Browsing for trout: Experiments on purpose & context


Preliminary Data & Discussion (Final poster will have results of more trials)

original image


Browsing has been recognized as important to scholars at least since the INTREX Conference of the 1960s. Rice, McCreadie, and Chang assert it is among “fundamental and pervasive human and social activities.” Browsing represents a significant or complete shift of the locus of representation from some external agency (e.g. cataloger) to the information seeker. Browsing is not a matter of dumb luck; it is, in varying ways and degrees, purposeful. Talking with scholars, submarine chasers, bounty hunters, and others for whom known-item searching is not an option, has done much to illuminate and establish the validity of the set of activities generally termed browsing.

In 1982, I was asked in one of my doctoral qualifying examination questions if browsing was significant for scholars and, if it is, could I design a computer program that would be of assistance to a scholarly browser. I cited substantial sources arguing for the importance of browsing to scholars, journalists, artists, and the like. I suggested a program that would make random insertions into a document collection, arguing that, except for the documents already known by the browser or the documents already known to be of no use (e.g., written in a language not read by the browser), all documents had an equal probability of being useful. In keeping with the early definition of browsing as the cropping of scant vegetation, there was an assumption that there might be a high ratio of effort to payoff. M.E. Maron accepted the response to the qualifying examination question, but wondered if there wasn't something more to be said about browsing.

Experiments have been difficult to design because of the very nature of the activities and the problems they typically address. We wished to devise some experiments that would stimulate thinking about more robust experiments. In order to focus on behavior rather than any specific environment, we devised some experiments that took us outdoors, away from books and computers. On the face of it, the work presented here might seem downright silly. However, we reasoned that if we performed two experiments at the outer bounds of reason, we might be able to begin to narrow in on useful and informative studies of browsing behavior. Two concepts guided our thinking: browsing is purposeful immersion in a state between stasis and chaos; browsing cannot act upon nothing, an environment empty of any likely targets.

Boundary Case Hypotheses

Fly-casting for trout on a golf course will yield at least one fish per four sessions. Looking for golf balls on a golf course will yield at least one ball per four sessions.

Experiment Design

On a golf course in Texas we conducted eight data gathering sessions aimed at elucidating the roles of purpose and context in browsing. Four of the sessions involved browsing for golf balls and four of the sessions involved browsing for trout.

Each of the golf balls sessions involved following a path through the golf course and picking up any golf ball that lay within one fathom of the path. Each of the golf ball sessions was conducted on a stormy day when no golfers were on the course and none had been on the course for at least one hour. The golf balls that were picked up were assumed to have been put into their positions of discovery by the actions of golfers and to have been abandoned by their owners. We cannot rule about the possibility that some of the balls had been moved by dogs, children, or raccoons.

Each of the trout sessions consisted of casting by a capable fly fisher. Trout catches in Montana and Argentina before and after the Texas sessions demonstrated the existing and continuing ability of both the equipment and the fly fisher. All fishing was catch and release. Each of the golf course trout sessions consisted of 15 to 30 minutes of casting, with approximately 50% of each session spent casting in one direction and the other 50% spent casting 180° away from the original direction.


The golf ball sessions returned 14, 8, 12, and 7 golf balls. Of the total 41 golf balls, 4 were damaged but counted as finds because they were sufficiently intact to perform the ordinary functions of golf balls. Partial golf balls and outer skins were not picked up.

The trout sessions landed 0, 0, 0, and 0 fish. There were no “nibbles” or other stimuli ordinarily taken by fly fishers to indicate a partial find or even a sighting.


Of course, the likelihood of catching trout on a golf course is essentially nil. Some golf courses do have ponds that may well have fish in them, so we could not entirely rule out the possibility that on some golf course one might be able to catch a trout. Similarly, of course one is likely to find golf balls on a golf course. However, we can say that for the cases presented, failure to catch a fish and the ease of finding golf balls suggest that browsing for some particular sort of thing in an environment not at all likely to hold that sort of thing is a waste of resources.

Yet these limiting cases represent only a very limited range of the behaviors termed browsing. What if we shift the stimulus to browse to a question that is less directly targeted? Two possible candidates might be: How can I acquire some food? What might I find on a walk that would make an intriguing sculpture? For the moment we will conduct these only as thought experiments. As to acquiring food on the golf course, one might notice the long-dead raccoon remains, reason that the are some small animals and set about finding rocks or golf balls to throw at the animals; or one might sharpen a stick with a rock and dig for roots of plants that might contain sufficient calories to be of use. (Recall that this is a thought exercise.) Simply fishing because one has fishing tackle would not be a reasonable match of behaviors to purpose; sampling the area for potentially available calories and making use of what is at hand (bricolage) to acquire the calories might be a better use of resources.

Wandering about the golf course with some idea of constructing a piece of art, probably a sculpture or a collage, one might keep an eye out for almost anything that could easily be carried back to the studio. A handful of golf balls, the bleached skull of a hawk, an interesting piece of wood split from a tree by a lightning strike, an abandoned t-shirt, and whatever else happened to be scattered about would be a fair target. It might be that finding some golf balls and a tennis ball and a small bone with a rounded end and a round knurled branch would inspire a sculpture of round forms, so that the t-shirt and hawk skull would be abandoned or set in a place for later retrieval.

How might we predict ahead of time that an environment is likely to be empty of useful targets, especially when we don't know what a useful target might be? How might we enhance the sampling processes necessary to determine if an environment holds the possibility of useful targets? Are there haptic elements evident in browsing paper stacks, hunting for submarines, or tracking down criminals that can be adapted to scholarly browsing online?