[For those not subscribing to CACM, Jakob Nielsen and I have come down on opposite sides of a usability debate. Jakob believes that the prevalence of bad design on the Web is an indication that the current method of designing Web sites is not working and should be replaced or augmented with a single set of design conventions. I believe that the Web is an adaptive system and that the prevalence of bad design is how evolving systems work.

Jakob’s ideas are laid out in “User Interface Directions for the Web”, CACM, January 1999.

My ideas are laid out in “View Source… Lessons from the Web’s Massively Parallel Development”, networker, December 1998, and http://www.shirky.com/writings/view-source

Further dialogue between the two of us is in the Letters section of the March 1999 CACM.]

Jakob,

I read your response to my CACM letter with great interest, and while I still disagree, I think I better understand the disagreement, an will try to set out my side of the argument in this letter. Let me preface all of this by noting what we agree on: the Web is host to some hideous dreck; things would be better for users if Web designers made usability more of a priority; and there are some basics of interface usability that one violates at one’s peril.

Where we disagree, however, is on both attitude and method – for you, every Web site is a piece of software first and foremost, and therefore in need of a uniform set of UI conventions, while for me, a Web site’s function is something only determined by its designers and users – function is as function does. I think it presumptous to force
a third party into that equation, no matter how much more “efficient” that would make things.

You despair of any systemic fix for poor design and so want some sort of enforcement mechanism for these external standards. I believe that the Web is an adaptive system, and that what you deride as “Digital Darwinism” is what I would call a “Market for Quality”. Most importantly, I believe that a market for quality is in fact the correct solution for creating steady improvements in the Web’s usability.

Let me quickly address the least interesting objection to your idea: it is unworkable. Your plan requires both centralization and force of a sort it is impossible to acheive on the Internet. You say

“…to ensure interaction consistency across all sites it will be necessary to promote a single set of design conventions.”

and

“…the main problem lies in getting Web sites to actually obey any usability rules.”

but you never address who you are proposing to put in the driver’s seat – “it will be necessary” for whom? “[T]he main problem” is a problem for whom? Not for me – I am relieved that there is no authority who can make web site designers “obey” anything other httpd header validity. That strikes me as the Web’s saving grace. With the Web poised to go from 4 million sites to 100 million in the next few years, as you note in your article, the idea of enforcing usability rules will never get past the “thought experiment” stage.

However, as you are not merely a man of action but also a theorist, I want to address why I think enforced conformity to usability standards is wrong, even in theory. My objections break out into three rough categories: creating a market for usability is better than central standards for reasons of efficency, innovation, and morality.

EFFICENCY

In your letter, you say “Why go all the way to shipping products only to have to throw away 99% of the work?” I assume that you meant this as a rhetorical question – after all, how could anybody be stupid enough to suggest that a 1% solution is good enough? The Nielsen Solution – redesign for everybody not presently complying with a “single set of design conventions” – takes care of 100% of the problem, while the Shirky Solution, let’s call it evolutionary progress for the top 1% of sites, well what could that possibly get you?

1% gets you a surprising amount, actually, if it’s properly arranged.

If only the top 1% most trafficked Web sites make usability a priority, those sites would nevertheless account for 70% of all Web traffic. You will recognize this as your own conclusion, of course, as you have suggested on UseIt (http://www.useit.com/alertbox/9704b.html) that Web site traffic is roughly a Zipf distribution, where the thousandth most popular site only sees 1/1000th the traffic of the most popular site. This in turn means that a very small percentage of the Web gets the lion’s share of the traffic. If you are right, then you do not need good design on 70% of the web sites to cover 70% of user traffic, you only need good design on the top 1% of web sites to reach 70% of the traffic.

By ignoring the mass of low traffic sites and instead concentrating on making the popular sites more usable and the usable sites more popular, a market for quality is a more efficient way of improving the Web than trying to raise quality across the board without regard to user interest.

INNOVATION

A market for usability is also better for fostering innovation. As I argue in “View Source…”, good tools let designers do stupid things. This saves overhead on the design of the tools, since they only need to concern themselves with structural validity, and can avoid building in complex heuristics of quality or style. In HTML’s case, if it renders, it’s right, even if it’s blinking yellow text on a leopard-skin background. (This is roughly the equivalent of letting the reference C compiler settle arguments about syntax – if it compiles, it’s correct.)

