Data Ethics Club: The Tyranny of Structurelessness#
What’s this?
This is summary of Wednesday 23rd March’s Data Ethics Club discussion, where we discussed The Tyranny of Structurelessness written by Jo Freeman (aka Joreen).
The summary was written by Huw Day, who tried to synthesise everyone’s contributions to this document and the discussion. “We” = “someone at Data Ethics Club”. Nina Di Cara and Natalie Thurlby helped with the final edit.
Introduction#
In this week’s Data Ethics Club we discussed the article The Tyranny of Structurelessness, written by Jo Freeman aka Joreen in 1972. The article explores the structurelessness in the development of women’s liberation movements in the 1970s.
“The source of this idea [of structurelessness] was a natural reaction against the over-structured society in which most of us found ourselves, and the inevitable control this gave others over our lives, and the continual elitism of the Left and similar groups among those who were supposedly fighting this overstructuredness.”
We see such structures appearing in more modern organisations, such as academia, startups (often in the tech sector) and, to some extent, Data Ethics Club itself. Whilst the idea of structurelessness to offset rigid hierarchies is admirable, the end result often leads to implicit hierarchies forming that reflect historial power structures. We discuss this and how these so called “natural” structures can often be equally problematic with the added drawback of being implicit and not out in the open.
What did you think of this article? Did you recognise any of the structurelessness that Freeman described in settings that you are part of?#
We really liked the article and thought it had a was good take on organisation theory - a great way of looking at how power is distributed. Some of us were surprised that they hadn’t come across this article earlier. It was particular interesting that we, as a relatively unstructured/informal group, went to read it.
We found ourselves agreeing with the things that the article points out:
It seems necessary to have some rules/structure for larger groups.
Large organisations can find it difficult to organise everything.
Academia is recognisable as a structureless group.
Co-ops say they are non-hierarchical, but the structure was concealed.
The piece makes a simple claim, but makes it in a very broad, less precise way. It doesn’t make it clear where the limits of the idea lie, and there is an argument to be had about what a good structure looks like as an alternative.
People with the most information inevitably become more “powerful” in the system. People with more access to resources (e.g. a darkroom for developing printing) also has this effect. This gets even more messy when we start to assume the natural hierarchy is the best hierarchy without discussing further. One suspects that natural hierarchies just reflect historical power structures.
In technology businesses we see these sort of structureless structures where hierachies are “natural”. What we mean by “natural” is not actually “natural”. We see this all the time when data ethics advocates butt heads with innovators who want less oversight and more scope for innovation.
If you have an informal system, there’s sometimes less scope for collaboration and cooperation. With very unstructured hierarchies you end up with a lot of fluff and a lot less focus. Some kind of structure is inevitable, it’s naive to claim otherwise. It’s up to you to decide on what structure it is.
If your movement is not taking significant action, then it doesn’t matter. If you are trying to get something done, then you need some structure.
Is there any positive use of structurelessness? Possibly, it can make it difficult for bad actors to navigate. But it could be very difficult to structure/document everything. People might find it very bureaucratic and it might prevent them from feeling part of the community and wanting to contributing.
Perhaps there is a distinction to be made between hierarchy and structure.
Academia very much has the structurelessness described! This can make it difficult to establish boundaries with colleagues and leads to implicit ideas. These hidden power structures operate in a way that nobody makes explicit attempts to address. Are you an early-career reserchers experiencing this in academia with career progression entirely dependent on their supervisors - bad supervisor? Tough luck!
For example, in the climate strikes, Greta Thunberg was assumed by the media to be the spokeperson and pushed into the “star” role despite being vocal about others being affected much worse and highlighting campaigners from low-income countries.
Another example is the occupy movement which was set up on non-hierarchical basis, but very much experienced the tyranny of structurelessness described in the article.
How can we make changes to the ethics of data science? Should those changes be cultural or regulatory?#
Maybe the best way of making change is to demonstrate the utility of structure by doing our own and having success. What kind of change are we expecting/hoping for? Regulation can be good when used to get a handle on certain implicit rules emerging. But regulation for the sake of regulation is often counterproductive.
If we’re trying to go shop to shop to influence different individuals in data science, changing the culture one practitioner at a time, this would be a lot harder (but still maybe worth doing) than affecting things on a macro level. In this area we can see the advantage of imposing regulation.
This distinction between cultural and regulatory change is really important. Here at Data Ethics Club, we are more focussed on cultural shifts as opposed to regulatory. Very rarely have we actively sought to influence policy.
If we do have a good idea of our aims (which would often need structure) and those aims are more specific then perhaps we would want to be more focussed on the regulation.
You end up with a feedback loop when having concrete aims. If you can’t see if you’re succeeding then you can’t see when to change the structure you’re working in.
