Google Drive
Gist
Clipboard
2023-04-20: Wer arbeitet an welchem Kapitel? Bitte eintragen
These guidelines are drafted by the AG Greening DH with the goal of fostering research practices that would be better aligned with the consequences of the climate crisis than it is the case today. They reflect the current state of the art and are conceived to evolve along time.
Over the past years, the awareness of the environmental footprint of higher education and research-related activities has risen around the globe and has given way to a variety of initiatives aiming at addressing this specific issue in the wider framework of the climate crisis and its consequences. While most of these initiatives are not to this day widely disseminated and implemented, we are convinced that each actor in the scholarly community should be actively steering towards a reduction of their impact on the planetary resources and that scholarly associations can contribute to a leveraging effort that has to be carried out collectively.
All human activities have an environmental impact, many of them intertwined in social and political practices which individuals can hardly leverage on a day-to-day basis. All the same, professional practices are embedded in institutional mechanisms that cannot be overturned single-handedly. The goal of these guidelines is to identify areas in which shifts can be initiated and carried out in such a way that they do not rely on the sole initiative of individuals. They reflect the values of our community of practice at large.
In the following, we sketch areas of impact that are particularly relevant to the Digital Humanities im deutschsprachigen Raum (DHd) association and its community of practice, propose immediately implementable measures, and initiate a discussion on long-term challenges.
The Digital Humanities community we address in the following has three domains that are key to its activity: Research, Teaching and Data Curation. In the context of the German-speaking countries, institutions are connected following a federative logic. Still, much is left to be decided by local institutions such as Universities, cultural heritage institutions, and research centers. This concerns in particular infrastructure, project management, research data management, and implementation of algorithms.
Large institutions generally supply their own infrastructure, research data repositories, and software. Even so, they are bound by outline contracts concerning software and office supplies. Data relevant to Digital Humanities research is often stored more than once because no collaborative or interinstitutionally tools are being used. This also impacts backup in the context of the 3-2-1 rule (i.e. every byte exists three times).[1] In terms of infrastructure, organizational structures are clearly impacting data usage and therefore the emissions related to this usage.
Regarding project management, resources are always limited in staff, time as well as money, and projects often have very strict schedules. Environmental issues are generally left out of focus. Fixed-term and project-based contracts also make it difficult to have an impact on structures. Data and knowledge will be generated but often only in the context of projects for which funders still favor innovativity over reusability. To this day, reusable data according to FAIR and CARE principles still concerns primarily project output and hardly project input. The reusable data is de facto not reused.
Additionally, supercomputing such as it is used in research for e.g. machine learning and AI as well as blockchain technology are particularly energy-intensive. This concerns not only the use phase, but also the production and end of life of the technology-based devices and the data transmission activity they rely on (IEA (2022), Data Centres and Data Transmission Networks, IEA, Paris https://www.iea.org/reports/data-centres-and-data-transmission-networks, License: CC BY 4.0, https://www.iea.org/reports/data-centres-and-data-transmission-networks) The Levante supercomputer needs 18 GW as energy per year which costs about 9 Mio. Euro.[2]
In this section we have highlighted a series of systemic issues. In the following, we present impact areas and possible solutions.
Before considering in detail the activities specific to Digital Humanities, it is worth pointing to some factors that lead to high CO2 emissions and apply to work in Academia in general. Every office needs heating and cooling, as well as running water and energy for computer hardware (link to tools for lifecycle assessments especially of computer hardware (production, use/power usage, disposal/end of life)). Hardware production is energy-intensive. While we lack WEEE strategies (HERE ADD REFERENCE TO WEEE SCHEMES - it is somewhere in the toolkit), hardware is often replaced after only a few years of use. Emissions also occur for eating together in a canteen or for the kitchen equipment in the office. Academic activity also requires to attend meetings that might take place somewhere else than the main workplace. Frequent travel to conferences by plane is particularly impactful, but online participation in meetings also requires appropriate electrical and electronic equipment as well as energy. In the context of academic activity, server capacities are used for storing emails, papers, working files and images, etc. The use of communication tools and cloud based collaborative tools or collaboration workspace/solutions, like Nextcloud with Collabora Online or Onlyoffice integration (internal resources) Google Documents (external resources) also requires capacities. The same applies for teaching in academia. As teaching plays an important role in shaping the collective knowledge of society, we identify teaching in academia as an impact area.
