Once again challenging conventional wisdom, Don Norman suggests that human-centred design is not a good idea. Norman’s wonderful essay is worth reading, so I won’t try to summarize it here except to paraphase his point in my own words: focusing on the human (presumably to the detriment if not exclusion of other important aspects of a design situation) doesn’t work because focusing on humans only ignores so many other critical aspects of a design situation, that the results can be catastrophic. (You really should read his essay yourself.)
And while I agree completely with him about not only the inappropriate nature of human-centred design (HCD), but also about the relative superiority of activity-centred design (ACD), I do have some reservations about how he presents his arguments.
Norman tells us that the automobile – one of the most ubiquitous products world-wide – was not designed using HCD. And yet, he says, pretty much anyone can figure out how to use it.
I reply: yes, but only after taking lessons, and a (reasonably hard, in most places) test – which underlies an entire industry of driving schools, branches of the judiciary and law enforcement, and its collateral underbelly. Then there’s the continued risk of loss of control when driving – part of which is due to the controls themselves. My sons are amazed that I can drive as well as I do, considering how “hard” the controls are to them. And this from two boys who can work magic with a game console controller. Even so, I have strong memories of my Learner’s Permit days and how I struggled to learn those ridiculous controls.
No, I think that a proper redesign of automobile controls would revolutionize driving. It would allow drivers’ educators to focus on the “rules of the road” and how to respond to surprise situations. It would also likely lower accident rates – much to the chagrin of the insurance industry (but who cares about them?). I don’t know what that design might be, but only because I haven’t actually done the design. Remember Rule 29: There’s always something better.
So I don’t think the automobile is a particularly good example.
Norman argues that many everyday products are just fine as they are, but weren’t designed using HCD principles.
Some of these objects are so simple – like cutlery – that it’s hard to imagine a need for something as complex as HCD. Still, I’ve noticed that some cutlery works better than other – good cutlery has good balance, the right weight, the right size, etc. Some cutlery is designed for show exclusively and is often nearly useless for anything but display.
And why was it so funny when, in the movie Pretty Woman, Julia Robert’s character couldn’t figure out how to eat snails with appropriate cutlery? Because we’ve all had those cutlery-malfunction moments. Why? Because we can’t get the cutlery to do what we need. Sounds like a design problem to me.
Another example noted by Norman is the typewriter. The typewriter is sufficient, sure, but back in the day when most documents were manually typed – wouldn’t it have made sense to develop a keyboard that allowed people to type faster with less repetitive strain? Maybe the Dvorak keyboard could have been made good enough if HCD had been used. We’ll never know now, of course, either way, but one cannot exclude the possibility.
Norman also notes cameras as continuing to exist without HCD. Again, if one looks at the “classic” cameras from the 1st half of the 20th century (after they’d developed into proper “products” but before their computerization, one finds incredibly difficult to use machines. Over the decades, cameras have undergone automation and computerization. This automation makes them simpler (per John Maeda‘s principle of “hiding” complexity from users). In this case, it seems that automation happened first and accidentally addressed HCD issues by simplification. But they’re still not designed right. There are many people who can’t take a good picture (this characteristic seems to run in my wife’s family) no matter what the camera looks like and no matter how it works; there aren’t cameras for left-handed people; people with long fingers (like me) have a very hard time using many new cameras;….
Again, we won’t know if HCD can make cameras better unless we try – but doesn’t it seem that they’re ripe for it?
And just because HCD wasn’t used doesn’t mean people will refuse to use cameras. Indeed, most of the consuming public never stops to think that maybe the products they use are horribly designed from the users’ perspective (e.g. Microsoft Windoze as opposed to Mac OSX) and that they deserve better. If they don’t know what’s possible, they could well continue to use un-HCD-based products, simply thinking that that’s as good as it gets.
ACD beats HCD
Having said all this, I’m still with Norman that activity-centred design is better than human-centred design, because in the long run ACD does more for humans than HCD. ACD is better because it assumes a systems point of view, and the best product is the one that best fits the overall circumstances, not just the circumstances of one entity or another in a given situation. I prefer to say it thus: ACD best balances the situation. By putting activities at the centre of things, no single entity from the situation has a privileged position.
This is important: to say that humans are more important that everything else in a situation is not correct. Put another way, focusing on any one aspect – such as the human user – will help one reach only a local optimum such that it may be impossible to reach the global optimum – which is pretty much the definition of a bad product. One might devise a product that is exquisitely tailored to the needs of its users, but it might cost so much no one can afford it. What good is the world’s best product if no one can buy it? Perhaps it isn’t sufficiently durable. Or perhaps it isn’t sustainable.
