Alan Dix - research topics


In 1987 I wrote a paper called "the myth of the infinitely fast machine". The central thesis was that designers of computer systems assume the machines are "fast enough" - effectively infinitely fast. Because of this various strange behaviours could be observed. Although computers are now an order of magnitude faster, the problems still remain. One reason for this is that functionality increases as fast as processing power. Another cause is the increasing reliance on networks. These induce delays and hence timing problems. Once timing problems have been identified they are often not too difficult to put right - there are various generic solutions which can be applied. The biggest problem is for designers to admit they have a problem!

Remote collaborative work brings its own temporal issues. In such situations bandwidth is often not a problem. For example, it is easy to transfer multi-megabyte files over the Internet and even posting floppy disks allows a large data transfer rate! Instead of bandwidth the limiting factor in these circumstances is pace - the rate at which people (and systems) interact. For the Internet, the pace of communication is of the order of seconds - too slow for rapid interaction, For mobile-workers, the pace of communication with central office may be measured in days. When the pace of interaction is very slow (whether because of network delays or because the application has a long duration) we can no longer regard the feedback from the application as being immediate. The Norman execution-evaluation loop is broken and it is necessary to consider general design strategies for remedying the resulting problems.

Work with Devina Ramduny and Julie Wilkinson has developed ways of analysing long-term interaction in terms of triggers, the events which initiate actions. Typically the trigger of an action is not the completion of the previous action in a process, but occurs sometime after, often due to environmental cues such as a pile of papers on the desk. In addition, our case studies revealed a pattern of behaviour which we have called the 4Rs - request, recept, response and release, which appears to recur in may long-term collaborative interactions. This strand of work has important implications for the design of CSCW systems, workflow systems and BPR.

See also my topics pages on web architectures, which considers software architectures for collaborative applications on the Web; status-event analysis, which can be used to discover potential delays during early design and understand their effects; and the ecology of information, which looks at various ecological paradigms in current HCVI and computing and in particular issues of artefact centred design and environmental cues.

Other places to look

Time Design Workshop at CHI'2004 (ACM DOI)
DIRC Timeliness Theme
Wikipedia Human timescales
M. Kutar, C. Britton and C. Nehaniv. Specifiying multiple time granularities in interactive systems. Palanque and Paternó (eds), DSV-IS 2000 Interactive Systems: Design, Specification and Verification. LNCS 1946, Springer 2001, pp. 169­190.
This paper is extended in Kutar's PhD thesis (University of Hertfordshire, 2001) and deals with formalisation of issues such as the fact that "I'll do it today" could mean "by 5 c'clock", "by midnight", "within 24 hours" etc.
Maria is now at University of Salford: Maria's home page and working with Anna Cox at UCLIC on Time Design
Richard Thomas, Long Term Human-Computer Interaction - an exploratory perspective. Springer, 1998.
This describes a long term study of over seven years of use patterns of a single application (an editor).
Day workshop on Time and the Web held was at Stafford on 16th June 1997 The full proceedings are available online and also a report will appear in the January 1998 issue of SIGCHI Bulletin
In June 1995 a workshop was held at Glasgow in connection with the Temporal Aspects of Usability project.
The workshop report was been published in SIGCHI Bullitin, 28,(2).
After the workshop a mailing list was established called TICHI for people interested in this area.
The special issue of Interacting with Computers (vol, 11, no. 1, 1999) on Temporal Aspects of Usability, edited by Steve Howard and John Fabre of Swinburne University of Technology, Australia.
Helen Parker, Chris Roast and Jawed Siddiqi. HCI and Requirements Engineering - Towards a Framework for Investigating Temporal Properties in Interaction SIGCHI Bulletin, Vol.29 No.1, January 1997
Jim Dabrowski's Masters thesis work on thresholds of detection for changes in a graphical user interface
Danny Chow's PhD work including a CHI2001 short paper on The effects of time delay in electronic commerce
Jacob Nielsen's Response Times: The Three Important Limits extract from "Usability Engineering"
On Bob Spence's home page he describes two experiments on delays and interruptions.
In the first [Spence 1981], experiments on delays in an interactive visualisation tool 'Minnie' found that, counter to expectations, delays less than 1 second and variability of delay had no serious effect on performance
In the second[Field and Spence 1994], experiments in a menu based information system found that interruptions increased increased search time by 46% and the number of frames accessed by 25%.

Some of my own online papers concerned with time

maintained by Alan Dix