[Sigia-l] card sorting: dealing with multiple placements

[Boniface Lau]

> From: sigia-l-admin at asis.org [mailto:sigia-l-admin at asis.org]On
> Behalf Of Todd R.Warfel
> 
> On Thursday, May 29, 2003, at 07:38 PM, Boniface Lau wrote:
> 
> > The fact is that card sorter was invented in the late 19th century
> > to sort census punched cards. Thus, the tool was intended for
> > ordering.
> 
> I'll take your word for it that it was originally invented in the
> 19th century to sort census punched cards, although I would be
> interested to know where you found that information.

You may want to get a copy of Knuth's classic "The Art of Computer
Programming, Volume 3: Sorting and Searching":

http://www.amazon.com/exec/obidos/ASIN/0201896850


> I'm not familiar with how it was used for census punched cards, so I
> can't argue towards ordering or grouping there. However, I do know
> that IBM made card sorting popular for IA as a means of grouping.

As I had commented at the beginning of this thread, IAs are using an
ordering tool (card sort) for grouping purpose.


> 
> As card sorting relates to IA as a practice it is for grouping, not
> ordering. 

Thus, earlier I gave the analogy of stirring coffee using a knife. It
works. But people need to be aware that they are using something not
designed for the task at hand. Otherwise, they may end up accepting a
distorted view of reality.

For example, in reality there are plenty of things that fit equally
well in multiple groups. It is like breathing through nostrils. People
typically do not care whether they use the left or right nostril. They
use both. But since card sort does not allow an item to be grouped
under multiple groups, users are forced into an either/or mentality.
It is like asking users to decide whether they prefer breathing
through the left or right nostril.

To deal with the above limitation, some people allow card sort users
to duplicate items and place the duplicates in multiple groups. But
such duplication means users are no longer sorting. A deck of playing
cards ends up being bigger after sorted is a bizarre way of sorting.
Furthermore, duplication changes the way the results are analyzed.


[...]
> 
> > BTW, many years later John von Neumann wrote a sort program and
> > used the card sorter as performance benchmark.
> 
> For what?

For justifying sorting using a general purpose computer, instead of
card sorter. Remember that the card sorter was invented long before
the general purpose computer.


> 
> > History aside, arguing that card sorting was not meant for
> > ordering things is like arguing that sorting was not meant for
> > ordering. It ignores the established meaning of words.
> 
> I don't agree. Sorting is meant for grouping - grouping is meant for
> ordering.

Couple things about sorting:

First, to sort things does not mean you have to group them first.
There are plenty of techniques not requiring grouping. For example,
selection, insertion, exchanging, etc. Thus, sorting is NOT meant for
grouping.

Second, regardless of the sort techniques, the ultimate goal of
sorting is order. Thus, sorting is about ordering, not grouping.


Boniface