Help isn't all we need


Jeremy scooped me in his recent post where he commented on a recent SXSW panel on social search that included Marc Vermut, Brynn Evans, Max Ventilla, Ash Rust, and Scott Prindle. Jeremy pointed out that in addition to asking for help and embarking on a solitary search, was the possibility (discussed many times on this blog) of embarking on (an exploratory) search together. Searching together, collaboratively, is often appropriate when faced with exploratory (rather than known-item, factiod, or trending topic) information needs. Collaboration works best when information needs are shared, and when the results need to be created rather than merely re-discovered.

In an exchange on Twitter, Brynn pointed out that instances of true collaborative search comprised less than 10% of the instances she and colleagues had recorded in their study of Mechanical Turk respondents. But that argument misses the point.

While a survey of random respondents in 1910 might have yielded a small number who had experience with automobiles, that was not sufficient evidence to dismiss the utility and wide applicability of the (then) new technology. Whereas incremental redesign should rightly draw on people’s experiences with a technology to inform future iterations, the situation with new or more innovative technology is more complicated. You can’t ask for people’s opinions about things they haven’t experienced and internalized, so designers of new technology must rely more on first principles and on inference rather than on concrete observations of that technology in use.

The real problem is that currently people don’t have many tools that allow them to engage in true collaboration, and thus we, as users, and as designers, do not have enough collective experience with the associated phenomena to dismiss them as unimportant based on lack of evidence of use.

Fortunately, in the case of collaborative search, there is considerable evidence of its presence, if one knows where to look. The library science literature is full of accounts of collaboration in information seeking, spanning twenty or more years. In more recent work, Amershi and Morris  report many instances of collaborative information seeking in the enterprise, one of the factors that surely motivated the creation of their SearchTogether tool.

In summary, although there are many instances of search that would be well served by the tactics described in the SXSW panel, we should not dismiss other related kinds of information seeking just because existing tools fail to support them well.

Share on: 

1 Comment

  1. Good point, Gene!

    At the recent Search in Social Media (SSM) workshop at the Web Search and Data Mining (WSDM) conference in New York last month, Jan Pedersen from Micrsoft Research drove this point home. I don’t know if he said it as a dig to Google or not, but during his keynote, he explicitly went out of his way to point out that just because you cannot (yet) observe a particular user behavior from the query logs doesn’t mean either (a) that need doesn’t exist, or (b) wouldn’t arise, if given the chance.

    Google, especially with its over-reliance on things like A/B testing, tends to only work on what they can see directly in front of them, i.e. what they actually observe users doing. Pedersen says that this misses the point, and I tend to agree. Yes, you should follow the user as much as possible. But there are times when you have to lead, too, because that’s the only way you’re going to be able to create something truly new and innovative.


Comments are closed.