Developer Attitudes and Practices Towards Optimizing Software Energy Consumption

A survey of 134 professional developers reveals that energy is rarely a first-class concern, with proxy metrics and organizational disincentives creating systematic blind spots.

Max Weber; Alina Mailach; Florian Sattler; Sven Apel; Norbert Siegmund· 2026· DOI 10.48550/arXiv.2608.30527

The core problem

Software fundamentally shapes how efficiently hardware resources are used, yet software energy consumption is seldom treated as a first-class concern in day-to-day development practice. While data centers and mobile devices face growing energy pressures, developers rarely receive direct feedback about the energy implications of their code. This study investigates professional developers' attitudes, decision-making, and development practices related to software energy consumption, with particular emphasis on how energy considerations are recognized, assessed, and acted upon during software development. The authors—Max Weber, Alina Mailach, Florian Sattler, Sven Apel, and Norbert Siegmund—argue that understanding developer mental models is a prerequisite for designing effective interventions, tooling, and educational materials. The central research question is: how do professional developers perceive and act on energy consumption, and what barriers prevent energy from becoming a routine optimization target?

Energy consumption is explicitly considered in only a minority of projects. More commonly, developers influence energy use indirectly by optimizing proxy properties such as execution time and CPU utilization. Responses to scenario-based questions reveal systematic blind spots in this mental model, including cases in which performance improvements increase energy consumption or exhibit no correlation. For example, a developer might reduce execution time by increasing parallelism, which can raise total energy use due to coordination overhead and additional active cores. The survey also identifies organizational disincentives, limited tooling, and educational gaps as major barriers to adoption. Developers reported that energy is rarely part of requirements, code review, or performance dashboards, and that they lack reliable, low-friction measurement tools. The qualitative analysis surfaced a common assumption that faster code is always greener code—an assumption the authors show to be unreliable.

Software fundamentally shapes how efficiently hardware resources are used, yet software energy consumption is seldom treated as a first-class concern in day-to-day development practice. While data centers and mobile devices face growing energy pressures, developers rarely receive direct feedback about the energy implications of their code. This study investigates professional developers' attitudes, decision-making, and development practices related to software energy consumption, with particular emphasis on how energy considerations are recognized, assessed, and acted upon during software development. The authors—Max Weber, Alina Mailach, Florian Sattler, Sven Apel, and Norbert Siegmund—argue that understanding developer mental models is a prerequisite for designing effective interventions, tooling, and educational materials. The central research question is: how do professional developers perceive and act on energy consumption, and what barriers prevent energy from becoming a routine optimization target?

The authors conducted an online survey with 134 software developers. The study combines quantitative analyses with a qualitative open-card sorting of free-text responses to characterize perceptions, practices, and reasoning patterns around energy consumption. Participants answered structured questions about their projects, their optimization habits, and the metrics they track. They also responded to scenario-based questions designed to expose their mental models of the relationship between performance and energy. The open-card sorting method allowed the researchers to group and interpret free-text answers systematically, revealing recurring themes and contradictions. This mixed-methods design is appropriate for an exploratory study that seeks to characterize attitudes and practices rather than to measure energy consumption directly. The survey instrument covered recognition (whether energy is considered), assessment (how it is measured or estimated), and action (what changes are made in response).

The findings suggest that energy-aware development is hindered less by developer apathy than by structural and cognitive factors. The proxy-metric mental model creates a false sense of alignment: optimizing execution time or CPU utilization feels like optimizing energy, but the relationship is not monotonic. Formally, if E is energy, T is execution time, and P (t) is instantaneous power, then E = \int_{0}^{T} P (t) d t . Reducing T does not guarantee reducing E if P (t) increases, for instance through higher clock frequencies, additional cores, or memory traffic. The authors propose three implications: (1) institutionalize energy-aware approaches through visible flagship deployments that demonstrate value; (2) expand research and education on energy-performance trade-offs; and (3) develop practical, developer-oriented measurement and feedback tools that lower adoption barriers. A feedback loop for energy-aware development can be represented as follows: The taxonomy candidates—Architecture, Cybersecurity, Network, Cryptography—are adjacent domains where energy-performance trade-offs also matter, but the study's core contribution is a developer-centric account of why energy remains invisible in practice.

CS practitioners and researchers

Opening member content…