views:

248

answers:

4

I'm currently doing research on adaptive interfaces and i'd like to hear some responses from fellow Web Developers on this subject.

We used to have static interfaces back in the day, where you couldn't change a thing and you had to learn the system the way it was build. After that came adaptable interfaces, where you could change some preferences to the way you wanted. Think of font-size, colors and layout. These are the interfaces we see mostly today. The next step in this area of research and development on interfaces is supposed to be adaptive interfaces.

An adaptive interface is an interactive software system that improves its ability to interact with a user based on partial experience with that user.

Example adaptive interface (linkedIn):

alt text

You see that the interface is making suggestions about your user profile. It tells you to complete some forms of your profile page in order to be found easier by fellow users.

An adaptive interface is the next logical step after adaptable interfaces, and research in this area has been going on for more then 20 years. However, these days i don't see alot of adaptive interfaces coming back in web applications.

This leads me to a few questions:

  • Why do you think that adaptive interfaces are still rare while the need for user personalisation in interfaces keeps growing?

  • Which problems do you see in the concept of adaptive interfaces?

  • Do you know any good examples of adaptive interfaces somewhere online that i haven't seen yet?

Thanks in advance.

+7  A: 

Psychological studies have shown that having excessive options makes people sad. I have certainly experienced this standing in the grocery store confronted with twenty slightly-different jars of pickles. I just want a dang jar of pickles. I don't need to become a pickle expert obsessing over the nuances of the various options.

User interface studies have shown that consistency is among the most important factors in a graphical user interface design. Older versions of Microsoft Office had a feature which hid unused menu items. Nearly everyone hated this feature. It was disorienting to have an interface that changed over time. It was even more confusing to go from one machine to another. New versions of Microsoft Office use the same ribbon GUI elements in the same order across all machines and for all users. The customization options are very limited. And as a consequence, it is much easier to use. (There was a lot of initial resistance to the ribbon design, but that also proves the point. It wasn't what people were used to.)

I don't like the thought of an adaptable interface. If I'm trying to post a message or check my bank balance or get movie ratings, I don't want to be fiddling around with colors and fonts. Leave that to the desktop publishing software.

And the thought of an adaptive interface, which morphs itself automatically in unexpected ways, sounds to me like Internet hell.

I don't think there is a need for adaptive interfaces. I think there's a need for high-quality GUI design right out of the box.

Jeffrey L Whitledge
You tend to think about changes in layout when talking of adaptive interfaces. This isn't necessarily true. Another example is collaborative filtering, a method to compare the interests and goals of different user with eachother. For example when you're on a specific dvd page from a webshop (e.g. 'Troy'), you get to see other dvd's that people who bought that dvd also bought (e.g. '300'). Like this, you get to share interest with fellow users and you can reach out for more content.
Paulo
@Paulo: That's more content-based, much like adaptive spam filtering. It isn't interface-based. I know how to see what my spam filters catch, and I know where to find Amazon's recommendations for me, and neither of those change unpredictably on me. If Thunderbird or Amazon decided I'd like an interface change based on my mail or what I order, I'd be really annoyed.
David Thornley
I agree its content-based, however this is still a part of adaptive interfaces. The interface adapts to the user's behavior and preferences. This is the core of adaptive interfaces. Their possibilities are not limited to the layout of a webpage like you and Jeffrey point out.
Paulo
+1  A: 

To build on Jeffrey's answer, another example of a changing interface causing problems is website redesigns. I have been involved in a number of redesign efforts. The first few days/weeks after a launch we will get feedback from people who can no longer find certain links and feel lost/disoriented on the site. This occurs despite our efforts to improve the layout and navigation of the website in question, and make it easier to get stuff done on the site.

Swoop
+2  A: 

The problem with the UI changing as it learns about the user is that the user is also changing as s/he learns about the UI. This makes usable adaptive UIs difficult to achieve because they result in inconsistency and user loss of control, violating two key principles of usable interfaces.

Adaptive UIs tend to work best as an unobtrusive supplement to a static UI. For example, split menus work well. Here, the system places the most frequent or recent menu items at the top or side of a long menu while the long menu still includes all items (examples include MS Windows XP’s Start menu or the font dropdown in MS Word). The key is that user can use the adaptive feature at any time, and can also ignore it at any time without any cost on using the static UI (e.g., no need to click to dismiss the adaptive feature). In contrast, it doesn’t work well to start moving or removing menu items in a menu (like MS Office’s failed Adaptive Menus).

Adaptive UIs also tend to work better at “lower levels” of automation. If something is in the UI in the first place, then it’s probably something the user cares about and therefore wants control over. It’s often helpful to adaptively suggest something (like Amazon suggesting you buy the latest Doonesbury book, when you’ve bought every other Doonesbury book), but often unacceptable to go ahead and do it without even giving the user an opt out (e.g., for Amazon to order and send the book to you).

Done correctly, adaptive UIs can make for a better experience in some situations. However, they are not appropriate for the “mark their territory” personalization that you’re interested in. The desire in such personalization is personal expression. When the system does it for the user, then it’s no longer personal. That is a case where users are unhappy if they don't have options.

For more about adaptable personalization, see

http://stackoverflow.com/questions/1814679/are-customizable-settings-important-in-web-apps

Michael Zuschlag
There are people who think that adaptive menus like you mention are a real pain. In that intermediate state between when you have to search each menu and when you just use the keyboard shortcuts (which differs a lot between individuals) it disrupts the mental model and muscle memory of the menus.
David Thornley
I’ve also seen users lose menu items in an adaptive menu. They remember seeing it on Menu A (right) but now it’s hidden, so they assume it must be Menu B (wrong). Split menus avoid both problems as long as there’s a constant number of items above the split. Each item remains in a constant position (below the split), so muscle memory works and you can always find the item.
Michael Zuschlag
+1  A: 

In the course of things, other people will frequently get glimpses of the interface you're using, and that has privacy implications. For example, if I were to hook my laptop up to a projector and started to type in a URL, the auto-completion suggestions might well be unprofessional. (In fact, I'd use a professional profile to avoid this.) Similarly, a person who bought tons of those bikini iPhone apps that Apple just pulled the plug on might not want the phone to suddenly switch the desktop image to something the iPhone noticed the user liking.

This is in addition to the "I just learned to do this/just got comfortable with this and the stupid machine changed it on me" effect.

David Thornley