Looking at Next-Gen Answers to Technical Debt — Enabling Sustainable Modernization (Part III)

In Part I and Part II of this series, we explored how debt is inevitable when undergoing digital transformation, no matter the measures you take and fail-safes you put in place. Part I specifically looked at the impacts of debt accumulation on your business and hinted at why Cognitive could provide a viable solution. Part II went back to the technical routes of debt origination, demonstrating the role developers have to play.

Let’s now consider the evolving definition of debt as we progress further on the digital maturity curve and why these new questions call for innovative answers.

Reacting to Debt: Models, Software, and Tools to Explore

With digital becoming the core of modern businesses, the problem is now pervasive, impacting more than just IT. Deloitte estimated that a typical application contains an average USD 3.61 in technical debt for every line of code — this is a massive number when you consider that software now determines productivity, marketing, sales, customer relationships, and on-ground operations.

To address this, organizations are trying to come up with their own debt resolution tactics.

Approached adopted by firms to manage technical debt

Usually, organizations use “structure change” as a resolution method: introduce a Product Manager responsible for not just storyboards and features, but technical debt as well. However, this approach is applicable only if the Manager adopts agile processes and development.

On the other hand, more mature organizations establish best practices such as Architectural Debt analysis. These are validated and followed-up by regular reviews of the architectural style, patterns, and principles, ensuring coverage for non-functional requirements, infrastructure, and deployment models when scanning for debt.

Another practice which organizations can look at, is blending technical debt with business value. This is encapsulated in the concept of “Business Value Debt.” Business Value Debt analysis covers application functionality, along with rules, processes, and policies to gauge their alignment and maturity. A key tenet of Business Value Debt is the analysis of ease-of-use and support ticket properties.

Finally, there are leading providers who have focused on building solutions and models around debt reduction. An industry-wide accepted benchmarking model is the one presented by the Consortium of IT Software Quality (CISQ), based on the parameters of Reliability, Security, Performance Efficiency, and Maintainability of IT software. CISQ is a Special Interest Group of the Object Management Group (OMG) established as an industry-led initiative to improve software risk management and software acquisition practices by measuring the structural soundness of business-critical IT software.

Coming to ready-to-deploy debt resolution solutions, standout products and providers include:

● CAST — The CAST platform comes with a system-held dashboard, an engineering dashboard, a security dashboard, and an imaging system to ensure end-to-end digital enterprise resilience. Its comprehensive nature makes it a top pick for debt resolution.

● SonarQube — This is an open-source code quality management system that lets developers manage, track, and improve their software. It has a number of dashboards and reports and is commonly used as the underlying framework for commercialized debt remediation platforms.

● SQUORE — This is a commercial product that covers more than only technical debt. It also predicts remediation costs, system reliability, and efficiency, providing a complete and accurate picture.

● Kiuwan — This SaaS product can be scaled according to enterprise requirements, and deployed in the cloud. The product features simple visuals and easy navigability making it applicable to users even without deep technical knowledge.

However, these tools are now facing a barrage of new demands with the advent of machine learning, AI, and other cognitive services.

The Evolution of Debt in the Cognitive Era

Every leading cloud player, from IBM and Microsoft to AWS and Google, has made some move into the cognitive domain. This is because the cloud provides a wealth of resources and computing prowess, while also making the resulting technology accessible to a broader user base — in other words, the commercialization opportunities write themselves.

Photo by NEW DATA SERVICES on Unsplash

However, AI and ML are notorious for incurring debt, the more efficient they become. These systems are constantly faced with decisions that ask them to choose from a set of actions based on contextual data. Typically, reinforcement learning techniques (trying each option and assessing outcomes until the right one is identified) are very effective in these scenarios. But this leads to an enormous amount of debt, piled up at every learning cycle.

Another factor adding to the problem is the sheer frequency of updates and upgrades, without a publicly visible roadmap. Consider, for instance, these facts:

- At the Google Cloud Next conference 2019, the company announced over 120 changes to its offerings!

- AWS has over 300 podcasts, each discussing tens of updates

- The Microsoft + Azure ecosystem is becoming increasingly more complex with new products and features added every day

All of this indicates the rapid maturity of these systems, making it extremely difficult for enterprises to keep up. To get ahead, it’s vital to take a proactive stance on debt — interestingly, leveraging cognitive as a “guiding light.”

First Incremental, then Holistic: Outlining a Path to Lower Debt

While cognitive technology has led to unprecedented maturity in digital systems, it can also be part of the answer. Using cognitive principles, it is possible to remove the element of human bias when refactoring software platforms. Large volumes of code edits can be completed in a fraction of the time via cognitive-led automation. Here’s how cognitive can help in both incremental as well as holistic models of debt remediation.

1. Incremental — In complex enterprise environments, it can be difficult to garner a 360-degree view of IT usage, pinpointing exact problem areas. Cognitive services can help identify the platform deployed in specific use cases, highlight performance challenges, detect UX gaps, and find out if the code is falling short of the required business functionality. Importantly, the incremental approach can be slightly resource-intensive, requiring a step-by-step roadmap to solve all of the pain points detected.
2. Holistic — You can also leverage the power of cognitive to uncover the business code in action, as well as define its usage frequency levels. Frequently used code can be replaced by an end-to-end SaaS platform, eliminating the entire code debt in one sweep.

Accenture used a cognitive layer in conjunction with CAST to remove debt for a leading retailer. Empowered by cognitive, CAST not only identified quality violations but also recommended an action plan, improving code quality by 47% and reducing technical debt by a staggering 82%!

However, replicating such results without massive investment continues to be a challenge for most enterprises. They focus their attention on business systems and processes, while systems or environments used by employees to delivery services or be productive continues to be neglected, piling up technical debt.

What’s needed, therefore, is a one-stop solution that can overhaul home-grown workplace systems and fragmented productivity platforms, offering a unified “application store” in its place.

The necessity for this idea is reinforced by recent research. Deloitte found that organizations are increasingly faced with –

a) an aging workforce: increasing the need to capture the tribal knowledge of soon-to-retire talent,

b) information overload: employees can’t find what they need, even with technology advances, and

c) the need for speed: employees need to work faster and collaborate more effectively to get their jobs done.

This is why digital workplace solutions offer enterprises a smart and cost/effort-lite answer to solving debt via the holistic model rather than taking baby steps. They combine the various productivity tools, monitoring & governance plug-ins, analytics capabilities, and backend integrations needed to run an enterprise. In fact, by linking to other systems like ERP and CRM, you can build a low-debt enterprise — relying on cloud-based SaaS and cognitive-led improvements, instead of on-premise stored bad code.

In our next blog, we discuss how digital workplaces could go a long way in helping enterprises reduce their existing technical debt, and why digital-born companies should start from the cloud to stay ahead of debt.