KM: False positives or False negatives

In a conversation yesterday with a technology analyst, I got into a debate about the desirability of false responses. Which is more desirable (from a KM perspective): a false positive (i.e., a response to your query which turns out to be wrong), or a false negative (the system telling you it has nothing relevant, when in fact it does but it isn’t connecting the dots properly)?

I argued for a false negative, on the assumption that users will eventually stop using a system that leads them down too many dead-ends. The analyst argued for a false positive, using the analogy to a doctor: at least with false positives, you have something to test to see if it’s right or wrong. A false negative, you assume that it’s non-existent and move on… which is when people die.

So I’m kicking it to you guys — which would you prefer? A system that gives you an answer that turns out to be wrong, or an answer saying there is no information related to your search? Why?

4 responses to “KM: False positives or False negatives”

  1. in KM, False positives definately. As a support guy (I used to do enterprise-level multivendor network support), I'd rather get some things that *might* be related and some straws to grasp at than nothing at all. More often than not, even if the articles weren't related they'd start a productive train of thought.

  2. I tend to err on the side of false positives as well. Quite frequently, one of those false positives will take me down a road I had not considered which ends up being relevant. In addition, a false negative often gives the researcher the impression that there is nothing on the topic at all, leading them to erroneously and prematurely end their research. False positives also teach researchers to build better queries for the future bc they see what they have possibly done wrong to get them to that point, whereas if there is no response at all, there will be no reason to question the query.

  3. I like both arguments for increasing the accidental and/or opportunistic knowledge discovery.I'm just a bit skeptical that this is how busy, non-tech-savvy users approach the problem. Don't they get discouraged when the system “wastes” their time with incorrect info?–Rick

  4. From a lawyer perspective, false positives. When I use technology to do legal research, I often receive false positives — that is, results to my searches which do not fit what I want. But when I review the results, I quickly get ideas that I didn't have before, can follow leads to the information I originally wanted, or learn why my search parameters are lacking.False negatives can lead to the worst possible scenario: Telling a court that no case exists on point when one does.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.