Consider the use of HTML headers and tables as layout tools. When these practices appeared, in 1994 and 1995 respectively, they infuriated partisans of the SGML ‘descriptive language’ camp who insisted that HTML documents should contain only semantic descriptions and remain absolutely mute about layout. This in turn led to white-hot flamefests about how HTML ‘should’ and ‘shouldn’t’ be used.

It seems obvious from the hindsight of 1999, but it is worth repeating: Everyone who argued that HTML shouldn’t be used as a layout language was wrong. The narrowly correct answer, that SGML was designed as a semantic language, lost out to the need of designers to work visually, and they were able to override partisan notions of correctness to get there. The wrong answer from a standards point of view was nevertheless the right thing to do.

Enforcing any set of rules limits the universe of possibilities, no matter how well intentioned or universal those rules seem. Rules which raise the average quality by limiting the worst excesses risk ruling out the most innovative experiments as well by insisting on a set of givens. Letting the market separate good from bad leaves the door open to these innovations.

MORALITY

This is the most serious objection to your suggestion that standards of usability should be enforced. A web site is an implicit contract between two and only two parties – designer and user. No one – not you, not Don Norman, not anyone, has any right to enter into that contract without being invited in, no matter how valuable you think your contribution might be.

IN PRAISE OF EVOLVEABLE SYSTEMS, REDUX

I believe that the Web is already a market for quality – switching costs are low, word of mouth effects are both large and swift, and redesign is relatively painless compared to most software interfaces. If I design a usable site, I will get more repeat business than if I don’t. If my competitor launches a more usable site, it’s only a click
away. No one who has seen the development of Barnes and Noble and Amazon or Travelocity and Expedia can doubt that competition helps keep sites focussed on improving usability. Nevertheless, as I am a man of action and not just a theorist, I am going to suggest a practical way to improve the workings of this market for usability –
lets call it usable.lycos.com.

The way to allocate resources efficently in a market with many sellers (sites) and many buyers (users) is competition, not standards. Other things being equal, users will prefer a more usable site over its less usable competition. Meanwhile, site owners prefer more traffic to less, and more repeat visits to fewer. Imagine a search engine that weighted usability in its rankings, where users knew that a good way to find a usable site was by checking the “Weight Results by Usability” box and owners knew that a site could rise in the list by offering a good user experience. In this environment, the premium for good UI would align the interests of buyers and sellers around increasing quality. There is no Commisar of Web Design here, no International Bureau of Markup Standards, just an implicit and ongoing compact between users and designers that improvement will be rewarded.

The same effect could be created in other ways – a Nielsen/Norman “Seal of Approval”, a “Usability” category at the various Web awards ceremonies, a “Usable Web Ring”. As anyone who has seen “Hamster Dance” or an emailed list of office jokes can tell you, the net is the most efficent medium the world has ever known for turning user preference into widespread awareness. Improving the market for quality simply harnesses that effect.

Web environments like usable.lycos.com, with all parties maximizing preferences, will be more efficent and less innovation-dampening than the centralized control which would be necessary to enforce a single set of standards. Furthermore, the virtues of such a decentralized system mirrors the virtues of the Internet itself rather than fighting them. I once did a usability analysis on a commerce site which had fairly ugly graphics but a good UI nevertheless. When I queried the site’s owner about his design process, he said “I didn’t know anything when I started out, so I just put up a site with an email link on every page, and my customers would mail me suggestions.”

The Web is a marvelous thing, as is. There is a dream dreamed by engineers and designers everywhere that they will someday be put in charge, and that their rigorous vision for the world will finally overcome the mediocrity around them once and for all. Resist this idea – the world does not work that way, and the dream of centralized control is only pleasant for the dreamer. The Internet’s ability to be adapted slowly, imperfectly, and in many conflicting directions all at once is precisely what makes it so powerful, and the Web has taken those advantages and opened them up to people who don’t know source code from bar code by creating a simple interface design language.

The obvious short term effect of this has been the creation of an ocean of bad design, but the long term effects will be different – over time bad sites die and good sites get better, so while those short-term advantages seem tempting, we would do well to remember that there is rarely any profit in betting against the power of the marketplace in the long haul.