In the context of Digital Humanities, large server capacities are needed to store and use datasets and applications. The energy needed for data centers is often underestimated: Only 40% of the energy costs are required for the servers themselves, plus 32% for the air conditioning, 13% for Uninterruptible Power Supply (UPS), 5% for storage, 4% for network like switches and routers, 2% for power distribution and 4% more for other infrastructure like lighting, alarm and monitoring devices. (Schomaker, G., Janacek, S., Schlitt, D. (2015). The Energy Demand of Data Centers. In: Hilty, L., Aebischer, B. (eds) ICT Innovations for Sustainability. Advances in Intelligent Systems and Computing, vol 310. Springer, Cham. https://doi.org/10.1007/978-3-319-09228-7_6, p. 115.)
Transforming non-digitized data to machine-actionable data is key to digital humanities work, which often involves AI. AI system consume energy directly in their development, training and application as well as indirectly through the energy and resources that are needed for the manufacturing, operation, and disposal of hardware.
Application development is one of the areas concerned by Digital Humanities. The process of building digital applications can have a notable environmental impact due to various factors. These include the energy consumed by servers where data is stored, the storage of unused files, the efficiency of the cache policy, the amount of data transferred between servers and apps, and the size and compression of images and videos used in the app. Auto-playing videos and pop-up images in default high-resolution are loaded everytime the concerned page is accessed, generating superfluous energy consumption. Furthermore, the choice of programming language used to develop the app can also significantly impact energy consumption. For example, programming languages such as Javascript may result in higher energy costs due to their reliance on a larger number of dependencies.
The GO::DH Minimal Computing Working Group is a working group of the Global Outlook::Digital Humanities which on the other hand is a community of interest who want to "address barriers that hinder communication and collaboration among researchers and students of the Digital Arts, Humanities, and Cultural Heritage sectors across and between High, Mid, and Low Income Economies" (ADHO (2013). Global Outlook::Digital Humanities (GO::DH) Special Interest Group Created. URL: https://adho.org/2013/01/13/global-outlookdigital-humanities-godh-special-interest-group-created/). Minimal Computing is a method where computing is constrained in its use of hard- and software, power, network capacity etc.
The Digital Humanities Climate Coalition is
Europeana Climate Community
The conference for digitalisation and sustainability - Bits & Bäume has been held twice so far: in 2018 and 2022 at the TU Berlin. In 2023 it will be carried out in North Rhine-Westphalia by the Bits & Bäume NRW subgroup. 13 organizations are part of the sponsoring group, they deal with sustainability, digital and network policy issues.
In 2019 the proceedings of the Bits & Bäume 2018 conference were published. It is available as an open access download, as are individual graphics from the book. The infographics on streaming, resources, open source and the footprint of the conference itself are very interesting, just to name a few.
Demands were developed during both conferences and subsequently published. The 2018 demands can be found in full on the website or in the aforementioned book (p. 137-143). There is a pdf- or html-paper available online for 2022. The demands of 2018 are divided into the following areas: social-ecological objectives in the design of digitization, democracy, data protection, control of monopolies, education, three different aspects of development and trade policy, IT security, longevity of software and hardware. The 2022 paper divides the political demands into five broad themes, each broken down into three to five sub-items: digitalisation within planetary boundaries; global justice and regional empowerment; redistribution of technological power, democracy and participation; justice in digitalisation, sustainable technology design and social issues; protection of digital infrastructure and IT security.
But how to measure the impact of using tools for OCR/HTR models as Transkribus? Maybe by using "Carbontracker": https://arxiv.org/pdf/2007.03051.pdf citation: "In this work, we present carbontracker, a tool for tracking and predicting the energy and carbon footprint of training DL models. "
Setting standards for good practices can be achieved in key areas such as scientific events, academic travel, project management, data curation, and awareness-raising.
Conferences, congresses, academic travel
The pandemic has set new standards in this particular area of academic activity. It led to a greater creativity in the organization of online and hybrid options, recreating even social meetings like coffee breaks and pub meetings online. While these digital tools have undeniable pitfalls (zoom fatigue, lack of social contact, rebound effect due to the purchase of additional devices and the use of energy-intensive internet connections), they not only make it possible to avoid long and carbon-intense journeys, they also make access to the events easier for people with limited mobility and for people from countries with visa restrictions and/or with care responsibilities. Offering hybrid options for academic events should be standard. Next steps in improving these standards include choices of videoconference systems that are not too energy-intensive (maybe encourage audio-based communications) and that guarantee the protection of personal data.
