Encountering the data lives of the dead

Catriona Brickel
05/25/2025 | Reflections

Researchers have long recognised the ethical challenges posed by internet research methods like web-scraping and AI-assisted analysis – sometimes known as Big Data methods. Such approaches “have put the various concepts of privacy and autonomy under significant strain” (Bechmann & Zevenbergen, 2020, p.38). This strain is intensified further when considering the ethics of research using the data of the dead. How should researchers understand the right to data privacy, when the subject is dead? Does it matter for research ethics whether their personal data was shared by the dead whilst they were alive, or whether it was shared by someone else after their death? In latter cases, is it correct to identify the dead person as the research participant or is it instead the living individual who shared the data? How could the principle of informed consent be applied in research contexts where the participant is dead? 

These are some of the questions I have considered in planning my PhD project, which explores the creation of digital memorials through charitable giving as a contemporary mourning ritual. I plan to combine data collection methods that are both intensive (semi-structured interviews) and extensive (web-scraping and AI-driven topic models). I want to understand both how users are operationalising concepts of grief and mourning in their online memorials, their subjective interpretations of these, and the factors that influenced their choices about what to include and how to present the identity of the deceased person. Data collected during the Big Data phase of the project may include: the name of the deceased person, their image, their biographical details (date of birth, date of death), their family relationships, their work, their hobbies or their personal qualities. This is some of the information I would expect memorial creators to include within the description of the page, much like a traditional obituary notice. Other data will include the total fundraised on the page, as well as the number of contributions. On the face of it, a ‘Big Data’ approach has several advantages: it minimises the observer effect as data is created within a naturally occurring environment; it avoids the oft-problematic difference between what participants say about their behaviour and what they do; it captures an otherwise inaccessible moment in time (since conducting research immediately following a bereavement may be considered inappropriate); and allows for the identification of patterns at scale. Web-scraping is legal in the UK when conducted for legitimate purposes, and the data of the deceased is not covered by the UK Data Protection Act (2018). For some researchers this is sufficient (Schnell & Redlich, 2019). Yet, for others, legal has never entailed ethical, and so I believe the questions listed above require deeper and more reflexive deliberation. 

Post-mortem privacy

The European Union’s General Data Protection Regulation (GDPR) of 2018, one of the most comprehensive and influential pieces of data protection legislation in the world, unambiguously excludes the data of the dead from its scope (see Recital 27). In its implementation of this regulation, through the Data Protection Act (2018) and the post-Brexit UK GDPR, the UK has similarly chosen to exclude post-mortem data. Other countries, however, have taken up the EU’s invitation to legislate for post-mortem privacy. Perhaps the most robust of these is The French ‘Loi pour une République numérique’, which permits individuals to set directives for how their data may be processed after death and to nominate a responsible individual to execute these. Here, ‘data’ refers to the information they have created online during their life. This might include email accounts, accounts with online retailers (including purchase histories), social media profiles with their associated media, blog posts, and many many more. Importantly, data controllers must comply with the directives or, where none exist, with the wishes of the person acting on behalf of the deceased. Such a strong approach to post-mortem privacy may be founded on the cultural norm that the dignity and wishes of the deceased ought to be respected (Ashley, 2020). It is also reflective of the findings of Harbinja, et al. (2024) that individuals want control over how the data they leave behind continues to be processed. Under the French approach, this desire is fulfilled since the autonomy of the individual transcends death. 

However, this legislation still does not represent a substantial challenge to traditional, non-digitised notions of personhood and agency for which some academics advocate (Arnold, et al., 2018). One of the most radical proposals is detailed in Kim (2001), which suggests that digital information ought to be conceived as a digital self that represents the physical self (albeit with filters), that documents the body’s interaction in the physical world, and that is infused with the mental self (including thoughts and feelings). This digital self persists long after the physical and the mental have died. Could it be then, that the data of the dead really is an enduring person to which rights of autonomy and dignity might be attached? This would have important consequences for research methods, since research with post-mortem data would need to be considered research with human participants and so all the usual requirements for consent, anonymity, confidentiality, among others, would apply. The French legislation does not go so far. Whilst allowing for a form of post-mortem privacy, the rights are still always attached to a living person whether that be the individual documenting their post-mortem wishes for how or the nominated individual responsible for seeing that these are delivered. 

The Italian Data Protection Code takes a softer approach still (Malgieri, 2018), in that the dying may not issue directives about how their data should be treated, but their data subject rights can be exercised by any interested party, such as a family member. Rather than rooting post-mortem privacy in the autonomy of the deceased individual, it is instead rooted in the interests of their living heirs, just as with copyrighted material. This is more in-keeping with the conceptualisation of digital data as ‘assets’ (Harbinja, 2022) which persist beyond death. Under this quasi-property model (Malgieri, 2018), the data of the dead is not a self in any sense and the dead has no right to their pre-death wishes about their data being enacted. Instead, digital data is more akin to financial or other property assets that may be inherited and managed by living others. This means that, unlike under the French model, the post-mortem processing of their data may, legally, be entirely at odds with what the deceased person would have wanted. 

