Current air quality visualizations predominantly rely on map-based representations and Air Quality Index (AQI) color coding to show outdoor air quality data. As outdoor air quality concerns grow in importance—due to rising health risks—the need for greater transparency in scientific communication becomes increasingly critical. Showing uncertainty improves transparency, and allows for greater public understanding and informed decision-making. This paper compares sensor uncertainty visualizations along dimensions of context and interactivity. Map-based, time-based, and sensor-only are each presented in static, interactive, and animated formats. These include several novel visualizations such as a map-based hypothetical outcome contour plot. In a crowdsourced user study, we assess the clarity and usability of these nine visualizations, focusing on how well they communicate both the air quality data and associated uncertainties. We find that there are trade-offs for data interaction design: supporting accurate air quality judgments versus understanding data uncertainty. Participants were signficiantly more accurate AQI judgements when using static visualizations, but they had a significantly better understanding of uncertainty when using interactive visualizations. Our work informs visual interaction design to support interpretability, communicate uncertainty, and contextualize data to help with more confident judgments.
Sustainable ICT is an emerging field that integrates ethical, environmental, and technical perspectives, yet there is little guidance on the role of computing professionals in advancing sustainable practice. Despite the increasing focus on sustainability in computing, there is no established framework to define the responsibilities of ICT professionals in this space. Existing research in ICT for Sustainability (ICT4S) has explored the environmental impacts of computing and the role of technology in supporting sustainability goals, but it has not adequately addressed how ICT professionals should navigate their ethical and professional responsibilities in this evolving field. This research introduces a story retelling methodology, drawing from Indigenous and decolonial storytelling traditions, to bridge the gap between sustainability principles and professional practice in ICT. Using an action-based, iterative process, we used storytelling cycles to critically reflect on their roles, responsibilities, and ethical decision-making in sustainable computing, capturing insights through qualitative narratives, and articulating a decolonised professional framework of practice for ICT professionals. This approach provides a decolonial framework that clarifies the role of computing professionals in sustainable ICT, offering practical guidance for future practitioners and contributing to a more inclusive and ethically grounded professional practice.
The advancement and demand for digitalization increase the environmental impact of software systems through the rising energy consumption and CO2 emissions from information and communications technology (ICT) aided by software systems. This necessitates the need for energy-efficient software development. Green coding has emerged as a key practice for reducing the carbon footprint of software systems by optimizing energy efficiency and minimizing computational overhead. However, the adoption of green coding practices among software development professionals in the industry remains limited. This study investigates the state of green coding practices among software development professionals in Finland through a survey with 53 respondents. The goal is to identify software development professionals’ green coding awareness levels, the extent of adoption in practice, the tools used, and the challenges faced in integrating green coding in software development projects. The results indicate a significant gap between the perceived importance of green coding and the actual implementation in software projects. Key challenges identified are a lack of awareness, knowledge, insufficient educational resources, limited tool support, and guidelines for green coding best practices. The findings highlight the need for structured education on green coding, improved tool support, and industry-wide standardization to facilitate the better adoption of green coding among software development professionals in the industry.
ICT increasingly supports healthcare through mobile health (mHealth) technologies. The prevailing narrative in the literature suggests that mHealth improves healthcare efficiency while reducing its environmental footprint. However, there is (1) an important lack of studies assessing the direct environmental impacts of mHealth technologies and (2) a lack of consideration of the indirect environmental impacts resulting from their integration into healthcare systems and society. Hence, the effective potential of mHealth in reducing healthcare’s footprint remains speculative. To help address this gap, this paper explores the premise of a multi-level framework for assessing the comprehensive environmental impacts of mHealth technologies. This is illustrated with the case study of continuous glucose monitoring (CGM) devices, a representative example of mHealth technologies. First, the focus is on their direct environmental impacts, which are estimated through a comparative life-cycle assessment considering several scenarios. Results show that CGM devices exhibit higher direct impacts than the monitoring solution they aim to replace by a factor of 1.1x to 5x, depending on the impact category. Only scenarios that extend CGM device lifetime show significant impact reductions. Second, this paper points out relevant tools such as augmented decision models from health economics and causal loop diagrams from systems thinking to account for the indirect environmental impacts of CGM devices and other mHealth technologies.
Tokenization is a fundamental preprocessing step in natural language processing (NLP) and LLM that influences both model performance and computational efficiency. Although extensive research has explored tokenization strategies in terms of accuracy, their impact on energy consumption remains underexplored. This study investigates the energy consumption and classification performance of three widely used tokenization methods: word-level, character-level, and subword-level on sentiment analysis tasks using BiLSTM models.Our experiments measure the energy consumption of tokenization, model training, and inference, along with classification performance using the F1 score. The results show that subword tokenization is the most energy intensive during tokenization, while character tokenization consumes the highest energy during training and inference. Word tokenization emerges as the most energy efficient method in all stages while achieving competitive classification performance. Despite its higher energy consumption, subword tokenization performs similarly to word tokenization in terms of F1 score. However, character tokenization, proves to be the least effective, exhibiting the lowest classification performance while demanding the highest computational resources.These findings emphasize the need to consider tokenization choices in Green AI research, particularly for energy-efficient NLP applications. Future work should extend this analysis to other NLP models, such as transformers, to provide a broader understanding of tokenization efficiency in deep learning.
