Good software is green software—and vice versa

In a win-win, the same factors that make software code efficient and effective also make it environmentally friendly.

In the news

Software programmers are asked to work within many parameters: functionality, reliability, performance, time to market, security—the list goes on.

Add another box to be checked: sustainability. With net zero and other environmental goals becoming vital parts of doing business, and “doing business” increasingly dependent on running software-driven digital technologies, the importance of sustainable software is now front and center.

Interestingly, software has two roles in the fight for planetary survival. It can play offense, helping the world adapt to and mitigate climate change by, for example, rapidly improving AI-based climate models and quantum processing.

But it can also play defense. As the Green Software Foundation puts it, creating green software means “reducing the negative impacts of software on our climate by reducing the carbon emissions that software is responsible for emitting.”

What goes into this carbon-reduction component? Energy efficiency, hardware efficiency and carbon awareness. Measuring and tracking each of these variables is a complex task requiring leadership and focus. The Green Software Foundation has created a maturity matrix intended to help organizations assess the greenness—or lack thereof—of their own software.

The Cognizant take

“Sustainable software is a hot topic for sure,” says Ivo Van Der Zanden, a Cognizant senior manager in sustainability. But the factors that go into its sustainability boggle the mind. “Some languages are more efficient than others,” Van Der Zanden notes. “Then there’s the architecture, code efficiency and the number of lines, hosting and auto-scaling, your data storage and retention policies. All this must be taken into consideration.”

Moreover, externalities such as energy supply to the facility (how much of it comes from sustainable sources?) and physical distance from the data center come into play. It’s a complex set of interdependencies.

While AI doesn’t change these factors, it does turbocharge them because of its compute intensity, Van Der Zanden points out. AI may be used to save energy in the long run, “but the training phase requires massive computational resources with large datasets and massive quantities of fresh water for cooling. The initial environmental cost is high.”

In the end, “good”—or what’s often called “elegant” software tends to be good for the environment. This is great news for developers who relish writing code that is efficient, effective and spare code that does everything it needs to do with minimal fuss. This is what green looks like in the software world: Essentially, less is more.