DJALCHEMI

Information foraging theory

Foraging rabbit,
copyright and cc-licensed by
caro wanders by mistake.

I came across information foraging theory via Jakob Neilsen, who described it as "the most important concept to emerge from Human-Computer Interaction research since 1993". Wikipedia has a reasonable overview, noting that "information seekers… use the same strategies as food foragers — informavores constantly make decisions on what kind of information to look for, whether to stay at the current site, trying to find additional information or move to another site, which link to follow, and when to finally stop the search." And there's a general introduction in a New Scientist article.

I'm interested in the theory for its possible applications to how people discover new music, in connection with the book I'm writing. Although Nielsen has referenced it a few times. There aren't that many other published applications of the theory on the personal page of Pete Pirolli, who is its principal proponent. There are a large number of academic papers on the theory among Pirolli's publications, of which I've read just the 01999 Psychological Review paper and the 02005 HCI International one. What follows is my assessment of what I like about the theory and what I don't like.

What I like

If you just read the words 'information foraging' or 'information scent', you already have a pretty good inkling of what the theory is about. I don't think you can overestimate how helpful it is to build a theory around a good metaphor — something that people already grasp and feel comfortable with (even if there's sometimes a downside when people over-extend the metaphor to areas where it no longer holds). It also means I can put a picture of a rabbit on this page, which, Jay Cross argues, means that you may learn it almost twice as well.

But what I most like about the theory is the way it puts the user and the user's self-directed activity at the centre. I've been reading recently about filters and recommendation systems, and — useful and necessary though these systems definitely are — the way they work still has an aura of 'push' about it. Foraging is clearly about self-directed 'pull'.

No matter how much you fine-tune systems for discovery, there will never be a perfect one. Do not count on foragers for loyalty. The nature of web use is that's it's quick and easy to go somewhere else (that's a big part of why people use the web in the first place). Nielsen has encapsulated this in "Jakob's Law: users spend most of their time on other sites" (quoted from about two thirds down this page, and at the top of this one).

Information foraging theory leads us to think in terms of people 'information snacking': picking the low-lying fruit, perhaps, and then moving on elsewhere. It develops models of 'information patches', where information relevant to a forager's needs tends to be found in clusters or patches, but once the yield of one patch starts to decline, so does the forager's contentment, and they start looking elsewhere for alternative patches.

I like, too, the idea of 'information scent', which models how how people sniff out when something looks interesting. The scent analogy allows for the role of coincidence, when you just happen to be passing through an area, and catch a whiff of something tasty. It also caters for some 'push' promotions, as when the marketers spray some sweet-smelling stuff around the environment to attract your attention and get you salivating.

The Psychological Review paper also points out that information foragers can mould the environment to suit their strategies. This might include bookmarking or using Furl and such-like, or it could be organising the piles of paper and filing cabinets around your desk.

What I don't like

Perhaps, not for the first time, one of the things that makes the theory so attractive also carries part of its potential undoing. The idea of wild animals foraging for food is, as I've said, richly suggestive. But let's beware of taking it too literally.

People's desire for information is not the same, and does not have the same dynamics, as animals' desire for food. Information feeds the mind and the soul in complex ways, not just the stomach. Information teases and lingers, longer even than the taste of kippers, and new information changes the value and sustenance of old information.

I'm not sure if information foraging theory captures this richness of motivation, desire and its satisfaction. It assumes a simple, single-minded focus on the part of the human user. This is, I imagine, also necessary to devise the detailed mathematical models of information foraging that are detailed in both the research papers I read. (I confess these formulae are all 'applied Greek' to me; I used to be able to handle this kind of maths 20+ years ago, but no more.)

In this respect the theory reminds me of another one, GOMS, which was also devised by Stuart Card, co-author with Pirolli of the Psychological Review paper. GOMS assumes what I call a Cartesian model of human subjectivity and desire: that it starts with a clearly defined goal and everything that follows, the tasks, formulation of sub-tasks and their execution, follows like a waterfall. But many, including me, would argue it doesn't often (or ever) quite work like that. Lucy Suchman's Plans and Situated Actions was one of the first from within Human-Computer Interaction to challenge the Cartesian, waterfall model. Suchman gave examples of circumstances where people set out with vague and fuzzy goals, and then improvised as they went along. Sometimes people invent goals post-hoc to justify these improvisations. And that's before we've started on unconscious motivations.

Goals such as 'discover new music I like' are inherently fuzzy, and I'm doubtful whether it lends itself to detailed mathematical modelling.

In conclusion, on information foraging theory: useful for the broad sweep user-driven perspective and its suggestive analogies, but don't mistake analogies for identities, and be cautious about the detail.