Deep learning's growing computational demands have led to considerable energy consumption and carbon emissions. In this paper, we explore on-device learning with sparse backpropagation as a means to improve training energy efficiency. We apply this approach to several CNN architectures (ResNet-18, DenseNet-121, GoogleNet, and MobileNet-V2) trained on popular vision datasets (CIFAR-10/100 and Flowers-102) using an NVIDIA Jetson AGX Orin (64GB) edge device. We present a sparse training algorithm that updates only a fraction of model parameters per iteration, reducing computation in the backward pass. We integrate the CodeCarbon library to measure energy use and associated CO2 emissions. Experimental results show that sparse backpropagation can maintain model accuracy within 1-2% of standard training while reducing energy usage and carbon emissions by up to 30-40%. We analyze tradeoffs between accuracy, speed, and sustainability and discuss deployment strategies for energy-efficient on-device learning. These findings demonstrate a practical step toward "Green AI" by cutting the carbon footprint of deep learning without significantly compromising performance.
The energy demand from data centers is expected to double in the next couple of years. With this in mind, increasing the energy awareness of data centers’ customers and providers is a key step towards realizing potential actions to slow down the energy demand. However, creating energy awareness is not an easy task, as data centers are primarily multitenancy environments where participants share physical and digital resources.We propose an energy accounting system for data centers called FairWatt that aims at allocating energy in a fair way and increasing the energy awareness of stakeholders. Fairness is integrated at design time by means of fairness policies and scenarios, and assessed at runtime through the novel Shared B2 fairness score and Fairness Rulings.
Computing and software engineering (CSE) educators constantly seek effective ways to integrate sustainability and ethics into teaching, bridging theoretical knowledge with practical and socio-environmental challenges. This study examines sustainability-focused hackathons as a pedagogical tool for fostering sustainable mindsets among CSE students. Using qualitative analysis of open-ended survey responses (N=59) and semi-structured interviews (N=20) from participants across three universities, it explores how these events enhance awareness of sustainability challenges and the incorporation of sustainability principles into prototype software solutions. Findings indicate that sustainability-focused hackathons provide a dynamic, experiential learning platform to engage participants in real-world sustainability challenges that demand innovative and responsible solutions. The study highlights their role in developing critical thinking, teamwork, and socio-environmental awareness while strengthening the practical application of sustainability concepts. By bridging theory with hands-on experience, sustainability-focused hackathons prepare students to meet the ethical and sustainability expectations of future software engineers. These insights position sustainability-focused hackathons as a valuable, structured extracurricular activity that complements formal education in sustainable computing and software engineering.
There is a growing interest in the energy impact of computing-related activities, which is expected to increase in the coming years. Modern software development usually relies on Continuous Integration (CI) to support short iterations, where code changes are integrated on a daily basis. The implementation of CI workflows usually relies on build automation tools, (e.g. Apache Maven or Gradle), to automate several activities such as testing or compiling. The large adoption of CI practices raises concerns about the impact of such tasks usually run on cloud environments, where the underlying hardware and its associated energy consumption remain intangible to developers. To better understand the energy footprint related to modern software development, it is essential to investigate the energy profile of the tasks managed by Apache Maven and Gradle. To achieve such goal, we performed a large-scale study with 1167 CI workflows implemented through GitHub Actions and mined from popular Java projects hosted on GitHub. After executing them locally, in a controlled environment, we analyzed in depth the energy profile of 183 355 tasks managed by Apache Maven and Gradle. These tasks represent a quarter of the total energy consumption associated with the CI workflows. We found that tasks from workflows of small-sized projects do not necessarily consume less energy than tasks from workflows of medium-sized and large-sized projects. We also found that testing-related tasks consume the most energy, and that the larger the project, the higher the percentage of energy consumption related to testing. Moreover, tasks of different categories have a different profile regarding energy consumption per task and per unit of time.