Hybridation cannot solve the most problematic aspect of conferences: the fact that we have reached a point where the wealth of conferences encourages Academics to attend many conferences in many places, year after year (this yearly congress and that yearly conference), if not even by adding new yearly conferences as time goes by. It is to the scientific associations and scholarly societies to reconsider the pace in which they deem salutary to convene their members - such reconsiderations should be encouraged: holding a conference every other year instead of every year, for instance. Academics themselves should be encouraged to less conference hopping and longer stays in the places they visit.
Practical recommendations
Project Management
Project management is embedded in the administrative structure in which the project is hosted. The suggestions below should be discussed with the concerned administration in the conception phase of every project. It strongly shifts the focus from performance to sustainability and in that sense, goes against what still is the modus operandi of many institutions.
When planning a project, every aspect should be considered in terms of its environmental impact: the devices that will be purchased, travels that will be undertaken, data curation and production, publications, communication in and outside the project team. Reviewing their impact does not mean giving up on every environmentally costly item, but consciously balancing environmental cost and scientific benefit. Traveling can be reduced by pairing archival research or team meetings with conferences and by deciding for online options. Resources can be prioritized by looking into solutions that make it possible to reduce storage space. The process of assessing the environmental impact of each area of project management is likely to be a long and difficult one as long as it has not become standard practice.
One key element in reducing the environmental impact of a project is reusability. This means not only making one's own data and tools reusable, but actually build on existing data and resources. Projects that reuse existing data or improve the discoverability of available data and tools should be those at the forefront of research, and will certainly be in a few years' time.
In terms of the electricity use of data-based projects, a few simple measures can be recommended to begin with. Running computers or other machines in standby mode in the night or weekend leads to power usage, that can be easily avoided. In addition to making sure that appliances and lights in offices are switched off when they are no longer needed, eco-power can be used in the office. Hardware should be limited, chosen wisely and used as long as possible, by extending warantees to the longest possible time span. The choice of servers should also be decided upon carefully. The energy usage of large or small and cloud or private data centers differs dramatically. (Hintemann, R. (2015). Consolidation, Colocation, Virtualization, and Cloud Computing: The Impact of the Changing Structure of Data Centers on Total Electricity Demand. In: Hilty, L., Aebischer, B. (eds) ICT Innovations for Sustainability. Advances in Intelligent Systems and Computing, vol 310 (pp. 125–136). Springer, Cham. https://doi.org/10.1007/978-3-319-09228-7_6,.) Research Infrastructures are likely to provide guidance in thisarea.
Practical recommendations
Foster discussions about mistakes that can easily be avoided (particularly for project planning).
Develop an environmentally-aware model for Data Management Plans.
Foster dialogue with research infrastructures.
Data production, curation, computation
Data play an essential role in Digital Humanities research but they are still seldom considered in terms of their environmental sustainability. This concerns formats, storage, transmission - areas that are central to the philosophy of Open Science.
COPIED FROM ANOTHER SECtiON For example it is important to think about what content has to be dynamic and what can be published statically. A static webpage would need less energy than dynamic content.
Another dimension concerns computation.
The approach of Minimal Computing can reduce the energy needed by software through working memory, processors and hard disks by avoiding unnecessarily complex processes. Software can be updated to new circumstances instead of creating new ones. A good ontology design should aim to use as few resources as possible. Monitoring the power usage of a project can create awareness for specific impact areas.
Finally, an important measure is to talk openly about one's own impact on the environment and individual possible solutions with colleagues in order to initiate systemic change and to change structures.
Awareness-raising
Verbandswebseite: schwarzer Hintergrund/weisse Schrift (oder Tag/Nacht-Modus mit Nacht=default) → mögliche Infos allgemein zu dem Thema bei web4nature.de
Von der impact areas-Gruppe:
Building Apps:
When Building Apps, 8 aspects are to be considered.
Overall rule: Only use, what is necessary.
Choosing a Green Hosting Provider.
Defining a efficient Cache Policy. This can be done by adding support both your apps and APIs of If-Modified-Since header field with HTTP 304 Not Modified.
By Optimizing your Media and Images. Implement only what is needed and delete unused media and images. Choose vector images over non vector-assets. Compress your images and remove all useless metadata.
Reduce the amount of data being transferred between your Server and your App. Reducing assets size, only load needed assets, implement lazy load.
Remove unused features from your App: by analysing usage data of each features within your app; remove if less than 5% of your users use a feature
Follow Energy Efficiency best practices for your App. adapt your App’s behavior according to the device power mode; When your app enters background, make sure you don’t let going tasks that don’t need to be done in background.
