure. We know that there is a lot of content on the internet and often what we are looking for isn’t of great value so we move on if we don’t immediately find what we want. We don’t commit a lot of time to any one task. As web users I believe that we often fit into the ‘time pressure’ category studied by Klein.

A quick look at the Eye Tracker article on The Google Effect shows the RPD process in action. We have become programmed to believe that the top few Google results will be what we are looking for and we employ the pattern recognition to dictate our actions the first time we see the results page. I, for one, will click on the top link that looks hopeful and quickly scan that page for something that signals that I am closer to my goal. If not then I will go back to the results page and start reading the results items. My behaviour switches drastically - to an comparative, evaluative model.

Support for the spread of this recognition primed behaviour on the web is the recent iProspect/Jupiter media study. In 2002 48% of users clicked on a link on the first page of their results. Today that figure has risen to 62%. More and more people are acting like experienced web users, displaying impulsive behaviour and, I think, behaving as if they were under time pressure. Our users are likely to become more and more impulsive, clicking first and reading later.

So what’s the point of all this?

• Use conventions wisely. If people regognise a pattern they are likely to click without thinking.
• Front and top load everything for quick scanning.
• Use sensible link text and headings for decision making out of context.

To be honest, I don’t think this is anything really new but it does add weight to the rules most of us obey already. I now think that I understand why my users behave the way they do and having this sitting in the back of my mind makes me think very carefully about the way I design my information…

In my redesign post I mentioned this ‘zooming content’ idea I’ve adopted. I don’t think the concept is new but I’ve not seen anyone write about it so I’m jumping in… Each page is based on what I’m calling a ‘zooming’ layout. By that I mean that the information view and relevance on the page zooms out as the page goes down. This is influenced by three things:

1. Derek Powazek’s Embrace your bottom! article.
2. Peter Morville’s Ambient Findability.
3. The Best of Eyetrack III: What We Saw When We Looked Through Their Eyes.

The Eytracker III study from a couple of years ago found that people aren’t afraid to scroll below the fold but that their scanning became more rapid. To cater to this behaviour I have limited the amount of information in the lower sections of the page. The page content becomes more and more brief and scannable as the page goes on. Take the weblog index for example: it starts with a complete entry, then moves to summaries of 5 recent entries, along with access to the archives, and finally to external links.

Every page follows this pattern. The top of the page deals with specifics and details while the bottom of the page presents an overview, encouraging scanning and, hopefully, giving people what they want, when they want it.

Why have I done this instead of going for the more conventional two column blog layout? Conventions are our friends, after all…

In Ambient Findability Peter Morville calls web pages ‘discreet findable objects’ that can be accessed out of context via search engines… Basically, there’s a very good chance that people are coming to a content page from Google having performed a search on a specific keyword phrase. The standard two column layout presents these nice new visitors with a lot of information they’ve not asked for and are probably not interested in. I wanted to let people get what they came for before bombarding them with choices, in line with Derek’s suggestions.

Content is king and will govern both incoming links and page weighting by the engines. If someone gets to one of my pages it will probably be thanks to the content so I’ve made sure I give that up front. The article, blog entry or portfolio item is there in full at the top of the page with nothing else in the way. The main navigation gives a broad overview of the site and the metadata to the right suggests the presence of more local and related posts but I’ve tried very hard to keep noise to a minimum. I’m hoping that I’ve left enough wayfinding and orientation hints at the top of the page to let the user reach a decision about the context of the site and content before continuing down the page.

After the full entry I zoom out one level.

On the index pages this steps out to an overview of the most recent posts, kept short and scannable; On the content pages this is lists of two different types of entry: the most recent ones, which I’d obviously like to promote to new visitors, and related (and hopefully relevent) posts, which are an attempt to push old content. (This second group is found by passing the article title into my site’s search and listing the top results.) Derek Powazek advocates using the bottom of the page to give readers somewhere to go but I had one more question I thought people might ask: Why should I believe the post I’ve just read? By providing a list of links to other things I’ve written I am trying to establish credability with my readers as well as cross-sell content. That’s also what the slightly anomolous ‘about me’ snippet is trying to suggest - that I do have some idea what I’m talking about.

The final portion of the pages is for less relevant or less valuable information - external links, comments, that sort of thing.

