[Sigia-l] Usability Testing comments from Giga

[Jared M. Spool]

A late contribution to this thread:

Having just returned from the User Interface 7 Conference, I found this 
thread to be very interesting.

Ziya asked an important question: does every decision need to be tested.

Others compared IA/Usability testing to software testing.

As someone who started as a software developer in the '70s, I noticed that 
nobody discussed the evolution of software quality in practice.

When I first started as a programmer, it was common practice to look at the 
results of compiled code. (For those not versed in the lexicon of computer 
science, a compiler takes human readable programs and converts them into a 
machine code for the computer to execute.) We would regularly print out the 
machine code, checking to see if the compiler had coded to our expectations.

This was especially true in the early days of optimizing compilers. These 
tools would often make poor guesses about the programmer's intentions, 
creating code that didn't work. We would sit with printed output from the 
compiler, going over every machine instruction, making sure that stack 
corruption, memory overflows, and inefficient code wasn't dragging down our 
results. I remember, more than once, hand modifying the compiler output to 
produce code that worked the way I needed it to.

In today's development environments, this is rarely necessary. In fact, 
most programmers don't have the skills to do this type of debugging and 
analysis. Because our knowledge of compiler development has advanced 
tremendously from those early days, those crude (yet vital) verification 
and correction techniques have all but disappeared from the profession.

This is all about confidence -- trusting that we have the underlying 
knowledge, practices, and methodologies to produce reliably effective 
results every time. Today, our confidence is weak. (In fact, I'd say that 
those who *are* confident are foolishly so.) We can not produce predictably 
good results with every design.

This comes back to the basic research we need to have to move forward. We 
need to know why certain design elements work sometimes, but not others. We 
need to know how different types of content dictate different types of 
elements. We need to know practices that produce effective designs -- that 
meet user requirements -- every time.

While an experienced navigator can trust his gut to say that the ship is on 
course, having instruments to verify position, speed, and conditions gives 
everyone on board the confidence needed.

Jared

Jared M. Spool        User Interface Engineering
http://www.uie.com    jspool at uie.com
Don't miss the UIE Roadshow: 4/22-6/12
DC, NYC, San Diego, Austin, and many more cities
http://www.uie.com/events/roadshow