Taxonomies, Invisibility and Management

In Science Daily, a report that a species of skate has been fished to near extinction because of a taxonomic error in the 1920s. Two species of skate were classified into a single species by an influential biologist R.S. Clark in 1926. This act effectively made it impossible to “see” both species as distinct, and so the over-fishing of one of them was simply not tracked. Taxonomies are designed to foreground salient things, and background un-salient things. If you make a mistake, you cannot manage or safeguard the things you render invisible. This error is a good example of why taxonomies in management are so important – it’s not simply a matter of rendering lots of documents searchable. It’s also about mapping the world of work to render it manageable. Next time I get an invite to put together a taxonomy by reviewing a pile of documents over a couple of weeks, I’ll think of the humble – and almost extinct – flapper skate D. intermedia.

4 Comments so far


This is a great example that displays the importance of taxonomy. And, for me, it stirs the discussion could a system-generated taxonomy create a more accurate taxonomy (semantics)? A human decided to classify these two species into one, a human error, right? And a semantic engine would help us to create and manage very granular categories. What do you think about this?

Posted on December 14, 2009 at 04:52 PM | Comment permalink

Patrick Lambe

Computers are not very good at handling salience, especially where there are “soft” issues of interpretation. In my book Organising Knowledge I cite the case of Walmart, who used an auto-classifier to suggest “related videos” to customers who bought videos from its online store. It got into trouble when its auto-classifier started doing things like associating the word “chocolate” with African-American themes. I have another post on a related topic tomorrow. Here’s the Walmart link:

Posted on December 14, 2009 at 05:01 PM | Comment permalink


I have to admit I am not experienced with semantic programs and so far categorisation in my working context happened manually as well. And the result is mixed. Therefore, I have developed an interest in semantic engines.
Looking at the Walmart example and some experimental tests within my working context, I can only agree that we should not entirely rely on machines to create taxonomies. What I can imagine in a library context is that engines could suggest improvements to an existing taxonomy as a the content is developing.
Let’s see what results are showing us.

Posted on December 15, 2009 at 03:42 AM | Comment permalink

Patrick Lambe

Yes, absolutely, in the context of suggesting categories, and managing validated relationships between categories, the semantic engines have a lot of value. The danger arises when they are taken to be magic black boxes, removing the human visibility into the system.

Posted on December 15, 2009 at 12:55 PM | Comment permalink

Page 1 of 1 pages

Commenting is not available in this weblog entry.

Comment Guidelines: Basic XHTML is allowed (<strong>, <em>, <a>) Line breaks and paragraphs are automatically generated. URLs are automatically converted into links.