Interface Hall of Fame/Shame
June 21, 2003 | Edward Tufte
31 Comment(s)
Bruce Tognazzini (along with a contributor, Joe Moran) has a brilliant and funny account of the interface design for an ultra-quiet dishwasher. “Ultra-quiet” turns out to matter. Lots of thoughtful advice about interface design in general. Posted at Asktog, which is always great reading on interface design.
Note particularly the point about the value of clutter compared to over-smooth elegant faux-simplicity (the BMW iDrive is mentioned). As Tog writes “Simple visual appearance doesn’t = simple interface.”
Topics: 3-Star Threads, E.T.
So, too, does “ultra-slow” without an indication that the machine (all too often, the computer I’m using at the moment) is actually doing something. Hence the hourglass, spinning rainbow beach ball, etc. I suspect “ultra-loud” would do the same thing if there’s nothing to indicate that noise is to be expected.
Perhaps the general rule is to indicate to the person at the other end of the interface when something is happening that may contradict expectations? Or even to set the expectations unambiguously? Or to just make things work how people would expect? The latter becomes difficult in times when we’re all so used to poor design that it becomes the norm instead of the exception.
I have a similar problem with my Sony Ericsson T39 Mobile. If you look at their web site and select “Try
the phone”, you can play with a flash emulation of the phone. You’ll find that it’s
really easy to get ahead of the emulator, and delete a message by accident. OK, but
that’s just flash being cranky, right?
Wrong. The emulator is as fast as a real phone – in fact the real phone
seems to have a “standby” mode when you haven’t touched it for a while, which
seems to add an extra 0.5 second to the pause before it responds to keypresses. This
makers it distressingly easy to delete the number or message you wanted by
accident.
Not that I’m giving up this phone just now – I couldn’t live without Bluetooth, even if
it does eat batteries. But lack of feedback does make usability suffer.
Here’s a start on interface design:
The Interface Hall of Fame (from Isys Information Architects via Wayback), and then link there to the Interface Hall of Shame:
http://web.archive.org/web/20021105045608/www.iarchitect.com/mfame.htm
Then 2 recent stories from The New York Times, about cell phones and about fancy
interfaces for very expensive cars:
http://www.nytimes.com/aponline/us/AP-Cell-Phone-Usability.html
http://www.nytimes.com/2006/05/28/automobiles/autoreviews/28AUTO.html
In extending this thread, let’s try to contribute more to the Hall of Fame than to the Hall of
Shame, since Shame Hall will be easy.
Also it will be helpful to read carefully over the links above and make comparisons among the interfaces described in those links and the new material in the contribution. It is easy enough to design a poor interface; the relevant issues are (1) what are the characteristics of a good interface (show actual examples) and (2) how can good interfaces be produced?
The graveyard spiral for airplanes and the pilot interface:
http://www.aero.ca/e_graveyard_spiral.html
and associated links.
And of course the enormously thoughtful Bruce “Tog” Tognazzini
http://www.asktog.com/menus/designMenu.html
http://www.asktog.com/index.html
Scott Zetlan’s musing from 2003,
strikes me as a nice companion to the “Principle of Least Surprise” I’ve seen mentioned by interface designers (I can’t remember who the originator is, or where I heard it).
The Principle of Least Surprise is that whatever happens after the question “I wonder what happens if I press this button?” should produce the reaction that is as far as possible from “Oh! I wasn’t expecting that to happen!”
Scott’s corollary is that even if what happens is what was expected, then the news should be delivered to the user immediately via at least one of the user’s natural senses, to avoid the user standing around fretting. Or, if what happens is not what was expected, walking away only to come back to a flooded kitchen, .
The fundamental flaw with several of the shameful examples is not presenting choices poorly but presenting too many choices to begin with. Barry Schwartz has written of the Tyranny of Choices, with research showing that the more choices people are offered, the less satisfied they tend to be with the choices they make.
Take the example of the Mercedes front seat adjustments, which have over a dozen buttons. This shifts the responsibility of designing a comfortable seat from the car manufacturer to the user, a non-expert in automotive seat design. While the user may be able to adjust the seats to find something immediately comfortable (difficult enough with 12 variables to control), he is not likely to know ahead of time whether that same setting will remain comfortable over the duration of a long drive. And if the user does feel uncomfortable in the seat, he hasn’t even the luxury of cursing the designer or the car salesman; he adjusted the seat, so he is responsible.
Customization is often desirable. But it should not be used as an excuse to foist fundamental design choices onto the end user. The designer’s job is to make good choices on our behalf.
Why do people talk loudly on mobile phones?