Through a detailed breakdown of the metals required to manufacture and interconnect IT components in a High-Performance Computing (HPC) supercomputer, we introduce a novel metric for evaluating its environmental impact: the Metal Usage Effectiveness (MUE). This metric is based on the abiotic depletion potential (ADP), an indicator of the non-renewable material consumption relative to remaining resources. We define this metric as the ADP (expressed in kg antimony equivalents) caused by the extraction of metals required to manufacture and interconnect IT components, divided by the supercomputer computing power (in FLOPS). The MUE, insightful on the supercomputer manufacturing cost relatively to its computing power, complements existing metrics mainly focused on the use phase cost, such as the Power Usage Effectiveness (PUE) or the Water Usage Effectiveness (WUE). It could be a first step toward a more complete Life Cycle Inventory (LCI) of supercomputers and, by extension, data centers. We provide models for estimating copper and gold content in compute hardware components and interconnects of an HPC supercomputer, and apply them to Frontier, the world’s first exascale supercomputer.
This paper discusses to what extent researchers justify their views on the relationship between ICT and sustainability (ICT/S) in papers published at the ICT4S conference. We argue that this would help clarify the nature of the disagreements underlying conflicting views. We first describe various ways in which ICT/S research has a political dimension. We then conduct an analysis of research papers published in ICT4S since 2020, reading abstracts and introductions. From this, we derive 1) a typology of the chains of justification used to relate research questions to sustainability; 2) a model of the problematization process in ICT4S papers; and 3) an evaluation of the extent to which views on ICT/S are explicitly justified in the analyzed corpus. Results show recurring patterns of justification, and a lack of justification of views on sustainability. We conclude by discussing actions the ICT4S community can take to address this.
The semiconductor industry has revolutionized information and communication technologies by enabling unprecedented advancements in computing, communication, and automation. However, alongside this rapid progress, the escalating power consumption and resource demands of integrated circuits have raised critical sustainability concerns. Traditional evaluation figures of merit, such as Power, Performance, and Area (PPA), no longer appear to be sufficient enough to address the environmental impact of these designs. Hence, there is a need for an evaluation paradigm that can assess a design based on Power, Performance, Area, and Sustainability (PPAS). This research proposes a novel Sustainability Framework to allow designers to assess, benchmark, and choose the most efficient and environment-friendly architecture for their product development at the early design phase (i.e., ‘schematic’ and ‘layout’). The proposed framework is applied on a wide variety of foundation cells to demonstrate it’s wide applicability and agility. Foundation cells have been benchmarked and the impact of design choices on sustainability is presented based on different applications like IoT, Automotive, High-Performance Computing, and Portable Electronics.
Industry and academia recognize the importance of computer scientists having sustainability awareness and knowledge. While bachelor’s programs in computer science provide a comprehensive education, the topic of sustainability is typically only briefly covered in a few elective modules. Consequently, most computer science graduates lack awareness and fundamental knowledge of software sustainability. How can we ensure that all graduates receive basic sustainability training without increasing their workload or removing other courses from the curriculum?To achieve this goal, we have developed the novel concept of Sustainable Bytes, a voluntary online course whose sections (the bytes) are linked from the educational material of existing mandatory courses. In this experience report, we present the goals of the course, the key decisions made during the design process, and an overview of its contents. Evaluation and usage statistics from the first instance of the course reveal that only 26 % of first-semester students accessed it and 7 % completed it. Consequently, we share lessons learned to help us and other educators increase student participation in the future.
Comparing the carbon footprints of economic sectors is difficult when relying on dedicated studies based on Life Cycle Analyses and the Greenhouse Gas Protocol. Input-Output Analysis, in contrast, allows yearly homogeneous comparisons based on Multi Regional Input-Output (MRIO) publicly available databases and institutional air emissions accounts.In this paper, we analyze the global carbon footprint of ICT by comparing its average over the 2010-2021 decade to those of sixty other sectors of the economy featured in the Eurostat-JRC MRIO database (called FIGARO). The three basic indicators used in our analysis are direct emissions, total embodied carbon footprint and final demand carbon footprint. They vary steadily from year to year, in response to the variations of final demand volumes. We also consider the two associated ratios of direct emissions and final demand footprint divided by total embodied footprint, and a third independent ratio, the self-consumption ratio of sectors. These ratios appear as structural features of the sectors value chains. We compare the growth rates of ICT to those of the other economic sectors, by distinguishing two periods, the 2010-2019 time span, considered as a "business-as-usual" period, and the extended 2010-2021 time span, which includes the effect of the COVID world crisis. The crisis impacted strongly a few sectors like aviation, boosted a few sectors like health and government, and had a more moderate impact on ICT and many other sectors.
The growing energy footprint of Information and Communication Technology (ICT) services has become a critical environmental concern. While current approaches mainly focus on optimizing existing architectures, this position paper advocates for investigating a more fundamental shift: moving from Cloud-centric to Local-First Software architectures, where data and computation primarily reside on end-user devices. Through a preliminary study, we examine the energy consumption implications of such an architectural shift. We first develop a framework identifying the main potential impacts across the service stack (server, network, and client devices). We then discuss these impacts through three real-world examples, demonstrating how Local-First Software approaches could reduce energy consumption. Our analysis reveals both opportunities and challenges in this architectural transformation. While most impacts could contribute to energy reduction, particularly through decreased server and network usage, some negative impacts emerge, mainly around synchronization and client-side computation. These initial findings suggest that Local-First Software architectures, when appropriately implemented, could significantly reduce the energy footprint of digital services. This position paper aims to stimulate discussion and research in energy-efficient software architectures, laying the groundwork for future empirical studies on Local-First Software approaches in sustainable computing.
