Jean-Paul S. Boodhoo

Sponsors

The Lounge

Wicked Cool Jobs

Syndication

Recent Posts

Tags

View more

Archives

Advertisement

Jean-Paul S. Boodhoo » And here's a number you can take to the bank!!
Images in this post missing? We recently lost them in a site migration. We're working to restore these as you read this. Should you need an image in an emergency, please contact us at imagehelp@codebetter.com
And here's a number you can take to the bank!!

In reference to the article on unit testing, here is an interesting excerpt from the following article written by one of my former colleagues Paul Julius:

“ The reason I was there was simple. Earlier that year, those same IT managers had performed a series of calculations to estimate how much it cost the department each time a bug made its way out of development, into SIT, into User Acceptance Testing, or all the way into production. For this IT shop, one bug into SIT cost them $12,605 (and you can imagine how much a bug into production would cost.) So, I did some quick math. The burn rate for the developer who introduced the bug was $70 per hour. Since the developer had already fixed the bug and committed the change, I deducted the $70 from the total and ended up with $12,535. Now, it cost them $70 (or less) to fix that one bug, instead of $12,605.”

So I’ll ask you again. Would you rather spend $12000 to fix a bug or $70?

Posted Wed, May 23 2007 10:39 AM by bitwisejp
Filed under:

[Advertisement]

Comments

Fregas wrote re: And here's a number you can take to the bank!!
on Wed, May 23 2007 3:11 PM

I don't get it.  How could one bug cost them over $12k?  

Trevor wrote re: And here's a number you can take to the bank!!
on Thu, May 31 2007 3:32 PM

The math in this example is incorrect.  So, he compares the cost of the average bug (which is ridiculous, this must be an extraordinarily high ROI application, or they are extraordinarily incompetent in responding to bugs) to the cost of a developer responding to the failure of a regression test in development.  Nowhere do we see the cost of developing the regression tests themself.  So without the complete numbers on both sides, the story is inconclusive and meaningless.

Not to say testing is not good, but this example is not something that  you can "take to the bank", simple as that.

Does this story honestly, to you, act as DEFINITIVE proof that TDD is the way to go?  I think in certain scenarios it definitely is, but we don't get enough details here to make that conclusion.  Zealotry.

Devlicio.us