The only difference between technical debt and financial debt is that costs are more often known in advance when taking on financial debt. Both types of debt are a tool when used intelligently with purpose and a plan to manage it, and both can take a devastating toll when used recklessly or imposed through misdirection or miscommunication.
Acceptable vs. unnecessary debt
The original heading here was “necessary vs. unnecessary debt.” On further reflection, though, I realized that the only good reasons for incurring debt are time drive. If time is removed as a factor, there is no reasonable need for debt. So then it becomes a question of when time is important enough of a factor to make debt acceptable. The only context I can think of where time is universally an acceptable driver for debt is in an emergency.
Beyond an emergency, the evaluation for whether debt is acceptable because of time becomes a value proposition. In our personal lives, the first car and house is generally considered to be a good reason to accept debt because both have a large-enough cost where they are likely to become more expensive over time, making it harder and harder to save for them in a reasonable period of time.
Similarly, building in-house custom applications rather than waiting for a commercial off-the-shelf (COTS) solution that will incur technical debt in minimally reviewed code and the inevitable maintenance costs is worth it for functionality that is key to business value. Having worked for software vendors, I can honestly say that, if it isn’t already generally available (GA) as at least a Patch 1, it should still be considered unavailable as a COTS solution.
The other common time driver that should generally not be an acceptable reason to take on debt is impatience. Using a home-equity loan to buy the latest television is a poor financial decision, and implementing a new solution without a thorough evaluation and proper training is a gamble that will usually result in higher maintenance cost or a potential system failure.
The old adage “patience is a virtue” is not only true, it is a vast understatement of the value of patience.
Stop debt before it happens
The reason technical debt is becoming an increasing concern at many companies is because it tends to grow exponentially, just like financial debt. And for the same reasons. Of the three drivers for debt mentioned previously (emergency, long-term value, short-viewed impatience), the most frequent cause is the least necessary: impatience. Problems arising from bad habits will grow until the habit has been replaced by actions that have a more positive effect.
Without getting too psychological here, impatience is a result of either wanting very much to move towards a reward or away from loss. For some odd reason, the drive forward doesn’t seem to repeat in the same context nearly as much as the drive to move away from. In technology, the drive to move away from is so common that the three key emotions related with impatience driven by escape have an acronym: FUD (fear, uncertainty, doubt). In the case of IT decisions, all three are essentially redundant, or at least a sequence. Fear driven by uncertainty and/or doubt. When the decision is around taking on technical debt, the fear is that business owners or customers will be upset if the feature is delayed or reduced and the uncertainty and doubt are the result of either not asking these stakeholders or asking only half the question.
Asking a stakeholder “Is it a problem if feature X is not in the release?” will frequently have a different answer than “Would you prefer we include feature X in a later release or risk certain delays to all future feature releases by pushing it before we have time to include it in a maintainable manner?” My experience is that most of the time neither question is asked, and it is just assumed the world will end if users don’t have access right now to a new option that only 3 percent will ever use. It is also my experience that when the trade-off of reliability and stability versus immediacy is explained to stakeholders they usually opt for the delay. I know many people believe that businesses have lost sight of long-term implications and I believe that in many cases it not because they are deliberately ignoring them but because the people that should tell them when and why to be cautious are afraid of saying anything that will be considered “negative.”
To summarize, the best way to reduce the accumulation of technical debt is to have open, honest communication with stakeholders about when decisions involve technical debt, the consequences of that debt, and the options for avoiding taking on the debt. Then, if the decision is to still choose the right now over the right way, immediately request buyin for a plan, timeline and budget to reduce the technical debt. Again, my experience is that when the business is presented with a request to ensure functional reliability they frequently say yes.
Getting out of unavoidable or accepted debt
Taking on some technical debt is inevitable. This is why I used the modifiers usually, most often, and frequently in the previous section rather than more-comforting-yet-inaccurate always, definitely, and every time. Even in a theoretically perfect process where business always opts for debt-free choices and emergencies never happen, there are still going to be debt-inducing choices made either from lack of information or use of imperfect vendor releases.
In the case where the debt is incurred unknowingly, once it is discovered be sure to document it, communicate, and plan for its correction. The difference with cases where the debt is taken on knowingly because it is unavoidable without a much larger cost in vendor change, monitor the item with every project, and, when there is a reasonable option to correct it, do it. I once had to build something that was a bit kludgy because the vendor application clearly missed an implication of how a particular feature was implemented. We created a defect in the defect tracker that was reviewed in every release. Eighteen months later, the vendor found the error and corrected it, and we replaced the workaround with the better approach in the next release. For major enterprises, it is a good idea to raise a support case with the vendor when such things are identified, which I did not do at the time because the company I was managing this application for was too small to get vendor attention and the feature was not in broad use.
This article is published as part of the IDG Contributor Network. Want to Join?