Erik Pukinskis

Linguistic theory of user-centered design

The HCI community and the Human Factors community before them have long known there are tangible benefits to user-centered design. But doesn't seem to be a great understanting of why computer-centered design has failed and where user-centered design is leading us. There are two syptoms of this lack of understanding: the lack of power exhibited in modern graphical interfaces compared to text-based computer languages and the difficulty in learning new skills even in systems design from a user-centered standpoint. The lack of power can be understood best when we plot the usability versus power of computer languages and graphical interfaces on the same coordinate plane (figure 1). The usability and utility of graphical interfaces is plateauing.

I would like to suggest a cognitive theory of human computer interaction that will give us an understanding of why computer-centered interface design failed, where user-centered design is taking us and why it is inadequate. The human mind is expert at processing a multitude of natural languages. English is one, the language of colors shapes and spatial positions is another. In addition, the human mind is very capable of creating pidgen languages when faced with using a new language. But natural languages are in most cases, not possible to be formalized. This creates a real problem for computers which are only able to understand formal languages, and relatively simple ones at that.

This is in part the reason that computer languages and computer interfaces are artificial by comparison...

12/12/2003 - Related to this topic is the situation, as in Pinker (1994, pg 27), in which a deaf boy grows up learning ASL from his parents, who use certain grammatical constructions incorrectly. Despite hearing an inconsistent mix of incorrect forms, he learns all by himself to use the correct forms. This implies he was somehow able to filter out the ungrammatical "noise".

I wonder if this is also the case in people learning computer interfaces. They filter out the noise of behavior which is inconsistent with some sort of natural way of interacting with the computer. Hmm.. tenuous thoughts. Must sort this out.


 
This page was last updated December 12, 2003 at 4:53pm.