It is not because they are using them in noisy environments. It is because there is no feedback between the microphone and the earpiece. It is present in landline and wireless landline phones. For some reason virtually all mobile phone have this basic interface design item missing. (Can anyone give a reason?) People are so used to having this small part of their own voice fed back, that when it is missing they automatically think that the person at the other end cannot hear them as well.
What an interesting point, explaining why we all get to overhear those loud banal
conversations.
I look forward with horror to cellphones on airplanes (apparently coming soon), although I hope to be protected by my iPod+noise-cancelling earphones. Perhaps Bose will now eliminate the gap in their sound cancellation that allows some voice frequencies still to come through (based on a theory-fantasy about hearing a voice in an emergency, a theory perhaps concocted by the legal department).
My hearing aid is fitted with snug earpieces so that the combination of microphone and speaker in the same package will not cause feedback howl, and we are advised to insert the hearing aid and then switch it on, rather than vice versa.
I suspect that the poor or absent user voice feedback in miniature mobile phones is because the phones are so compact compared to landline handsets that the designers feared a similar positive feedback loop would cause whistling in the ear.
The mobile phone vendors’ paramount concern is to avoid irritating the buyers of their product so much that they will advise others not to buy that model. They don’t worry so much if the buyer irritates complete strangers as a result of their design decisions.
Cat user-testing MacBook Pro:
https://www.vimeo.com/clip:77426
A remarkable account of trying to cancel AOL by Randall Stross at The New YorkTimes:
http://www.nytimes.com/2006/07/02/business/yourmoney/02digi.html?pagewanted=all
Demos of the splendid iPhone interface.
The gray circle moving on the iPhone screen is your fingertip. At the Apple web
page, which has somewhat mysterious navigation, animations in each category are at top
right.
The main iPhone
design issue is keyboard usability, although users have sometimes accommodated
themselves to small and difficult keyboards over the years. At least email messages will be
shorter with iPhone.
After
moving through the demo, I found it a drag to go back to typing HTML commands.
ET
conflict of interest statement: The demo includes, by good luck as far as I know, a cameo of my
Beautiful
Evidence cover.
Would it improve the usability of the iPhone’s e-mail client to use voice recognition software like Dragon Naturally Speaking in conjunction with the speakerphone to dictate e-mail? NPR reported DNS 9 is ready to use out-of-the-box.
My oven, by GE, has a delightful dial control on the front panel, which is used to set the oven temperature, clock, and timer. It couldn’t be simpler or more intuitive. Push the button for the function you wish to set (labeled “Bake” “Clock” “Timer” and one more I can’t remember because I never use it), then spin the dial. Clockwise moves the numbers up, counter-clockwise takes them in the other direction. When setting the timer, turning the dial slowly moves upward in 5 second increments. Once you get over a minute, it switches to 10 second increments. Shortly after that, it switches to minute intervals. Once you stop turning it, you can then go forward or backward at 5 second intervals again. The clock works very similarly. It is an absolute dream to use, the best interface for that purpose I’ve ever encountered beyond a full-fledged number pad.
Outstanding display of lots of data:
http://www.gapminder.org/
And a lecture using the gapminder
http://www.ted.com/tedtalks/tedtalksplayer.cfm?key=hans_rosling
another incredible gapminder video: http://video.google.com/url?docid=7996617766640098677&esrc=gvpl&ev=v&q=google+tech+talk+gapminder&vidurl=http://video.google.com/videoplay%3Fdocid%3D7996617766640098677%26q%3Dgoogle%2Btech%2Btalk%2Bgapminder&usg=AL29H23eOJh6H8aTYmK39vH6nA5E3JSxiA
Abstract
Science Fiction movies have been a source for speculation about the future of technology
and human computer interaction. This paper presents a survey of different kinds of
interaction designs in movies during the past decades and relates the techniques of the
films to existing technologies and prototypes where possible. The interactions will be
categorized with respect to their domain of real-life applications and also evaluated in
regard to results of current research in human computer interaction.
http://w5.cs.uni-sb.de/~butz/teaching/ie-ss03/papers/HCIinSF/
An interesting and elegant multimedia interface involving some data-graphical
intelligence:
http://www.portfolio.com/news-markets/multimedia/
Parts of the design combine the news graphics style of Spy (readers of a certain age will remember Spy Magazine) and the New York Times.
Alan Cooper and I recently traded copies of our new books. His new About Face 3 (with
Robert Reimann and Dave Cronin) is better than ever. Alan has all sorts of profoundly wise
things to say about interface design:
“No matter how cool you interface is, less of it would be better.”
“Define what the product will do before you design how the product will do it.”
“Imagine users as very intelligent but very busy.”
“An error may not be your fault, but it’s your responsibility.”