Public doesn’t mean just anyone

Charitable digital memorials are straightforwardly public online spaces, since they are available to everyone with an Internet connection and no registration is required to view a memorial (Sveningsson Elm, 2008). At the point of memorial creation, some platforms allow users to make their memorial private, although they warn this will affect other mourners’ ability to find and contribute to the page. Public is the default setting, and so the data could be understood as available for research. 

As we know from, however, the technical reality of publicity does not always perfectly match the user’s understanding or expectation (NESH, 2019). For example, users may only expect their content to be viewed by those in their network, even if they understand that the data is, technically speaking, accessible to all. There are offline parallels to this too: a conversation between friends occurs in a public space, and is vulnerable to eavesdroppers (even researchers), but the expectation is that others will not be deliberately listening in. Research ethics reflects such an expectation. 

Challenges of informed consent

Gaining consent from the deceased person would clearly be impossible. Thankfully, it is not required under the quasi-property model presented above. Gaining consent from the living heirs, whilst not impossible, would be both “manifestly impractical” (franzke, et al., 2020, p.10) and undesirable, as previous studies demonstrate. One way of gathering consent would be to post on the memorial pages themselves, declaring the presence of the researcher, and asking the creator to respond if they consent for the data to be included. This was the approach taken by Hudson and Bruckman (2004) in their chatroom studies. The results were predictably poor, with only four of a potential 766 participants opting-in to the research. Some of the responses the researchers received also conveyed clear anger and annoyance at being asked, to the extent that the researchers argued the harm of asking was greater than that of the research itself. The danger of such harm seems even greater with digital memorial pages, where such messages could easily be interpreted as spam that spoils the curated, and possibly sacred, memorial space. This approach would also alter the environment from one of natural interaction to something close to an observed social behaviour laboratory. A stated advantage of Big Data methods, that they reduce the observer effect, is thus undermined. 

1. There is some evidence to suggest that users of online platforms are not uncomfortable with their data being used for research purposes. Whilst consent should not be straightforwardly inferred from the acceptance of Terms & Conditions since many users do not read them (Jessop, 2017), a majority of Twitter users (76%) are aware that their data may be accessed by third-parties and that 84% of respondents to their survey were not at all or only a little concerned about their data being used in university research (Williams, et al. 2017). 

2. It is not obvious that online posts, including blogs, social media content or digital memorials, are instances of social interaction. Instead, they might be seen as publications, the use of which in research does not require informed consent from the authors  (Bassett & O’Riordan, 2002). This would treat digital memorials in much the same way as newspaper obituaries, which have been analysed without the need for consent (Hume & Bressers, 2010). 

3. Similarly, web-scraping and other computational methods of gathering online data do not involve the direct participant interaction that characterises human participant research, and for this reason, many Institutional Review Boards at universities have chosen not to hold it to the same standards of consent (Brown, et al., 2024). 

A way forward

I propose the following: the data of the dead that is collected through my project should be treated as the property of their living heirs, which they have chosen to post online. In this case, online does not mean public, even though this is the technical status of the memorial content. I respect that they have posted the data for the use of a specific audience of mourners, of which I am not a part. However, the process of gathering informed consent poses more harm to both the bereaved creator and the success of the research project than proceeding without it. Other mechanisms will be in place to ensure that individuals cannot be identified through the data collected through web-scraping. As with any research project, it remains important to weigh-up the risks to research subjects against the benefits of the research. By developing an understanding of how people are using charitable digital memorials and working with charity fundraisers to develop best-practice guidance in this area, my research aims to ultimately improve the experience of bereaved creators of digital memorials. I take it this social good justifies the use of a minimally invasive computational approach to data gathering and analysis to identify broad patterns in mourning behaviour. 
 

Catriona Brickel is a PhD researcher at the Centre for Death and Society at the University of Bath. Her research investigates the expectations and experiences of bereaved people who create digital memorials through charitable giving. She also seeks to understand how the perspective of mourners giving to charity as an act of remembrance compares to the meaning that professional fundraisers ascribe to this same act. Recently her research interests have expanded to consider the creation of ‘self’ online, the ontological status of the data of the dead and the implications of this for ethical research practice. With five years as a professional fundraiser, including time leading the in-memoriam giving programme at a UK mental health charity, and now a freelance innovation and fundraising consultant, her research is informed by her experience within the fundraising sector and continues to be shaped by practitioners.

Published: 05/23/2025