Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!uniwa!DIALix!metapro!bernie From: bernie@metapro.DIALix.oz.au (Bernd Felsche) Newsgroups: comp.human-factors Subject: Re: adaptive user interfaces Message-ID: <1991Jun29.064425.1909@metapro.DIALix.oz.au> Date: 29 Jun 91 06:44:25 GMT References: <4488@jethro.Corp.Sun.COM> Organization: MetaPro Systems, Perth, Western Australia Lines: 69 In <4488@jethro.Corp.Sun.COM> toone@looney.Corp.Sun.COM (Nolan Toone) writes: >Alot has been said concerning adaptive vs. adaptable interfaces. I think >my approch would be to use the adaptive approch to set the default action >and the user can choose it with little or no effort but ALLOW the user to override >it at will, but he user should also be allowed to set the defaults as well. >That my humble opinion. The trick (IMHO) with adaptive interfaces is to figue out when and how to adapt. Taking a leaf out of nature's pocket book, it seems like the time to change the interface's behaviour is on startup, not while the application is running. Having said that, I have to back off a little and consider what happens with long-running applications, and of course the nature of the application. It is probably most natural that applications adapt at a very slow rate, and never by more than one iteration (however that may be defined) in each function. Some of the things which come to mind in adaptive applications are the presentation of error and status messages. For example, initially a user may be required to take some action (press a key or click on a gadget) before the application continues. Later, that may be down-graded for the less-important conditions, to a time-delay, of gradually reducing magnitude. When the user exits, interface adaptation parameters should be saved for later re-use. On the next startup, it is intuitively appropriate to downgrade the adaptation a little, and allow the user to come up to speed again. Unlike machine tools, where adaptive control is relatively easy because the controller knows the desired result, the adaptive user interface has absolutely no idea of the target. In fact, the interface is not the controller, it's the user. The interface is a tool used to perform a function, which is probably another clue as to how adaptive interfaces could adapt. If the tool adapts to the work at hand, then the user will adapt to the tool, seeking the easiest path to a solution, as long as it's not obfuscated. The tool must therefore adapt to the type of work and how the user perceives the interface. Using this, the interface can be modified to present the most likely functions in a way which the user can easily perceive and use them. In terms of control systems, the situation is most interesting, with two adaptive systems (hopefully) in synergy. I'd rather not sit down and work out the mathematics, thank you. In situations like this, I prefer to stand back and use some intuition to come up with a usable "machine". It could also be likened to two people building a brick wall. The first knows what he wants to do, but only by starting to excavate for the foundations, does the other know to dig as well. Then when the brick laying starts, how does the second know to cart the bricks or even lay them, or when and how much cement to mix? If the second is over-enthusiatic is some respect he could mix too much cement or building the wall too high. Incessant nagging and "stupid" questions may lead to a very unfriendly interface. -- Bernd Felsche, _--_|\ #include Metapro Systems, / sold \ Fax: +61 9 472 3337 328 Albany Highway, \_.--._/ Phone: +61 9 362 9355 Victoria Park, Western Australia v Email: bernie@metapro.DIALix.oz.au