Ethics of digital heritage and culture

This reflection of what we mean by ethics and ethical practice when creating, using and sharing so-called digital heritage and cultural data emerged from my work with colleagues on the Towards a National Collection (TaNC) research programme funded by the Arts and Humanities Research Board (AHRC).

Specifically, it is a set of thoughts and responses to the excellent Ethics as Practice Report written by Ananda Rutherford, Anna-Maria Sichani, Katrina Foxton and Sara Perry published in September 2024.

The specific context for this work came out of the lack of consideration and programming around ethics at the outset of the TaNC programme and a realisation across the different projects comprising it that the weight of responsibility towards the people behind the data – and each other – was too great to ignore. One brilliant phrase in the report has stuck with me: the “privilege of obliviousness” cited by the authors from one of the participants in the workshop that preceded and informed the report.

The drive to use new technologies to publish, share and reuse so-called collections data deriving from museum, library and archive catalogues and other documentation, has been done in the name of opening access for the good of all people, but the consequences of doing this, and the manner in which it is done, is seldom given due consideration which is what inspired the authors of this report to press pause and think.

Protecting reputation or people?

In another recent critique of research ethics in a UK university context, the authors of ‘Rethinking Research Ethics in the Humanities: Principles and Recommendations‘ (2023) discuss the standard ethics approval processes that all researchers are obliged to complete when undertaking a new project. University Research Ethics Committees (URECs) oversee this process with specific protocols against which new research is judged.

The authors emphasise the overly bureaucratic and hierarchical nature of this process, a procedure more about mitigating brand and reputation risk and demonstrating compliance for the institution than it is about the wellbeing of researchers and participants. They highlight the problems of how power within these committees is exercised and accounted for.

Throw-away culture

One specific issue both the TaNC Ethics in Practice report and the article above have in common is the urgent problem of unethical post-project legacy planning when it comes to the relationships formed between researchers, institutions and participants. The urgent problem being, there just isn’t any, and there isn’t any accountability from funders either, usually beyond requesting that websites stay online for a certain period.

It’s all about capturing the data, and often not even demonstrating much care around that, as we have witnessed from the myriad, repetitive digital heritage and humanities projects claiming to do ground-breaking things for the first time.

Our research culture is throw-away and amnesiac, in spite of the irony that more past research is easily findable and available to researchers than ever before.

The TaNC Ethics as Practice report highlighted a number of challenges, values and recommendations. These are addressed to all members of TaNC project research teams, including my own in UAL’s Transforming Collections project led by the Decolonising Arts Institute. A number of questions arose for me and my own practices that may be relevant to others and their practices.

I am sharing here in the spirit of spreading the word that the whole digital heritage, culture and humanities ecosystem within and without universities needs to start paying attention and thinking more critically and less hubristically about the real value and impact of their/our work.

Challenges of data ethics

The authors have identified four areas of challenge. I have not reproduced the whole text, please read the report for full context. My questions arising follow.

1. Collection and use of data.
   What does “inadvertent misuse” mean in the context of AI access, development, use and abuse since Tanc started? Here I am thinking of the way in which the concepts of giving, taking and ownership are completely different when it comes to content ‘found’ on the internet, especially with the unreferenced uses of data scraped and downloaded from various sources for AI training.
   
2. Use of technology to “problematise” and recognise bias.
   How might we “own” our biases rather than side-step them? What is unethical is when these remain side-lined or unprioritized within and without a project team––our assumption being that the bias inherent in the data is someone else’s.
   
3. Concern for people.
   How might honesty in relationships and actual negotiation e.g. recompense and recognition be emphasised? See above on URECs.
   
4. Climate impacts.
   Absolutely the single most unethical act is diving headfirst into this kind of work without a Life Cycle Assessment of our uses of these invisible technologies. This needs to happen at project conception and be tested by research funders. If ethical practices are about identifying and mitigating harms, surely this is the single biggest one we are knowingly perpetuating? At least an acknowledgement would be a start so people can start to figure out whether it is really worth it or not?

Ethics of value-setting

The authors proposed a series of five values to underpin ethical practice in programmes like Tanc. Read the report for the full context.

