Sunday, July 24, 2011

No test - inconceivable. Not Technical Debt.

Technical Debt

The software term "technical debt" is getting a lot of play on the air waves.  But I do not think we are using it the way Ward was when he invented the term (a metaphor) to explain to his buisiness team why creating software fast to get feedback was a good thing.  But that they had to be willing and able to sustain a pace of repayment on the debt of doing just good enough design to get product feedback.  Their form of repayment was constant refactoring.  Always keeping the software model moving toward the best possible business model, which modeled the real world.  Using many XP practices to enable the repayment plan of the debt they were consciously assuming.

In this vain, technical debt does not cover the process of writing bad code, of poor design, of skipping steps (such as testing).  Those behaviors would be considered to be incompetent design and implementation.  That behavior results not in debt at all but a breach of the inherent contract between development team and the business. The warranty of merchantability.

The warranty of merchantability is implied, unless expressly disclaimed by name, or the sale is identified with the phrase "as is" or "with all faults." To be "merchantable", the goods must reasonably conform to an ordinary buyer's expectations, i.e., they are what they say they are. For example, a fruit that looks and smells good but has hidden defects would violate the implied warranty of merchantability if its quality does not meet the standards for such fruit "as passes ordinarily in the trade".

This industry (software development) is deceiving its customers and pass off bad products as if they just have a little bit of debt to be repaid.  As if the product was a 3 year old car with a 5 year loan.  One purchases the car and the debt.  But I suggest that is not the same thing as purchasing software the has been created in such a poor fashion as to have no unit test, or no acceptance test. Little if any way to become the application it appears to be.  A car on the outside but with a faulty electrical system, blown engine, and bad transmission, but really good tires and paint.

A case in point - a team wishes to upgrade the compiler that produces their application.  They have little to no tests for the application - it just works.  They then ask for a test group to provide the effort required to prove that the compiler upgrade doesn't cause any bugs.  I do not think one can pass off the warranty of merchantability to the test team and expect good things to come from this - pass the hot potato behavior.  This is not technical debt - it is incompetency - it is inconceivable.

"You keep using that word. I do not think it means what you think it means."
-- Inigo Montoya

Ward Cunningham on the creation of the "Technical Debt" metaphor.

I think we keep using the phrase technical debt - but it doesn't mean what we think it means.  Technical debt means a conscious decision to defer up-front design and research in the product development, in order to get to market with a model that is capable of becoming the the desired solution, and capable of eliciting the customer feedback that we desire, which proves that the product is evolving in the proper direction.  And provide return on investment earlier.

See Also:

A Technical Debit - Collateralized Debt Obligation you should not invest in

The SQALE method is particularly devoted to the management of the Technical Debt (or Design Debt) of software developments. It allows:
  • To define clearly what creates the technical debt
  • To estimate correctly this debt
  • To analyse this debt upon technical and business perspective
  • To offer different prioritisation strategies allowing establishing optimised payback plan.
Managing Software Debt - by Chris Sterling

Post a Comment