There are many organisations that are still structurelessness. The UK Green Party deliberately doesn’t have a leader. But then again, how are they really doing at the moment? In the 70s it was very much a feature of left wing politics but these days its quite a libertarian feature.
How many open-source software apps have these problems? One of our members used to be involved in such communities and it was very dramatic with lots of squabbles. Perhaps these disputes would have happened less often with more formal structures in place? Linux has good procedures and a set of quality assurance guidance. Some sort of constitution is very helpful!
Our impression from using open-source software is that there are structures there, most contributors (e.g. using GitHub) have some sort of management environment around them.
Many passion-led silicon valley companies (e.g. Google) started out vaguely with “don’t be immoral” but once they start making money, some things change. The system of capitalism that these companies are working within rewards them for dropping their vague ethical statements, past a certain point they almost have to be immoral to continue “succeeding”.
Academia is a bad interaction of a toxic environment of hyper-competitivity that rewards selfishness with the structurelessness and all the problems these entail. We would need to be fighting against the structure that do exist, even though you are only rewarded (or even just sustained) for being selfish. The high-level goal of academia to contribute knowledge is directly stymied by this environment.
We see rules such as research councils not allowing research funding to pay publication fees in the high-end (over-charging) journals. This is an example of embedding elitism by ensuring that only wealthy institutions can publish in the “best” journals.
We could refuse to work with the big publishers and move to open source journal systems. We could take it in-house at universities and commit to green open access etc. But this would have to be en masse across disciplines. The trouble is university management stands directly against that.
Even if we removed a load of university management to make way for this, we would then make the structurelessness worse! We would also need a plan to replace it.
The boundaries between structural and not are quite hard to pin down but there is quite a lot of places where one can see things through the framework of there being quite a lot of structurelessness. Even so, structures still arise.
What structures do we see in data science fields and specifically in data ethics clubs?#
Data science as a field has a different kind of structurelessness to that described in the article, because we have an amorphous community working across academia and industry. How can we organise some more structure specifically for improving ethics in data science?
Perhaps one answer is data managmement and data processing plans and procedures. It would need a plan that is written down (including intentions) that is independent of individuals - no relying on “Bob is the only person that can do [x]”. The problem is that default setting of researchers is quite selfish. Perhaps need to encourage collaboration more broadly? What inscentives could we use?
We found ourselves asking, what can Data Ethics Club do differently? It was interesting to really be intrespective about the structure (and the amount of structure) within the club.
The website has all the links to organisers etc. - is the organisation independent of the individuals? Is there a bit of structurelessness when it comes to suggesting “other activities” apart from the fortnightly discussion topics?
The paper that some of us wrote was a little bit “structureless” in how it came together but how could that have been improved? Can we write down a set of general rules for future “non-fortnightly session” activites? Should we?
One of the perks of our relative structurelessness is that we can welcome more people but we can form off into our own specific groups. The key thing is not excluding people, welcoming alternative viewpoints and be open minded. This is the best way to drive this cultural shift.
Less structure means more people welcome but the group will be less aimed at a goal. This would be an awful structure for a military for example. In less structured groups, we inevitably see a culture emerge (slightly left leaning politics for example). This will be less appealing to certain people.
This bought our discussion to echo chambers; people like to group together with commonality in opinions. There’s only so much we like to have disagreements reguarly. Even if the group is by design not-hostile, it wouldn’t be surprising if people didn’t want to stick around just because they were disagreed with.
We’re all big nerds. There’s an implicit expected education level which also excludes people but we feel the club is quite good at making people ok with asking questions. This stands in contrast to academia.
Data Ethics Club is not absolutely permissive but this is inevitably not possible. The reality is that we have to draw a line where if we can’t agree on a certain point (e.g. people deserve to be treated nicely) then maybe we can’t be permissable to those kind of views. In order to participate in this particular circle, all you need to “turn up and be nice”.
Attendees#
Name, Role, Affiliation, Where to find you
Natalie Zelenka, Data Scientist, University of Bristol, NatalieZelenka, @NatZelenka
Nina Di Cara, PhD Student, University of Bristol, ninadicara, @ninadicara
Huw Day, PhDoer, University of Bristol, @disco_huw
Euan Bennet, Senior Research Associate, University of Bristol, @DrEuanBennet
Ismael Kherroubi Garcia, MSc Philosophy of the Social Sciencesm LSE, @hermeneuticist, @Ismael-KG
Emma Tonkin, Research Fellow, University of Bristol, @emmatonkin
Chantel Davies, MSc Health Data Science, University of Aberdeen, @cjjdavies, @CDavieSTEM
Sergio Araujo-Estrada, Research Associate, Aerospace Engineering Department, University of Bristol