1. Put people first.
   Who? Which people? Something missing on clarity regarding people as human beings vs. institutions which act in a specific communal interest underwritten by a clear hierarchy and set of rules based on an interpretation of legal compliance. This thought arose from the pacific use of words like ‘people’ and ‘communities’ that require detail for the value to be convincing and workable.
   
2. Be transparent.
   What opportunities are there to ensure the ethical framework itself is based on clarity and accountability? Sometimes being transparent exposes when there is nothing to see – this should be ok too but data-driven methods constantly hide and smooth-out gaps and absences; if you are going to use the term “ethical” you need to define it. Another case where a glossary with a set of interpretations or definitions would help to frame the authors’ intent.
   
3. Consider context.
   What are the definitions of terms such as “data affordances” i.e. what is possible vs what is the right thing to do and the wrong (unethical) thing to do? This question arose from the same thought as above, other terms that need some kind of framing are: stakeholders, resources, for example, are stakeholders the same as the people in no. 1?
   
4. Invest in legacy.
   How might we acknowledge that the entire research funding landscape and internal economy/environment within universities is not conducive to this? How many conferences have been recorded and never seen the light of day? We act with legacy in mind but not deed. This particular value needs investment of mentality as well as time and money. In a cyclical existence where you are planning your next research project straight after starting the current one, what hope is there for this?
   
5. Care for climate.
   How might we include humans in environmental ethics? The ethical thing to do is not to separate the two which perpetuates a hierarchy where the needs of the human are always on top (realistically they will be because that is how our species has evolved to be). The use of words here also matters given that both ‘climate’ and ‘environment’ are weaponised and politicised. Specific examples of definition would also help.

Reflecting on recommendations

The authors made three recommendations to convert into action. Read the report for their full context.

1. Establish values.
   What opportunities are there to suggest that funders ask for this piece of work to be done during the application process?
   
2. Embed ethics.
   How might this be achieved in practice? Such as during programming, dataset building, in a funded development phase prior to project commencement? This suggestion arose from thinking about how research teams, especially project leads, set the tone for the work to be done; the red lines that won’t be crossed; the areas likely to cause tensions for researchers and participants and partners; the clarity around who should be prioritised.
   
3. Define and monitor redistribution of funding.
   How might this be achieved if university or funder administration will not allow for it and be entirely inflexible i.e. the reality?

What is missing from practical ethics guidance?

A list of suggestions which might support use of this report:

1. Discussion of why ethics are important, valuable and relevant, not just because “there is a gap.”
   
2. More emphasis on ethical methods and their limits e.g. dialogue and negotiation with partners before, during and after [but no you can only charge us what we say you can and we will only pay you when and how we say].
   
3. How to hold institutions (universities, museums) who are already signed up to various codes of conduct and ethical practice to account on those before creating new ones.
   
4. Specific examples, even anonymised, supporting reflections and findings; cross-references with comparable research programmes.
   
5. Specific discussion of the nature of accountability when proposing a common framework.
   
6. Apparent harms need to be articulated e.g. risk of psychological, emotional, physical harm or deception (people and sources of information) or violation of confidentiality.

Common themes

Some themes recurring in ethical discussions whether they relate to digital research projects or others. These themes stood out for me and ones I feel should be taken forward into new iterations of this guidance.

Tensions. Expected, good, as it demonstrates navigating through a dilemma using dialogue and consensus while also recording unresolved dissent.

Democratisation. Access to digital heritage data. What does this imply in practice? Who has a right, and who gets to decide? This in itself should pose an ethical dilemma.

Conflict between law and ethics. GDPR e.g. the data of living people (community generated content) and “open access” and “reuse.” Who decides “public interest” over rights of individuals to decide what can happen to data they create and publish digitally?

Negotiated ethics. Where the needs and desires of communities (of origin, affinity and interest) especially creators of content outside the research setting are balanced with the needs and ethics of researchers and their institutions (legal agreements, financial arrangements); includes project partners (museums, libraries).

Power (im)balances vs representation. Positive action vs profiling to satisfy institutional and research agendas e.g. Global Majority (SL), diverse representation (TC), under-representation (Tanc)

Application of humane ethical practices outside the generation and uses of “data.”

Disconnect between values-setting and actual practice causing aversion. Values as a burden rather than a tool to assist or unite; the confusion of ethics with compliance.