If you followed us on LinkedIn on Halloween 2019 you may remember our Karsun “Zombie Slayers” pledged to eliminate a common legacy system scourge, dead code. Six years later our zombie code toolkit has expanded to include artificial intelligence (AI). The ReDuX AI-powered modernization platform and its Blueprint Lenses help our agile teams identify threats early and address their presence as part of the modernization process rather than risking cascading issues after deployment.

Three Zombie Code Threats

Zombie code typically refers to sections of source code in a software project that are no longer used or needed but still exist within the codebase. These unused code segments may have been part of previous versions of the software or were once essential but have become obsolete due to changes in requirements, design, or functionality. This creates system scar tissue that further complicates and constrains modernization efforts.

We identified three threats that still haunt enterprise teams today: dead weight code, living dead code, and vulnerable dead code. 

  1. Dead Weight Code is not related to any application functionality. Its lack of function leads to resource drain. Remediating this non-functional code siphons away developers’ time and energy creating rework and technical debt.
  2. Living Dead Code is dormant code that retains the ability to be activated. Perhaps it supports a feature no longer used by the system. However, this code poses a direct risk because it can negatively impact system output when activated by mistake. In 2019 we recounted Knight Capital’s catastrophic loss of $440 million in 30 minutes. This code woke up a dead zombie method, which incorrectly generated orders, flooding the stock market with trades.
  3. Vulnerable Dead Code consists of old libraries or packages. Vulnerable dead code introduces security risks because these old components have known security issues that are easy targets.

Identifying Dead Code with AI

Over the past six years we have honed our zombie fighting skills building the ReDuX platform. Accelerating modernization with AI, the ReDuX platform begins by providing in-depth legacy system insights. This not only helps Karsun agile teams design better modern systems, it helps us identify zombie risks early so they do not impact development later. 

The first step in our process is building a Blueprint of the legacy system then diving deeper with Blueprint Lenses. Legacy subject matter experts (SMEs) upload documentation, codebases, schema and other materials into ReDuX. Using AI ReDuX agents identify the relationships between system code and components. These agents then create several views, or lenses, to understand the system better. 

For instance, a team viewing their system in ReDuX could view their API calls, scroll through requirements and compare the code citations to ensure accuracy. As part of this process they can also identify orphan code. Thus beginning the dead code analysis process.

Addressing Zombie Code Threats with System Blueprinting

Here’s how we use AI-powered Blueprint agents to address the three types of zombie code.

  1. Dead Weight Code is addressed using AI to generate system requirements. By matching those requirements with specific sections of code we can also determine when code exists that does contribute to meeting these functional requirements. By only extracting functional equivalents, rather than automatic conversion, we can then abandon this type of zombie code in the legacy system graveyard. 
  2. Living Dead Code is addressed with AI by speeding up the discovery process. Now legacy SMEs can focus on the work of understanding how every component of the legacy system interacts with internal and external components. And, by fully generating system business rules and business logic they have a clearer picture of potential unintended consequences of living dead code.

    Teams can then proactively determine the code and system functionality they will extract from the old system to build the new system. More importantly they can do this incrementally, preserving the old system to build new components one functional slice at a time. This provides an additional backstop against living dead code.
  3. Vulnerable Dead Code is also targeted by extracting only the code we need. Further by focusing on functional equivalents we can modernize to optimize. Using AI code companion agents for development, we replace these libraries with modern architecture that is secure by design.

Whether moving away from risky code or outdated components, time is of the essence when trying to outrun zombie code. Providing four times faster delivery with two times less rework ReDuX is an essential tool for our zombie slayer teams. If you’re facing a similar challenge, our AI empowered teams address complex legacy challenges for government organizations and enterprise teams with complex mission-critical applications. Learn more about our capabilities at https://karsun-llc.com/solutions/.