This paper describes some observations of compensatory actions
taken by users when response time on the University network and the
World Wide Web were severely limited. It shows how different users
tackle the same problem in very different ways. The more technically
able users look for a technical solution to the problem by 'tweaking'
the hardware and looking for help in the more 'esoteric' areas of the
browsers available to them. The less technical (some would say more
practical) users apply a more pragmatic approach, looking for ways to
by-pass the problem. (saving the pages when they perceived the
response time to be reasonable) or looking for ways to speed up their
access to information which is relevant to them and their computer
configuration (i.e. ignoring sites displaying lots of graphics and
using frames and adhering to mainly text limited sites). The
observations suggest that a full experiment would be a useful
follow-up to the initial findings cited below.
The origins of this paper stem from the outcomes of two assignments set around the Web for two totally different groups of students. The first assignment was set for the second year Higher National Diploma (HND) students which required them to consider the Netscape browser from the viewpoints of both Novice and Expert users, writing a critical appraisal of their findings. The second assignment was set for a group of final year degree students (BSc Computing, BSc Software Engineering and BSc Combined Honours) requiring them to design Web pages for a small but flourishing company. The underlying aspect of both of these assessments was that of communication and usability with respect to the audience and the equipment being used. However at the time of these assignments first being tackled by the students, the University was having severe problems with the speed of access to the Web. Consequently the users were forced to compensate for the limitations of the system. It is interesting to note that even after the speed of access to the Web was improved, many of the users still took compensatory actions 'just in case' the old problems recurred.
The techniques used to compensate varied from making better use of the functionality provided by the browser to visiting those sites on the web that used simple media types (e.g. Jakob Nielsen's www.useit.com/alertbox/subscribe.html) and/or provided information more 'personal' to the user (e.g. News sites subscriptions such as - listserv@dispatch.cnet.com).
These resultant compensatory actions can be sub-divided into two classifications. The first classification is where the users generally have the knowledge or the location of the page but are trying to get there as efficiently as possible. This is frequently combined with knowledge of the hardware platform and/or the browser being used. We do know that the users' level of experience has a large effect on how they use the system (Nielsen, 1995) and that their ability and confidence level increases with further use. We are also aware that as the user builds up knowledge of the system and that systems is upgraded, the user expects to be able to transfer the knowledge gained from the earlier version of the system to that of the later one (Tognazzini, 1992).
In the second classification, the users do not know what
information they are seeking, but are intent on minimising the time
wasted in finding it. According to Landauer (1995) 'In striving for
work productivity, we fundamentally want to reduce the amount of time
that a user needs to spend to accomplish a given amount of work'.
(a) Multi-threading was used to enable the user to initiate multiple sessions thereby following more than one thread. This caused the users some problems since if the access speed was slow this wasn't improved by multi-threading. When the sessions were in vaguely similar areas the users became even more disorientated since often they weren't sure which results from which session they were currently examining.
(b) The users would download the pages of interest and save them away to view them later on a browser not attached to the Web. Users often found that they were unable to see the graphics, since only the text had been saved to disc. The method for saving the images to disc was by via the use of the right mouse button: the users were not familiar with using the right mouse button since on the system they were used to, most operations were carried out with the only the left mouse button.
(c) Some users expanded the cache memory to help with the assimilation of information, thus allowing the user to quickly access pages viewed earlier in the session. Users are generally only able to remember 7+/-2 (Miller's number) chunks of information in their short term memory at any one time. Increasing the cache memory and disc space allows the user to navigate backwards and forwards without revisiting the Web sites.
(d) Many users chose to deactivate automatic image loading. This causes a problem to the users when the author has not included text alternatives. Most browsers now perform partial image rendering, which at least means that the users can decide whether to wait for the full image or whether sufficient of the image has been downloaded for them to make a decision on where to 'jump' to next. Thumbnail images help in a similar fashion; when linked to full size images, they reduce the potential download time.
(e) Many users jumped to another link even though they were not
sure that it was a relevant link, but they felt panicked into
reacting before they 'lost the speed'. This was found to lead to an
unsystematic approach to Web navigation, cited in 'Hypertext and
Hypermedia, Theory and Applications' (Woodhead, 1991) as
'wandering'.
(f) Some users find a site or author who has some credibility (the author being known in academia for having produced relevant papers through the normal peer-review system) following their links, and maintaining contact with that site (e.g. subscribing to Jakob Nielsen's alertbox). However, even though the users may have faith in the author and their links, the fact that the associations are dependent on the author's view and their interpretation, means that the users' understanding does not necessarily tie in with that of the author. Webcosm, a product available from Southampton University, attempts to compensate for this by allowing the user to make their own links rather than (and/or including) those of the author.
(g) Some users chose to avoid sites with large amounts of graphics and/or other media types together with those that made heavy use of frames. Even though we know that from a usability point of view, frames are good; from a navigation point of view they cause much confusion. Users found that the use of the right hand mouse key lacked their expectation of consistency, i.e. they voiced the expectation that 'back' would take them to the previous frame. They also found that, during the download of frames, the feedback message at the bottom of the screen was very confusing: as they moved the mouse around the screen, they could not understand why the message changed from an indication that the download was complete to an indication that it was less that 100% complete!
(h) The use of a search engine was seen as an essential means of avoiding time-wasting through browsing, although many users declared their need to be familiar with the search engine to know how best to use it - and that learning takes time and effort. Novice users could not understand why they got a different search engine each time they used Netsearch; Expert users had their own favourite search engine which they used, and so could not understand the Novice users' problem.
(i) The use of 'me-mail', personalised information feedback and the use of agents were seen as potential future time savers. But even these show evidence of problems: intelligent agents are only able to monitor small amounts of what the user learns in the course of a day, thereby limiting the knowledge of the user which they are able to build up.
(j) Some users returned to a heavier use of FTP wherever possible.
This was certainly the case for those users who were familiar with
FTP and particularly with that site.
The above selection of points is based on observations of users'
compensations over time, which arose as a result of bandwidth
limitations. It would have been useful if we had been able to change
the criteria for the students' assessment allowing us to follow up
the initial observations, but that would have been educationally and
professionally unsound. Since this was not a professionally conducted
experiment, the next step is the organisation of an experiment using
a range of users (students are ideal in this situation as they cover
18 upwards with varying ability levels) to validate the observational
findings and to see if there is a logical reason for the users'
attendance to one of the two classifications noted above.
T. K. Landauer (1995)
The Trouble with Computers, Massachusetts Institute of Technology.
J. Nielsen (1995)
Multimedia and Hypertext, The Internet and Beyond, Academic Press
Inc.
B. Tognazzini (1992)
Tog on Interface, Addison Wesley Publishing Company.
N. Woodhead (1991)
Hypertext and Hypermedia, Theory and Applications, Addison Wesley
Publishing Company.