I don’t think this is a radical idea - far from it - newspapers have been doing this sort of thing for ever, but I am very surprised that it’s not been applied to more websites. When I sat down to think about what information I should be presenting to my users at any one time this zooming idea just seemed to make sense. My biggest concern with the approach is the way it dresses up normal in-site navigation to look like more content. When I first sent the new design round to people I was expecting lots of complaints about access to the archives. So far no one has said anything about that…

To sum up: people tend to scan the page below the fold so the further down they go how about offering them more scannable content? Zoom detail to overview.

This redesign has been 18 months coming. I did the original design over Christmas 2004 but I just couldn’t find the time to follow it through. Since then the design has been through several revisions, getting simpler each time. This version is the product of an anti-shadow and -gradient phase I was going through about 6 months ago, when all the shiny new Web 2.0 sites were coming out. I’ve mellowed a bit since then… I also really wanted to see a return for natural textures. They’ve fallen from grace since their heyday, back about 4 years, on sites like DNA (now redesigned) and the Hayward Gallery.

For the IA and information design I have experimented with a couple of ideas: zooming content and weighted navigation.

Zooming content, which I’ll write up properly another time, is basically content becomming more general as you scroll down the page. This gives easy and intuitive access to recent posts but makes the pages very dense and limits access to older content. To try and balance this I have experimented with the function of search. It’s normally an afterthought, consigned to the top right corner where everyone can find it but no one is encouraged to use it. With this design I wanted to try bringing search to the forefront of my navigation so it’s given pride of place in the dead center, right next to the primary navigation. This random access navigation is is supported by a fairly standard taxonomy, as well as more granular tagging. I still have about 200-odd posts to tag and perhaps 350 to categorise but I’ll be getting on with that behind the scenes over the coming weeks.

There is also liberal use of weighted lists, something I’d resisted up until now. I’ve used the same treatment for everything: categories, tags, chronology and projects all use size to convey importance. I’ve not seen them used as extensively as this and I’m not convinced that it will work but we’ll see…

The biggest change round here, for me anyway, is under the hood. I’ve finally diteched the ropey old CMS I’d been using. I wrote it about 5 years ago as a database editor for Access and I haven’t updated it since. Not that useful, really… So I am incredibly happy to welcome Symphony to donotremove. It was a long time coming but it’s been a pleasure to work with. It’s XML/XSL based, which gave me a fairly steep learning curve but now I’m on top of it I love it. XSL is incredibly powerful, even if it is somewhat clunky, and there is no end of documentation available on the web. I’ve tried Wordpress, Textpattern and Expression Engine, and none come close to Symphony. It just thinks the way I think a lightweight CMS should.

And with Symphony has come commenting. This is a new thing for me so we’ll see how that goes…

MySpace has become incredibly popular in the last year. It now boasts somewhere over 55 million members with another 150,000 people signing up every day. That’s simply incredible. It’s approaching the population of Britain and growing 150 times faster! And it’s launched its first (as far as I know) number one act on this side of the pond, the Arctic Monkeys, as well as propelling Clap Your Hands Say Yeah to half a million sales in the first week of their album’s release. It’s getting to be more society than social…

Despite its success, everyone I know who’s signed up slates its design. Now, the web development community puts a very high value on great design but we’re obviously not getting the big picture… The site is built on a great idea. Wood for the trees. I read somewhere recently that if every entrepreneur stopped to make the right decisions then many of our farourite products wouldn’t exist. I’ve seen this first hand ar Wordtracker: Mike and Andy wrote the system in 3 months and got it out there. Since then they’ve been working hard to do things again, the right way, but it’s an incredibly long process. If they’d done it that way from the beginning they would never have been able to afford the time and the product woudn’t exist.

This reminds me of Peter Morville’s user experience diagram with the interlocking hexagons combining to make a useful application. The sucess of a website doesn’t depend comletely on any single thing. An acceptable design will satisfice if the end result is great. All those folks out there writing webapps on the Web 2.0 wave would do well to remember this. Start with that great idea that solves a real problem and people will come.

.Net magazine asked me (amongst dozens of others) to give them my predictions for next year. The article’s now out and I feel a little bit mis-represented - they seem to have missed my point entirely. To set the record straight here’s what I wrote for them:

I think that next year is going to be incredibly exciting for interaction and user experience design. Right now we are seeing the reinvention of the web application. AJAX has been met with rabid enthusiasm and Dale Dougherty’s Web 2.0 label has crossed over into the mainstream press.