Business Process Management (BPM) provides organisations with a systematic approach to model, execute, analyse, and continuously improve their business processes. Traditionally, BPM practices focus on the economic performance of business processes. With the advent of Green BPM, environmental sustainability emerged as an additional performance dimension, allowing it to support organisations in achieving their sustainability goals. However, most existing Green BPM initiatives focus on specific phases of the BPM life cycle or particular environmental performance indicators. As a result, organisations only have a limited understanding of the sustainability of their business processes and lack the means to control their environmental impact during process execution. This paper proposes a Green BPM life cycle that integrates environmental sustainability into every one of its phases without being restricted to specific indicators. An extension of BPMN, one of the standard notations for modelling business processes, is defined to allow process models to capture the environmental performance indicators that should be monitored during process execution, together with target values that can optionally be specified for each indicator to control the environmental impact of the process. An architecture is further presented for integrating these indicators into existing BPM systems by connecting calculator services to the process life cycle events. As a first step towards achieving full life cycle coverage, a prototype modelling tool for the BPMN extension and a prototype calculator service for carbon emissions have been developed.
Many network operators have committed to be net zero by 2040. Yet understanding what does a net zero network looks like, or more generally what would a sustainable digital world look like, is still an open question. This article focuses on understanding what could be the smallest subset of the current digital world that should be kept in a long-term sustainable evolution. The low tech approach, which relies on the principles of usefulness, sustainability and accessibility, is explored to tackle that question. The study focuses more specifically on usefulness, scrutinizing the Manfred Max-Neef’s fundamental human needs, whose work is cited by many low tech thinkers. Through various workshops, some essential digital needs by 2040 are highlighted. These needs would be the ones to fulfill if the digital world would have to get narrower to fit within planetary limits. In this case, the digital technology would only serve to preserve the social fabric.
Current approaches for sustainability analysis face significant methodological challenges in comprehensively assessing impacts across business processes. They are constrained by time-consuming data collection and allocation, inflexible, manually created models, and difficulties in relating impacts to root causes. This paper considers Object-Centric Process Mining (OCPM) as enabler for more dynamic and comprehensive sustainability analyses mitigating these challenges. We explore the integration of sustainability data into the current standard for Object-Centric Event Logs (OCELs) used by OCPM techniques and introduce a prototypical tool to assess impacts based on an OCEL and allocate these impacts to events and objects. We demonstrate how OCPM can support the specification of inventory data, impact assessment, and impact allocation for a semi-automated, multi-perspective sustainability analysis. This exploratory work provides a concrete illustration for sustainability assessment using OCPM, thereby, laying the foundation for further research towards a data and process driven technique to identify and measure company’s negative sustainability impacts more effectively.
This empirical research seeks to identify the scale of Green ICT jobs in the UK and of ICT jobs in Sunset industries, defined as fossil fuel-intensive industries which must come to an end. No previous studies were found in the academic literature though there are useful reports from non-academic sources. Research was carried out using data from the UK Office for National Statistics particularly the Annual Survey of Hours and Earnings (ASHE). The findings contribute new empirical knowledge to the ICT for Sustainability field. The findings show 0.092% of the GB (Great Britain) workforce in Green ICT jobs in the narrowly defined Environmental Goods and Services sector (Greening by ICT), 0.043% in ICT jobs in Sunset industries and 1.259% in Green occupations in ICT industries (Greening of ICT). The paper discusses the methodological restrictions of working with the current Standard Industrial Classification and Occupation codes which often do not distinguish Green industries from Sunset industries. The findings will also be of use to educators in Computing / Computer Science to consider skills and knowledge which students will need for successful careers in Green ICT.
As Information and Communication Technology (ICT) use has become more prevalent, there has been a growing concern about how its associated greenhouse gas emissions will impact the climate. Estimating such ICT emissions is a difficult undertaking due to its complexity, its rapidly changing nature, and the lack of accurate and up-to-date data on individual stakeholder emissions. In this paper we provide a framework for estimating ICT’s carbon footprint and identify some of the issues that impede the task. We attempt to gain greater insight into the factors affecting the ICT sector by drawing on a number of interviews with industry experts. We conclude that more accurate emissions estimates will only be possible with a more more detailed, industry informed, understanding of the whole ICT landscape and much more transparent reporting of energy usage and emissions data by ICT stakeholders.
