Last week at Linja we were having discussion about intelligent UIs - how to make intelligent but not irritating user interfaces?
We got inspired about an article in Engadget: Intuitive's smartphone UI concept thinks so you don't have to.
__________________________________________________________
"It looks like you are writing a letter..."
Well, that's not completely fair. It seems to based on time, so I suppose this might work for people with very predictable life patterns. The key to this intelligent stuff is to make sure the prediction does not get in the way, or makes completing other actions more difficult. In this case, seems like apps are being pushed in front of other apps making it more difficult to do those actions, hence it's a pretty dangerous concept.
Think about the difference of intellimenus (the item-hiding menu) and the xp start menu's "most used apps"-menu. The first makes life really difficult, the latter doesn't. In fact, most people don't even realize the latter is an "intelligent" menu, it's that unobtrusive. Since it's as easy to access the "all programs"-menu as it has ever been, the latter doesn't get in the way or lengthen the process of launching apps even if the intelligence "fails" to recommend the right app. The former is about hiding stuff, the latter is about promotion -- it's a fine line but the effects are exponential.
Regards,
-- Martin G.
__________________________________________________________
What other examples could we find about a good intelligence in UIs?
- "Most used applications", "Most used bookmarks" idea, as Martin said.
- Switching web page language to the country where the browser is.
- The Wii Fit provides new experiences even after months of playing. For example, the repeating action of measuring your weight is done with your eyes closed.
- Facebook. They have changed the UI several times, going with the flow of how the concept of their service is evolving. It used to be more about branding yourself; now it's much more about discussing other's posts and links. Could be said that the service evolves according to its context.
- Why couldn't software record the ways I'm using it, and provide an alternative customized way of use based on that information?
- Instead of those irritating tips there are in image editing programs, could they measure my level of expertise and actually provide something useful for my habits, if I'd want to? For example, they could link to some interesting tutorials in the web, related to the type of work I'm doing.
- How could a company website evolve... One core thing is of course to keep the content up to date, and provide new information, and announce if there is something new happening. One other basic issue is to provide different versions of the site to different browsers, or mobile browsers, so that they can have an optimal experience. But could we actually change their experience to a good direction over time? Could the browser information be used for profiling the users to for example clients or developers, and display different focus for the two groups?
Cheers,
Vilja
__________________________________________________________
Letter-by-letter search results
My favourite adaptive UI pattern is letter-by-letter search results, especially for email and web addresses. Typing exact, complete addresses is a pain, but so is hunting and clicking through an address book or bookmarks menu (where you may need to run a search to find the item anyway). Instead, I’ve come to rely on URL/email address bars that offer reasonable suggestions from my bookmarks, address book, browsing history, sent/received emails etc. Also useful are search boxes that offer typical search keywords similar to the one you are currently typing (e.g. Google). For example, you type “currency conv” in the Firefox search bar, and up comes a bunch of popular searches for currency converters. Besides saving you typing, sometimes the suggestions include a clever way to phrase your query that wouldn’t have otherwise occurred to you.
The beauty is that you don’t need to actively do a search (or even know about the feature) to get the benefit. Your input into a normal text field is implicitly treated as a search query, the results are presented in context, and if you notice what you were looking for in the results, just hit the down arrow once or twice instead of finishing what you were originally typing.
As already pointed out, the key is to be contextual and readily available (discoverability) without being disruptive. The difference lies in careful interaction design, plain and simple. If those same search results would be presented using, say, a modal dialog box instead of ignorable dropdown lists, I’d be cursing them instead of praising.
- Lauri