Optimize your use of Location Services: The unnecessary use of location can prevent the device from sleeping and drain the device’s battery. There are multiple ways to optimize use of location services, but above all, you should only request location updates when your app needs location informations. Stop location services when you don’t need them. Keep in mind that the more accuracy you request, the more energy it’ll consume.
Adapt your use of Timers & Notifications: Timers have an important energy cost, and are often used unnecessarily in Apps. For state changes monitoring for example, apps should respond to adequate events instead of using timers. If you use timers, consider if you really need them or if you could use energy-efficient APIs instead
Local Notification are a good way to fire time-based events, and it also works if your app is killed. If your app requires time-based notifications without external data from the internet, use of Local Notifications is recommended.
For remote notifications, you should always prefer the deferred delivery method for secondary notifications. The deferred delivery method waits for an energy-efficient time to deliver the notification.
https://medium.com/swlh/green-it-a-sustainable-approach-to-app-development-1ef4234faf51
https://www.researchgate.net/publication/323626620_Software_development_methodology_in_a_Green_IT_environment
The field of Digital Humanities (DH) has the potential to undergo significant transformations towards greener and more sustainable research practices. By prioritizing ecological aspects and adopting a holistic approach to sustainability - encompassing social, economic, and environmental dimensions - DH can contribute to a more sustainable digital future. One crucial aspect is the integration of Greening DH guidelines across various levels and contexts. This section outlines a vision of what would be great to achieve within the next five years, envisioning a consistent, self-sustainable research landscape of the humanities in a digital environment.
A fundamental change requires increased awareness and engagement at the individual level. Researchers should develop a deeper understanding of the environmental impact of their research activities and embrace sustainable practices within their personal capacity. Institutions and organizations should provide resources and training to support researchers in adopting greener approaches. By raising awareness, facilitating discussions, and sharing success stories, the DH community can create a culture of sustainability that transcends individual actions and fosters collective change.
DH conferences should embrace the theme of greening and make it an integral part of their program. Conference organizers should actively encourage participants to present papers and workshops focused on green models, algorithms, and methodologies. By sharing best practices and innovative approaches, the DH community can collectively work towards reducing the environmental impact of conference activities, such as travel, accommodation, and resource consumption. A dedicated track on greening DH can facilitate knowledge exchange and stimulate further research in this area.
Within the next five years, a Greening Toolkit/App should be developed and widely adopted by DH researchers. This everyday tool will enable researchers to monitor and report their greening activities, measure their carbon footprint, and compare their progress against standardized parameters. The Toolkit/App should provide practical guidance on implementing sustainable research practices, reducing energy consumption, optimizing data storage, and promoting open access principles. By integrating the Toolkit/App into their workflows, researchers can easily incorporate greening aspects into their daily activities.
To promote the intersection of DH and sustainability, research projects dedicated to exploring the environmental impact of DH activities and identifying sustainable solutions should be initiated. These projects can provide valuable insights into the carbon footprint of different DH methodologies, data storage practices, and computational approaches. Finally, researchers can contribute to reducing the overall ecological impact of DH activities by conducting environmental analyses and implementing sustainable strategies.
Policy development and collaboration across DH-relevant communities are crucial to achieving sustainable research practices. Entities such as NFDI, DARIAH, Leibniz-Gemeinschaft, Fraunhofer Institute, Akademienunion, and Max Planck Institutes should integrate ecological sustainability as a core component of their research agendas. This can be achieved through the establishment of dedicated working groups, the creation of positions for sustainability stewards, and the implementation of certification programs. Collaborative efforts can lead to the exchange of knowledge, resources, and best practices, fostering a shared commitment to greening DH research. Additionally, institutions should establish sustainability guidelines to reshape research, teaching, administration and infrastructure practices.
Research funding agencies, institutions, and associations should incorporate sustainability principles and guidelines into their frameworks. By setting specific sustainability requirements and criteria for grant proposals (which, up to now, focus only data sustainability), research funding bodies can encourage researchers to consider greening aspects in their work from the outset. This is also possible on smaller scales, e.g. university or association grants.
It is essential to regularly evaluate the progress made in greening DH research and identify areas for further improvement. Institutions, funding agencies, and organizations should conduct periodic assessments of their sustainability initiatives and adjust strategies accordingly. This iterative process ensures that sustainability remains a dynamic and evolving aspect.
AG Greening DH
Individuell
Institutionell
DH-spezifisch
Politik
Forschungsförderung
Allgemein
-
Any changes
Be notified of any changes
-
Mention me
Be notified of mention me
-
Unsubscribe
Subscribe