“Contextualize information.”
These are a few of the general design principles; much of the book is right down in the
intensely detailed practicalities of interaction design.
Professor Hans Rosling ( Gapminder ) is bringing data exploration to
the masses in short segments akin to an evening weather report.
Health, Money & Sex in Sweden – GapCast #1 with Hans Rosling
After some incredible videos of dense content and fast moving
figures, it seems that the “gapcast” may have been dumbed down
the general public.
Hopefully we will get to see the real Professor Rosling
in Gapcast #2 without the Monty Python effects – or at least
if they have to be there for the masses, then perhaps we will
get more of a flavour for his rigorous approach to data and
subtle attention to detail.
NOTE – There are links on this thread to some of Professor Rosling’s
other presentations.
Macnify, a widget in Apple’s dashboard, enables zooming into any element on the computer
screen. Zooming goes all the way down to pixel-resolution. I believe this feature is also now
built into the most recent operating systems for Mac and PC.
There are so many dreadful interfaces out there – it’s a nice surprise to discover one that is exceptionally easy to use.
The online residential real estate brokerage company Redfin (www.redfin.com) has received some press coverage lately (if 60 Minutes still counts) for introducing competition to the traditional realtor-centered real estate business. After they expanded to the Washington DC market this month, I decided to check them out and was impressed by the interface of their website for searching properties.
Anyone who has ever used traditional real estate sites has endured the pain of searching for properties with klunky forms and limited options for displaying results.
Although not perfect, the Redfin interface is a welcome improvement. It is evident they considered how prospective home buyers use the internet to scout properties rather than designing an interface based around off the shelf web application software. After a few days of using it, it dawned on me that perhaps traditional brokers don’t want customers to use the website (so why make it easier?) and instead want you to rely on an agent to justify the commission-based business model. So maybe the better interface design on Redfin isn’t so much about striving for excellence in visual display of information as it is about supporting their distinctive electronic business model. Take a look and see what you think…
Features I liked:
Physically select the geographic area of interest instead of using zip codes.
See location patterns of desired property types by zooming in or out.
Filter out extraneous information from the map with one click (street view, satellite view, etc)
Subtle color changes in the house icon to let you know if you’ve viewed it before.
Graphed property value history (would be nice to see a sparkline application here with high/low and most recent value point and a “normal band” for the values in the area).
List pricing history (not just sale price history) for each property. (Just how desperate are they?)
One click actions that let you export the data into a variety of formats.
Very subtle table grid borders and row markers that emphasize the data and not the grid.
If only the site would automatically remove all real estate euphemisms from listings such as : charming, cozy, well-maintained, newly updated, high-potential, etc.
Happy House Hunting – DMA
Professor Mark Liberman gives a hilarious example of bad software interface design here. For further explanation, see his entry on the Language Log. From links that don’t do what they say they do, to forms that you aren’t supposed to fill out, Facility Focus seems to have it all.
Hans Rosling gave another presentation at TED 2007. Some of it is similar to his 2006 talk, but the visual impact of the statistics is just as strong. His unforgettable finale of “The seemingly impossible is possible” should not be missed.
He also uses one PowerPoint slide in this presentation, which he pokes fun at with the comment “But I have to get serious. And how do you get serious? You make a PowerPoint — you make bullets.”
http://www.ted.com/index.php/talks/view/id/140
In another presentation at the OECD forum he showed a slide of “beautiful music”, a page of sheet music to which he commented:
“Few people will appreciate the music if I just show them the notes. Most of us need to listen to the music to understand how beautiful it is. But often that’s how we present statistics. We just show the notes we don’t play the music.”
Beautiful Evidence, indeed!
I thought I would add some commentary to some really interesting things people have contributed to this thread.
Patrick Martin said: Customization is often desirable. But it should not be used as an excuse to foist fundamental design choices onto the end user. The designer’s job is to make good choices on our behalf.
I couldn’t agree more. I used to work for a portal company (in fact it used to be the #1 portal company, but its decline is a story unto itself… thankfully, it re-found itself and went back to its strength and still remains in business). Customers of this company wanted endless upon endless flexibility for their end customers that ultimately the consultant part of the company that I worked for helped develop or sometimes outright developed. Endless customization seemed to signify to me that they didn’t know exactly what they should present that would be of benefit. Some of these were intranet portal sites and they wanted the option to show the weather or external news sources of the staff member’s choice, or many other non-business relevant item. I could not understand this. They were spending sometimes hundreds of thousands of dollars and they didn’t want it to work towards the bottom line?
Jose Silva said: Science Fiction movies have been a source for speculation about the future of technology and human computer interaction.
