Jakob Nielsen

I have mixed feelings about Jakob Nielsen.

On the one hand, he’s fabulously famous (for an HCI practitioner, anyway). People who can barely spell usability are still quite likely to know the name and may even be subscribers to his Alertbox newsletter. He’s also basically sound. That is, the advice he gives is pretty much in line with what most of us would offer our clients. And, hugely to his credit, he is a great promoter of the idea that designers should base their decisions on evidence, gathered from real users.

On the other hand, the evidence he gathers is hardly what you’d call scientific. In fact, he is a good example of how people in HCI get away with abysmally low standards of evidence for their claims. There is no sign of rigorous experimental design or procedure in his reports, and the merest, occasional nod in the direction of hypothesis testing, replication and peer review. This is where his popularity becomes a problem.

It has happened a few times to me in recent years that I have had a design decision challenged on the basis that ‘last month’s Alertbox said…’ which then kicks off an hour-long discussion about the contingent nature of research findings, the need for proper hypotheses, the need for controls, other evidence from more reliable sources, the special circumstances of our own user, task, technology mix, and so on and so on. Not that I really mind having these conversations – it’s not often you get a chance to convince your client that you really do think about what you’re doing – it just bothers me that most people who read Nielsen take it as usability gospel, not just one man’s particular experiences.

I’m sure Nielsen would be the first to agree that blind faith in his pronouncements is a bad thing – but I suspect he might argue it’s not so much of a bad thing as not reading any usability guidelines at all. And, given that he is, after all, on the side of the angels, and he is, on the whole, pretty sound, I suppose I’d have to agree. Besides, the guy is a businessman with a living to make, and that no doubt involves presenting his slightly dubious evidence as rather more authoritative-sounding than it really is. So there is no point expecting him to qualify everything he says the way a scientist would.

Still, it makes me uneasy. The whole field is dogged by what amounts to anecdotal evidence being passed off as serious research. There is almost nothing in the way of solid theory in HCI and nobody seems to think this is a problem. Essentially that makes us a collection of craftspeople – not even engineers, and certainly not scientists. And the field is full of ‘gurus’ whose advice comes mostly from personal experience and not from good research.

Oddly enough, his business partner, Donald Norman, is someone I have considerable respect for and who has made some (of the very few) significant contributions in our meagre attempts to develop HCI theory.

Déjà vu

I’ve just read an interview between Jared Spool and Hagen Rivers and I’m feeling pretty depressed by it (even though the transcript was brimful of laughter!). I know that this interview was a thinly disguised advert for a report that they’re trying to sell (someone should do a book on ‘Salesmen and Showmen in HCI’) but that wasn’t what was so awful. It was the content of the report that upset me.

I must come clean here and say I have not read the report. Frankly, if it has in it what these two say is in it, I never will. It seems to concern an insight that Ms Rivers has had that the structure of Web applications can mostly be described as a set of ‘hubs’ and ‘interviews’. A ‘hub’ is just a page from which you can reach several other pages, and an ‘interview’ is a linked sequence of pages. Now I expect most of you are saying something like ‘Well, duh!’ or ‘Surely there has to be more to it than that!’ – which is pretty much what I said too.

The fact is that anyone who has been in this business for five minutes, or has ever used a Web application for that matter, will see that this is a statement of the blindingly obvious. Yet, for some reason, Ms Rivers believes this to be a profound insight and so does Jared Spool. I hate to sound like a bitter old fart but this was the kind of insight that was old hat more than 20 years ago, in the early days of hypertext, before the Web was even invented. I published a paper back then describing metrics for analysing a number of common navigational structures (and rather more sophisticated ones than the ‘hub’ and ‘interview’ structures that have just been ‘discovered’ – see Canter, D., Rivers, R. and Storrs, G. (1985). “Characterizing User Navigation Through Complex Data Structures”, Behaviour and Information Technology, 4(2),93-102.) If Ms Rivers or Mr Spool had bothered to look, they would have found many other papers on the subject too. In fact, a moderately large literature on hypertext navigation and structure that makes this exciting new report of theirs look rather silly.

If this was an isolated incident, it wouldn’t be so bad but the Web is full of ‘gurus’ like this who are discovering things that were well known and even well understood may years ago. God knows, I’m no scholar but I get the impression that these ‘experts’ never open a journal to see what has been done before. They just have their amazing ‘insights’ and splurge them out as if it was news.

Human-computer interaction (oh sorry, we’re calling it ‘interaction design ‘ this year) has always been an almost theory-free zone but there are people out there doing proper studies and seriously trying to accumulate understanding. The least we could do is look at their findings from time to time. Isaac Newton famously commented that, if he saw a little farther than others, it was because he was standing on the shoulders of giants. Well, there are very few giants in this field but there are plenty of midgets we could be standing on if we were really interested in seeing a bit more.

All-in-all, the discipline could do with a bit more Newtonesque humility and a little less guruism.