Meet Chris Hartgerink, a.k.a. ‘the Bernie Sanders of Open Science’: Scientific Objectivity, Inclusiveness and Socio-technological Frameworks

In this installment of the Road to Open Science, Stefan returns to the roots of Road to Open Science: a podcast. He interviews Chris Hartgerink, a Shuttleworth Fellow and meta-science enthousiast. They talk about Open Science, inclusiveness in Open Science, objectivity in science, and socio-technological frameworks of Open Science.

[00:41] Perhaps somewhat fittingly, you had the honour of being part of the first full digital promotion at Tilburg University. Your dissertation focuses on ‘either understanding and detecting threats to the epistemology of science or making practical advances to remedy epistemological threats’. For those listeners unfamiliar with your work, could you briefly outline what your research was about and what your findings were?

• My main finding was that we wouldn’t communicate science with ‘after-the-fact’ articles if we started today. If redesigned properly, how we communicate research could immediately help address pernicious issues like selective publication and access that threaten the sustainability of science. I propose one such design building on that of Elsevier researchers in 1998, where we shift towards an ‘as-you-go’ communication approach, sharing each step of the research when the researcher feels ready to.
• I meandered to this after I tried to use statistical methods to better safeguard the quality of published research, and finding out that we need to step in much sooner to address issues of data fabrication and questionable research practices.

[05:07] How does your research relate to the Open Science movement? Do you think the name is Open Science is adequate? Is the movement inclusive enough? In the Anglo-Saxon world, the word science is often associated with the social and natural sciences. Some humanities scholars I have talked to say they don’t feel part of the Open Science movement because, in their perception, it focuses too much on on the social and natural sciences.

• I really like the Knowledge Exchange approach to calling it Open Scholarship to be more encompassing.
• Even though the name Open Science has stuck, I regard it as relevant for research. In all the work I do I ask myself ‘How does this apply to the humanities? Physics? Chemistry?’ We are currently testing Hypergraph with theoretical researchers, because it is important to move beyond just the traditional empirical cycle research.
• Moving on to open science as a movement.
• My research sense was definitely inspired by other people in open science, but I always have a hard time understanding what that movement is supposed to be. There are many sides to it, and there are a lot of people saying they are part of that movement that I do not agree with in many ways.
• It is so varied what people subscribe to, but also what people want to achieve. In that sense, I find it difficult to understand whether I subscribe to it as well because lumping myself as part of that movement would immediately make those things reflect it. It is a very porous community in that sense. I find it difficult to judge whether the movement as a whole is inclusive exactly because of it as well.
• There are pockets of inclusivity and pockets of exclusivity, and there are pockets of inflammatory remarks accusing one another.

[10:21] In 2016 you wrote a critical article about the deal between the Association of Dutch Universities and Elsevier regarding open access. In May 2020, Dutch research institutions formed an open science partnership with Elsevier. This means that scientific authors can also publish open access in most Elsevier journals at no extra cost, with some notable exceptions such as The Lancet. How do you feel about this current deal?

• For those unfamiliar with the 2016 saga, this was at the beginning of the long drawn out process that now resulted in this 2020 deal between the VSNU and Elsevier.
• In other words, there is quite the history about this deal.
• Back in 2016, the VSNU was saying in newspapers that they were prepared to launch a full on attack on Elsevier — cancel big deals completely, encourage a boycott for all work. They did not relent, and pretty much said let’s keep talking and no we won’t boycott
• Here we are in 2020. The goals of the VSNU haven’t changed: Full open access publishing. In that sense, this drawn out process has given them success, although I will admit i haven’t read the fine print myself and much more critical reflections and investigations are out there by Sarah de Rijcke, Alastair Dunning, and Sicco de Knecht.
• The landscape has changed nonetheless, since 2016. And we know that Elsevier, or any other self-interested party, is not going to give up something without getting anything in return.
• So what is Elsevier getting in return this time?
• DATA. No surprises there either, because this has been flowing around for twelve to eighteen months as well. But it is critical.
• It puts Elsevier in a more central and monolithic position to capitalise on the surplus data produced by research, reducing competition.
• One might argue that they will share that data for others to build on as well, their collection only being a first step. I refute that by looking at Elsevier’s track record in CrossRef, where their metadata coverage is one of the worst for abstracts and references at 0%
• All in all? I think the people at VSNU and Elsevier have put in a lot of hard work to make this deal. I congratulate them on finding common ground after all.
• Are there kinks in the fineprint? Probably. Is this agreement good for the sustainability of science in the long run? No. This agreement is open science in bad faith because it does not actually progress community owned infrastructures.

[19:23] One of the methods you use in your own research is data mining. The combination of data mining and Open Science can be sensitive due to privacy concerns. A common slogan of Open Science is ‘Open as possible, closed if necessary’, but with increased data triangulation options it might seem as if the ‘closed if necessary’ clause will be increasingly important, somewhat negating the whole idea of open science. How do you see this friction?