Microsoft recently announced that they will be moving Office online and unveiled live.com to tap the online services market. Google has already made its intentions clear with service offerings like GMail, Google Maps and Google Reader. Yahoo! has entered the fray with some very interesting experiments in user experience, Flickr being my favourite but Yahoo! Mindset is an ingeniously simple enhancement to organic searching. Then throw delicious, Basecamp, Listal, Remember the Milk, Sproutliner, Netvibes, Technorati, Num Sum, Writely, Rojo, ProtoPage, TiddlyWiki and all the other independent apps into the mix. Finally, add the new or forthcoming offerings from high profile web designers like 37 Signals, Adaptive Path, Firewheel Design and Shaun Inman as they establish themselves as application developers. Everyone seems to have a web app in the oven. That’s a very rich and diverse online platform in the making…

These applications require more intense workflows than anything we’ve seen before on the web. I expect to see masses of experimental interaction accompanying new and existing web apps in the name of user experience while everyone figures out what works and what doesn’t. If anything, I think the high end Flash designers probably have a head start, having dealt with interaction issues like latency and interface feedback before. I am looking forward to plenty of experimentation in rich interaction with some blazing successes that change how we use the web but many, many dismal, unusable failures.

From a user experience point of view this is a magical development… The good folk at Yahoo have identified that people search with different tasks in mind and come up with Yahoo Mindset. It’s only a small tweak on the surface, and not something that had ever occurred to me to want, but after battling with search results noise while trying to buy something online recently, I think this is a very exciting addition. And it’s incredibly simple: a slider to describe the kind of search you are performing. Are you researching or shopping? Two distinct mindsets. I will be playing with this in the coming weeks.

This reminds me of something I’ve been battling with recently - gleaning user requirements from user feedback. Identifying what’s not said as well as what is; tracing a series of requests for interface tweaks to their source in a deeper-rooted problem. I’ve still not got this nailed down though. I can sit here and stare at feedback notes for hours and not come up with anything until my walk home when it suddenly occurs to me what’s been staring me in the face all day. Interface design is a fine art but interface redesign is a black art…

This is an absolutely fundamental question, and one I thought I had a pretty good answer to, but an interesting post Miss Manners Guide to Opening Links in New Windows on Subtraction (my new favourite web design) got me thinking… What right do we have to force our user experience decisions on our users?

I know how I use the web but I also appreciate that I’m not a typical user: I live on the internet, and that obsession has moved me into a kind of super-users group. I use a non-standard browser, I know what all 6 of my mouse buttons do and I even know the keyboard shortcuts for the most common actions. While writing a reply to the Subtraction post I realised how easy it is to patronise users - making assumptions about the user experience without properly considering the alternatives and just assuming that I know best. It’s so important to keep asking how your decisions will affect the user and why you’re making them…

All that said, I’m also not sure I entirely agree with Mark Hurst’s The Page Paradigm either. His view that users are ALWAYS looking for something on a website seems to be more aimed at ‘proper’ sites - e-commerce, education etc, rather than entertainment ones, where exploration is half the fun. I still think I’ll be coming back to his Practicing the Page Paradigm steps as a way of refocussing my attention on what’s important though…

I’ve been noticing a certain degree of homogeny across the websites I visit of late. It’s probably always been there and combined with my relatively limited selection of subject I shouldn’t be surprised. That said though, I would love to be awed, or at least charmed, a bit more often. I see a LOT of websites every day - it’s my job - and it’s very very rarely that something really jumps out, either for design, technical trickery or personality. Novelty just doesn’t lend itself to every subject and with my lack of patience novelty without some thought for the user guaruntees that I won’t stick around. By the sounds of things Carol has been feeling frustrated with Netdiver submissions… Enough to write an editorial about it.

In the article Reading Online Text: A Comparison of Four White Space Layouts it is revealed that people read fastest when text has no margins and sub-optimal leading. Conversely, comprehension is best with some nice margins and leading. That sounds a bit bizarre to me. but it reminds me of an article I read a couple of years ago. I can’t find the original link but the same findings are given in What is the optimal line length when reading prose text from a monitor? The 1999 experiments by Youngman and Scharff found that text on screen is read faster and more reliably with longer line lengths (8 inches at 12pt) but people preferred shorter line lengths (4-6 inches), while Dyson and Kipping determined that shorter lines (1.8 inches) were hard to read.

The obvious conclusions are that users prefer onscreen text displayed in blocks of 4-6 inches with margins and leading (even though they read fastest with longer lines and no margins).

I’d suggest that these findings support fixed line-length layouts over fluid layouts. I browse the internet on a wide-screen monitor with my browser maximised and fluid sites generally look terrible. Fixing columns, at least using ems would give me comfortable readability despite my extra-wide browser, and would take into account those folk who are viewing with larger text browser settings…