Coding Horror: The Cost of Complexity

November 07, 2004

The Cost of Complexity

There's an interesting eleven page article in the Economist considering the cost of software complexity:

The economic costs of IT complexity are hard to quantify but probably exorbitant. The Standish Group, a research outfit that tracks corporate IT purchases, has found that 66% of all IT projects either fail outright or take much longer to install than expected because of their complexity. Among very big IT projects—those costing over $10m apiece—98% fall short.
Gartner, another research firm, uses other proxies for complexity. An average firm's computer networks are down for an unplanned 175 hours a year, calculates Gartner, causing an average loss of over $7m. On top of that, employees waste an average of one week a year struggling with their recalcitrant PCs. And itinerant employees, such as salesmen, incur an extra $4,400 a year in IT costs, says the firm.

It's a great article; be sure to read through all the sections.

I think everyone agrees that usability is a desirable goal, but what they don't acknowledge is that usability is expensive. Both in terms of development time and absolute budget dollars. And worse, it is a soft cost: it's hard to quantify how much money your company will save if users spend 10 minutes less per day flummoxed by the software they're using. So usability tends to get short shrift.

It's technically more "efficient"-- in terms of budget dollars-- to build a quick and dirty solution that's hard to use. All you're really doing, though, is hiding the true cost of your software behind hours of user pain, and perpetuating the vicious IT cycle that causes users to avoid computers in the first place. We have to do better.

Posted by Jeff Atwood View blog reactions

« Is DVD the new VHS? Captcha Control Coda »

Comments

You should take a look at Jakob Nielsen's website (every usability info you need)

Misconceptions About Usability:
http://www.useit.com/alertbox/20030908.html

Usability 101: Introduction to Usability:
http://www.useit.com/alertbox/20030825.html

Benjamin Francisoud on November 10, 2004 04:31 AM

actually, the most alarming fact is, that the problem has not changed a bit since becoming initially evident as "software crisis" over 30(!) years ago.

for one of the first public mentions apart from the NATO conference in 1968, read dijkstra's turing award lecture (EWD340). you might also want to have a look at EWD1304 (written in 2002). (archive at www cs utexas edu /users/EWD - sorry, your blog sw regards this link as "questionable content"...)

iirc there further are several reports from the late 70ies and early 80ies containing nearly the same percentage of failures etc. (no source at hand, sorry)

obviously, the attempts to solve the problem only in terms of software engineering techniques have failed.

now enter usability, so everybody's got a nice buzzword to rave about. but one popular misconception about usability - which is btw not mentioned by nielsen - is, that usability engineering would be the solution to the problem. unfortunately, it isn't - at least it's not the whole story. usability in the first place is about knowing, what's wrong, not about how to make it "right". while it can and should reveal the weak points of a product, you still need design methods and knowledge (i.e. synthetic thinking) to tackle them. heavily recommented reading: cooper/reiman, about face 2.0 (www.amazon.com/exec/obidos/tg/detail/0764526413/104-4374149-1260717?v=glance) not flawless, but still the very best book on the issue, i know.

somehow, there has to be a way out of this mess. more multi-discliplinary and human-centered approaches might help a little bit on the way. ;-)

sascha brossmann on November 13, 2004 09:58 AM

addendum: while the focus of the early software crisis years laid more on engineering complexity and did not automatically include the end user's view of "complexity", i don't think this changes the general line of the problem.

sascha brossmann on November 13, 2004 10:22 AM