Tuesday, February 06, 2007

Was I Right When I Was Right?

After saying I was right on our decision in dotp not to use inline links and especially after saying it in such a provocative way, I expected to get some comments. And indeed I did, all sensible and informed: "And anyway, why are inline links so inaccessible? Surely they provide user with important context whether using a screen reader or not? and Surely the problem lies with inadequate link text, rather than with inline links? " Saying "it's a philosophical thing" plainly won't get me very far in this company so I resolved to try and explain myself better. I've touched on this very topic before - 2 years ago and more. Broadly I agree with the latter point - one of the manifestations of poor thinking on links is recurrences of "click here to visit the page on benefits" or similar. Whilst thinking about this post and doing some reading I came across this very nice rebuttal to the "click here is ok" theory. But I also see too many links inside a paragraph of text as being confusing, I find myself asking "do I have to follow these links to understand the point that is being made?" If I were using a screen reader, the constant interruptions of links - and the uncertainty of whether I should follow them right then and there (and if not how to get back to them) - would probably drive me mad. Wikepedia is clear on its recommendations:
  1. Do not overlink. Screen readers put each link on its own line.
  2. Create good link descriptions, especially for external links. (avoid "click here!" or "this" kinds of links)
  3. Avoid putting links in section headings, unless the link text is the only text in the title. Screen readers will stop reading the heading title when they encounter a link, and if the link is the first part of the heading title, they will only read the link text. For example, a heading title of "The Simpsons" will be read as "The", and a heading title of "hackers in popular culture" will be read as "hackers".

We're talking here about creating a site that is going to be heavily used - and I hereI mean "used" in the sense of many dozens of authors rather than millions of users reading the pages. Here's my logic for no inline links as a flat statement:

- I believe the more of the editorial policy you can fit into the system logic rather than relying on common sense or documented procedures, the better. As you spread your authors around the country (for governments) or the world (for corporates), in the same way that you want the default style sheet to be chosen centrally, you want some of the editorial policies to be enforced centrally - and simply, without human interference.

- Approving editors are busy people and whilst I imagine that they try and read everything, they can't be expected to - if they did, you wouldn't see so many errors in even mainstream press. So the more they can concentrate on "house style" and "does it address the brief" and the less they worry about spelling, grammar, link standards and so on, the better.

- I think text should be standalone and self-explanatory. Generally I link to another site to provide source context rather than to add further to what I'm saying (although I have to say my own editorial policy is a little vague on this and could certainly do with tightening up). Links at the end of a paragraph reinforce this - those links give you a chance to provide places where further information can be found or where the transaction itself can be carried out, or to address obvious questions that may result.

- Broken links occur all the time, even inside the same site, without much warning. A big site could see dozens of broken links every day, internally and externally. As a result, they need to be fixed with as little human intervention as possible. The only time someone needs to look is if the link is intact but the text on the page has changed materially (i.e. may no longer be relevant). The DWP is a good example of both poor standards and inconsistent application - links are both inline and in lists, page templates vary as you navigate through a site, titles are often links, PDFs are used throughout (don't get me started). Here is an example of bad practice from the DWP's site - it's a single line in the middle of a page: "Find out what has been included in previous editions of the Journal." If this breaks, and the link in the word "journal" is removed, what are we left with? A hanging sentence that makes no sense; is it an exhortation? A test? As Helen commented in her response to my previous post, this can be covered by better editing, but I just don't see those standards applied across government sites, let alone the web generally.

- Following on from the point above, the ability to remove a link entirely with no consequence on checking the text is a winner for me. The goal must surely be autonomous removal of links - if one is removed, no one should ever know that it was there in the first place. That means that breaking links out and keeping them in lists to the side of the page or below the appropriate paragraph of text keeps everything simple.

- Finally, a clean standard that is simple to navigate for all users - disabled or not - and follows a consistent set of design and editorial policies is likely to get more users and, importantly, more return visitors. The dti site is perhaps a good example. It looks to me like they've followed the example of the BBC site. Their page setups are clean and simple - the main body of the text is in the middle of the screen and on the right hand side you can see all the related links, background material and so on. The BBC is a news site, the dti is a government site. Is there a difference? Probably not.

Does that help clear up my position?

2 comments:

  1. Stell Ent9:13 pm

    Good answer. I stand corrected.

    ReplyDelete
  2. Anonymous9:40 pm

    Hmmm yes. I don't do arguing with Alan anymore.

    I saw an interesting advert on the jobserve today. (http://www.jobserve.com/W1CBC924C4D1422FA.job)

    Whilst I'm sure I'll never have DV clearance, I can see my tax being thrown down the toilet on consultants already. This is one of those great wastes of money that management have been conned into stumping up for.

    The simple fact is, you either know at the start of a job there's a limit to the load of a system or you don't. If it's an internal tool for the staff of your typical universal export company or someone, then you know it has to meet that load (say 3000 people.) So you simply design it to handle ten times the capacity, which is child's play, and then get it doing it, thus avoiding wasting money on performance consultants.

    If it's an unknown quantity, but transactional, then you simply design it so that it's completely scaleable by just adding parallel hardware. Then if it isn't fast enough, which you'll be able to see when you soft launch, then you buy more machines, much cheaper than stumping up for performance consultants.

    If it's an unknown quantity and offline, then get some hobbyist maths Phds in, and let them get on with it, because it's going to be beyond the performance consultant's level of ability.

    Either way, I just don't understand this new sector of the IT world.

    Ian

    ReplyDelete