• I am one of the most privacy aware people in my network. I read privacy policies A to Z and reject a lot of services because of it. I am that person who asks you to install Signal on your phone because I don’t have WhatsApp.
• So when I say I research methods in and of themselves are not privacy invading (they can still be problematic for other reasons), I say that with due consideration. In that sense, I take data mining to be just as concerning as statistical regression or correlations — not at all.
• I do understand privacy concerns surrounding the data that go into these methods and how those are shared.
• Whom is the data about?
  • Is it a person? Or not?
• Who “owns” the data?
  • The collector or the person who the data is about? Ownership is key.
  • Who decides to share it? Does someone decide to share something about someone else?
• What are the considerations to share?
• Researchers are not trained to ask these questions thoroughly and mindfully.
• But I also think the need for raw, individual person data is not necessary in a lot of cases. It is not a binary question whether you can share, but more a question of at what level can you share.
  • For statistical regressions or more complex models, it is perfectly okay to share summaries of the data in the form of covariance matrices without sharing individual person data. Of course there are many considerations, like in the case of missing data, and if we have better public documentation of those steps, I think we can alleviate a lot of these issues thoroughly.
• Data triangulation is a serious concern, and I’d like to return to data ownership for that issue. Cuz if somebody gets to decide FOR someone else whether their data gets shared, it sets up a paternal system.
• We need to give people the most direct agency to decide whether they want their data shared or not.
• That means giving them complete insight in what data is being stored, and individual level control about sharing or not. I think ultimately, researchers don’t need to be the ones who own the data.

[30:08] You have been called the Bernie Sanders of Open Science, which you seem to embrace. How do you see the concept of objectivity in science? Should scientists take normative stances? If so, when? How do you navigate the thin line between improving science and advocating personal political-philosophical preferences?

• Everything’s political.
• Objectivity in science is the result of a political process in and of itself. That objectivity is constructed, just as facts are constructed at the end of the day.
• After all, objectivity is only approximated by our agreement of subjective experiences. Whose voices are listened to and heard in that process? Or maybe more importantly, whose voices aren’t listened to and excluded?
• Scientists by definition take normative stances. You always take a normative stance, except that it doesn’t feel that way if it is the one that has been accepted already.
• So I encourage people to take informed normative stances, and explain why they take those stances. You need to be cautious to not take stances because you took them before, because dogma definitely is the bane of
• Which is also why I embrace being called the Bernie Sanders of Open Science; I think the struggles he highlights, those of inequality and injustice, also affect how we do science. It is not without reason the oligarchic publishers are problematic; it is no surprise that black people are being maltreated in academia; it is no surprise the veneer of racism is still quite alive in science.
• I encourage everyone to reflect on what their personal political-philosophical stance is in these systems, because it is a first step to rearticulating how your everyday life looks like.

[35:09] Can you tell us a bit about the Hypergraph project you’ve been working on?

• Hypergraph is the first implementation of the “as-you-go” research communication I talked about earlier.
• By september, you’ll be able to share your research “as-you-go” with your peers, for free, and with complete access. Data, code, videos, text — all are part of the process.
• All of the information shared goes into a common knowledgebase (that we call the peer-to-peer commons) and Hypergraph interfaces with that.
• By communicating this way, we can always retrace the previous steps of whatever we’re reading in that moment. That means that if you’re reading results, you can go back and look for the data underlying those results. You can reanalyse them and add your own results, and others can do so with your work.
• It is all about creating an open cooperative space for research to evolve.
• Ultimately, we want to go beyond just sharing content but also providing researchers with valuable tools to make their everyday lives easier. So much work is overhead that detracts from the research itself, and that’s ineffective in times where we’re already strapped for time with all the pressures surrounding teaching, managing, and publishing.
  • An example is that we will provide tools to find replication materials in methods, and ultimately even provide ways to order replication packets with the materials needed.

Relatedly, the Hypergraph superficially reminded me of Blockchain, inasmuch that they both seem to be digital distributed ledgers. How do you feel about the hype surrounding the usage of Blockchain in Open Science and related areas such as peer review and scientific publishing?

• I previously wrote “Concerns About Blockchain for Science” after I researched the usage of blockchain for scientific review and publishing.
• I also organised a session during the Open Publishing Festival on “Is blockchain living up to the hype?“
• I personally feel like the usefulness is primarily overblown. Most of the issues are governance related in science, and blockchain does not necessarily change that governance. I found with the available projects, it was too much a “let’s throw tech at it” approach, which is not productive.
• The biggest part of progress in science is about collective action, and tools won’t create that. Hypergraph won’t create that in and of itself either, which is why I focus much more about building it as a community space to collectively address issues than as a tool to address issues.

Thank your for reading and listening to this blog entry. If you want to stay updated, do keep an eye on this website, the Open Science Community Utrecht newsletter and follow us on Twitter.