What’s the point of inventing a human-centred product if it cannot also be durable, and affordable, and sustainable, and….? And yet, in an HCD setting, given a choice of human-centredness and any other factor, the human-centricity must win, by definition.
Norman’s example of Southwest Airlines is a great example of designing as balance-finding (as well as ACD). The airline doesn’t provide reserved seating and inter-airline baggage transfer (which are commonly noted as customer complaints) yet remains a very successful airline. This is because, in the situation in which Southwest exists, other forces – cost and reliability – more than offset the forces of the complaints. By balancing the design of their service this way, Southwest remains successful (i.e. lots of people use it) in spite of the complaints. I think the customers know it too, else they wouldn’t continue to use the airline.
Southwest’s solution would be unacceptable in an HCD framework.
Norman then proposes that technology never adapts to humans, but we always adapt to technology. I have a problem with this because “technology” is only a concept. You can’t point to “technology;” you can only point to instances of it (products).
Products can’t adapt to users, yet, because they’re just objects. Nor has any product yet been devised that is economical, functional, and as adaptable as humans are. What does change, over time, are generations of product, as a result of design and engineering. That is, there are various forces that are pressuring humans to make products that are not particularly adaptable, because we do not know how to do it well/cheaply yet. (The obvious exception is some kinds of software that adapts to different users, Google being a classic instance.) So successive versions of a product can adapt to users.
Norman then gives some examples of how we have had to adapt to technology.
Clocks & watches. It’s interesting to think of the clock actually changing us, but we wouldn’t have allowed it if there hadn’t been an extant need. The clock is just the means by which we seek to become more (and, increasingly, too) efficient, or to impose efficiency on others. We are really prisoners of our own desire to get more and more done. The clock is just the lock on the prison door. Because of this, I’m not sure the clock is the best example of humans adapting to technology – unless I’ve misunderstood Norman substantively.
Writing systems. Norman suggests that even Palm’s Graffiti system – designed specifically with humans in mind – is an unnatural way of writing. I would ask instead for someone to name a single natural way of writing. As far as I can tell, there aren’t any. So maybe this isn’t a good example either.
Musical instruments. These items are left in their woefully unergonomic states because special privilege is afforded those forms for the “purity” of the undertaking of music. Musicians have to adapt to their instruments because they are generally thought of as already perfected. Here’s a wonderful instance, I think, of how the arts exist in rather stark denial of the physical universe. I lump instruments with cars – they remain as they are because there are cultural, political, and societal forces balancing the need for a more “human-centred” design.
Norman writes: “Show me an instance of a major technology that was developed according to principles of Human-Centered Design, or rapid prototype and test, or user modeling, or the technology adapting to the user. Note the word “major.” I have no doubt that many projects were improved, perhaps even dramatically, by the use of these techniques. But name one fundamental, major enhancement to our technologies that came about this way.”
Right. I can’t think of one either. But perhaps that’s not how it has happened in practise. That is, while I cannot think of a single major technology that was devised with an HCD perspective, I can think of lots of major technologies that benefitted from HCD after its introduction, in one way or another, just as Norman writes. Consider the Wright brothers’ airplane, Ford’s Model T, or ENIAC, as compared to an Airbus A380, a Smart ForTwo, and an iPhone.
So it seems that HCD doesn’t apply to new and highly (even disruptively) innovative products?
I can think of a couple of potential reasons for this.
- Highly disruptive innovations are a surprise, even to their developers. Maybe they didn’t think their innovations would be adopted so broadly so quickly. Maybe they thought they’d have time to revise their designs before wide-spread acceptance occurred.
- Developers don’t necessarily know how disruptive innovations will be used. If the innovation is not necessarily targetted at specific groups, then how can HCD be applied? Of course, the same can be said of ACD, though to a lesser extent, because the /activities/ at the heart of ACD will include human users without making them central.
There may be other reasons too that I haven’t yet thought of.
Is there something better than ACD? Probably…. Remember Rule 29. But whatever it is, it isn’t obvious to me. I’m pretty sure it will be systems-based – until we find a better way of managing complexity than the systems perspective. I’m also pretty sure it will be as quantitative as possible. It will have to be rooted in understanding the other, yet incorporate the expertise and experience of the one (the designers). I’m also pretty sure it will be collaborative and involve many people – the era of the lone, genius designer is waning.
Here’s one last thought. In the same way that ACD beats HCD by focusing on activities and thus putting no single entity in a privileged position, there are some things that ACD privileges – activities. Is that a good idea? Is there something that should be privileged instead of activities? Has anyone done a comparison of a variety of ways of privileging items? Is there a way to think of design that privileges nothing?
It seems that we will continue to live in interesting times.