Last updated on August 24th, 2018 at 02:17 pm
Few performance management systems provide guidance with respect to behaviors relating to technical debt, perhaps because technical debt is not widely understood, or perhaps because technical debt isn’t seen as a concern for the performance of anyone but engineers and their managers. Still, organizations that expect to gain control of technical debt must ensure that performance standards are clear about expectations with respect to behaviors that could affect technical debt. In organizations in which technical debt currently plays a minor role, if any, in the performance management system, policymakers can advocate for effective changes, if they understand what the appropriate role for performance management is in controlling technical debt. This post should be helpful.

A fundamental premise of many performance management systems is that incentives can encourage desirable behavior and disincentives can discourage undesirable behavior. Unfortunately, serious questions have arisen about the effectiveness of these behavioral control mechanisms in general [Kohn 1999]. The problem is that employees find ways to harvest incentives without exhibiting the desired behavior. Likewise, they find ways to circumvent disincentives while continuing to exhibit undesired behavior.
Moreover, specifically for technical debt management, behavioral control is especially problematic, because some of the behaviors that must be controlled are inherently immeasurable. For example, the design of an incentive structure to encourage legacy technical debt retirement is debatable, given the technical difficulties involved in even defining legacy technical debt, let alone measuring its size.
Managing performance vis-à-vis technical debt, therefore, presents a problem of the kind Austin calls partially supervised [Austin 1996]. Supervising engineers whose work touches on assets that bear technical debt can only be partial, because measuring technical debt is only partially practical given the state of the art. Austin shows how partial supervision frequently leads to dysfunctional performance management, but the problem is especially vexing for managing technical debt. For example, in some cases, engineers’ work can incur new technical debt that remains unrecognized for months or years after the work is completed. To fully supervise such work would require inventing retroactive incentives and disincentives, which not only do not exist, but which are of questionable legality in most jurisdictions.
Although incentives and disincentives cannot serve to manage performance relative to technical debt, a very effective model is available. Enterprise leaders could communicate their intentions relative to technical debt, and empower the people of the organization to take steps to reduce debt. In the United States military, and others as well, a doctrine that implements this approach is called commander’s intent [Mattis 2008].
Gen. Mattis offers five principles that guide what the military calls “effect-based operations.” For technical debt management, the effect we seek is rational control of the technical debt portfolio. Here are his five principles, transformed to the field of technical debt.
- Technology development, maintenance, and cyberdefense in the future will require a balance of conventional and unconventional approaches.
- Technology evolves rapidly, and we must be willing to adapt our methods.
- Technologies are dynamic with an infinite number of variables; therefore, it is not scientifically possible to accurately predict the level of technical debt that will result from any given effort. To suggest otherwise runs contrary to historical experience and the nature of modern technological assets.
- We are in error when we think that what works (or does not work) in efforts involving one technology in one enterprise will be universally applicable to all technologies in all enterprises.
- Finally, to paraphrase General Sherman, “Every attempt to make technical debt management easy and safe will result in humiliation and disaster.”
Most organizations rely on supervisors to communicate the analog of commander’s intent to their subordinates. Currently, it’s fair to say that few supervisors outside the technology-oriented elements of the enterprise communicate much about technical debt to their subordinates.
That situation might explain why most performance management systems encourage behaviors that unwittingly expand the body of technical debt, especially for non-technologist performers. There are situations in which the widely applauded actions of the outstanding performer are such as to incur technical debt strategically and responsibly. Technical debt so incurred is what McConnell calls Type II [McConnell 2008] and what Fowler calls Deliberate and Prudent [Fowler 2009]. But most performance management systems, especially for non-technologists, say nothing about technical debt, and thus risk encouraging behaviors that indirectly exacerbate the problems associated with technical debt.