Not that I disagree about the movies (or TV for that matter – witness Star Trek!), but I think books are even more accurate predictors. Read The Foundation Trilogy and you will get a very quick sense. Another great author is Larry Niven. His tasp is exactly targeting how addiction works and yet most doctors had not subscribed to this theory at all when he wrote his book. How insightful!
Edward Tufte cited come principles in Alan Cooper’s book:
“No matter how cool you interface is, less of it would be better.”
“Define what the product will do before you design how the product will do it.”
“Imagine users as very intelligent but very busy.”
“An error may not be your fault, but it’s your responsibility.”
“Contextualize information.”
I’ll start with the second phrase and this will lead to a question for ET when I revisit that point. In software development, particularly where it is done iteratively, we concentrate on describing the functions needed by the application software; called use-cases in the approach we use, which is the Unified Modeling Language (UML). This is entirely independent of the user interface(UI) design. This is not to say UI design is unimportant, but knowing what you are trying to accomplish independent of any predefined UI allows open thinking on the approach – it defines teh domain so to speak. So I would like to follow up with some questions/comments on software design that I think are important as examples to the points Cooper is making. I can only use the points as presented by ET as I have not read his book:
Less cool interface: I would be interested in understanding ETs and other’s thoughts on things such as “wizards” where small questions and inputs lead to an overall presentation of data or entry of data into a system. What are the strengths and weaknesses of such an approach?
Definition: We use UML extensively and I personally find it a very dense format for presenting a lot of data that a software developer needs. I am curious what other’s thoughts are on this? I know in the software world it is sort of king right now, but is it really the best?
Really busy users: This means varying things based on the systems with which you are working. I know ETs work mostly concentrates with data presentation, but what if the data is centered around entry? I would say a very clean minimalistic interface with only feedback on errors during entry you need to correct are preferable to overly complex displays that show the effects of the data entered, when it is not truly relevant to the data entry itself. Focus on data accuracy and validation, not on presentation when it is data entry.
Errors: In software development, the true errors of entering the wrong type of data (characters when integers were only to be allowed), data outside the acceptable range of values, and the like are the responsibility of the developer to catch (or at least should be). You should explore these in your requirements and the resulting tests.
Context: Finally, when dealing with content in web applications and/or web sites, while content is the most important, context is second. How relevant would budget data on a Government website be if not shown against the program it was supporting? What would be the rationale for even showing keyword tags (tag clouds) if they were not considered relevant? Relevance is the second most important item after the content itself. Some would even argue more so, although I would argue that if the content were not in context, at least you could determine that by reading it if the content itself was substantial enough.
uh oh:
http://www.componentart.com/charting/gallery/
I think the component art designers got carried away.
You know…..I’ve read through this thread a few times now and it seems that a single topic has spun out into three distinct topics ( or perhaps more – my mind quit counting @ 3). On the Original topic, Ed made mention to an article @ AskTog that was (I believe) intended to emphasis the simple (yet more often than not) over-looked principle in Interface Design: Start Simple and broaden out from there ONLY as absolutely required.
There have been few instances in consumer products that have really taken this approach; rather most things have to have all the “Bells & Whistles” from market introduction/product inception. Why? Have they sold better in the marketplace? Have they become prime/classic examples of what/how to do it? It seems to me (albeit I’m a admitted neophyte on these subject matters) that this approach is taken as a means for folks to justify their jobs/positions…or the easy way out. I don’t get it….help me understand?
At the risk of waking a long-dead thread, there’s this post from the “Coding Horror” blog (a good read for
programmers of human interface applications), which points out that there’s a good complement to Fitts’s Law of
making items in the interface easy to click. It is, simply, that there are some interface components that really should
be hard to get to.
The Opposite of Fitts’s
Law
Moki.tv has put together an infographic
blog post of top 20 movie ratings, which, from a design standpoint, approaches the New York Times infographics.
Unfortunately, they slightly undermine their own position by overreaching from “top 20 movies” to the more general
“movies”. It would be at least as interesting to know if independent cinema was gaining ground.
As it stands, it appears Michael Bay has taken a real leadership position.
One of the best interface designs I’ve yet seen was a bit more than ten years ago, when I was working in foodservice, in
the form of a commercial microwave oven. I can’t remember the brand, alas, but I bring it up because it serves as an
example of what works well in design.
The oven had but one button, a mechanical one, that opened the door. The only other control was a dial that set the
time. Just turn the dial to the desired time and it would start. The time on the dial was on a logarithmic scale, because
the difference between 30 and 60 seconds matters more than the difference between 14 and 15 minutes.
The beauty of it was the utter simplicity. No series of buttons to press. No hitting “time” and then setting the time.
Much could be learned from this, I think.