Distinguishing responsible and irresponsible behaviors is possible only if understanding of the nature of technical debt is widespread in the organization, even beyond the technologists. Here’s an example:
It was ambitious, what advocates called a “stretch goal,” but the VP of Marketing approved the plan to get the new app released by the end of the next fiscal quarter. After a month of meetings, and much jawboning, the CTO agreed to find a way to make it happen, despite serious objections from the VP of New Product Development. Engineers and testers were able to meet the date, but they had to incur significant technical debt, and when they asked for resources to retire that debt after the release, the VP of Marketing opposed the request, because she needed additional resources for the promotional campaign due to our late entry into the market.
Stories like this illustrate scenarios in which technical debt considerations are consistently assigned a lower priority than goals related to market timing, market development, and revenue generation. Standards for setting priorities closely parallel the standards defined in the performance management system. Indeed, the goal of performance management should be to support enterprise goals. In the scenario above, the organization might meet the immediate goal of a successful release, but it does so by incurring technical debt, thereby imperiling the next release. In this scenario, it’s evidently necessary to change the performance management system to achieve a better balance between immediate goals and the near-term future goals.
Since anyone in the enterprise can take actions or make decisions that lead to incurring new technical debt, or cause existing technical debt to remain in place, organizations need performance standards that guide employees with respect to technical debt. To provide guidance for distinguishing responsible behavior from irresponsible behavior, performance management systems must acknowledge the potential of any employee to affect technical debt, constructively or otherwise. Performance management systems must be reviewed with respect to alignment with technical debt policy, and adjusted to encompass a mechanism analogous to Mattis’s vision of commander’s intent.
References
[Austin 1996] Robert D. Austin. Measuring and Managing Performance in Organizations. New York: Dorset House, 1996. ISBN:0-932633-36-6
Contains an extensive discussion of the consequences of partial supervision of performance. Since technical debt can only be partially supervised, the concept is relevant to understanding the effects of performance management systems on technical debt. Order from Amazon
[Fowler 2009] Martin Fowler. “Technical Debt Quadrant.” Martin Fowler (blog), October 14, 2009.
Available here; Retrieved January 10, 2016.
- Technical debt in software engineering
- Team composition volatility
- How performance management systems can contribute to technical debt
- Unrealistic definition of done
- Spontaneous generation
- Legacy debt incurred intentionally
- Controlling incremental technical debt
- Refactoring for policymakers
[Kohn 1999] Alfie Kohn. Punished by rewards: The trouble with gold stars, incentive plans, A's, praise, and other bribes. Boston: Houghton Mifflin Harcourt, 1999. ISBN:0-395-71090-1
[Mattis 2008] James N. Mattis. “USJFCOM Commander’s Guidance for Effects-based Operations,” Joint Force Quarterly 51, Autumn 2008 105-108.
Available: here; Retrieved November 9, 2017.
[McConnell 2008] Steve McConnell. Managing Technical Debt, white paper, Construx Software, 2008.
Available: here; Retrieved November 10, 2017.
- Nontechnical precursors of nonstrategic technical debt
- How performance management systems can contribute to technical debt
Other posts in this thread
- Non-technical precursors of non-strategic technical debt
- Failure to communicate long-term business strategy
- Failure to communicate the technical debt concept
- Technological communication risk
- Team composition volatility
- The Dunning-Kruger effect can lead to technical debt
- Self-sustaining technical knowledge deficits during contract negotiations
- Zero tolerance and work-to-rule deliveries create an adversarial culture
- Stovepiping can lead to technical debt
- Unrealistic definition of done
- Separating responsibility for maintenance and acquisition
- The fundamental attribution error
- Feature bias: unbalanced concern for capability vs. sustainability
- Unrealistic optimism: the planning fallacy and the n-person prisoner’s dilemma
- Confirmation bias and technical debt
- How outsourcing leads to increasing technical debt
- How budget depletion leads to technical debt
- Contract restrictions can lead to technical debt
- Organizational psychopathy: career advancement by surfing the debt tsunami
- The Tragedy of the Commons is a distraction
- The Broken Windows theory of technical debt is broken
- Malfeasance can be a source of technical debt