data-hazards
data-hazards copied to clipboard
Could there be a hazard for not doing the research/work?
It came up from one of the talks I did introducing the hazards to NHS analysts where an analyst made the great suggestion of the hazard of not doing the work. That could, I suppose, fall into General Hazard and it might also be that this falls into the usual justifications for research but because we often don't need to make those justifications for data analysis as analysts in health and care we don't list that anywhere as such. I'd be very interested to hear people's views on this 🙂
Hi Zoe, this actually isn't the first time this has been mentioned to us. Thank you for documenting it here!
My thoughts are:
- Hazard labels currently come from the perspective of having decided to do the project or at least having identified some reasons that you'd want to do it. It doesn't have to stay that way, and maybe if it does, we should state that a little more clearly somewhere. Perhaps the best analogy is: you wouldn't put a hazard label on disinfectant that says: if you don't use this, you might have not kill bacteria and get ill.
- Motivating a project is already something that we are very well trained at (required for making a business case, or getting some grant money, etc). It's a good and useful skill.
- I'm scared that the entire time you could end up thinking of ever-expanding utilitarian thought experiments, such as: if we didn't do this project, what else would we do? Would the thing we'd do instead be EVEN BETTER. Is not doing that project even more harmful, etc.
I'd love to find out if other people have different opinions. I'll share this on the slack and see if anyone else wants to add their views.
I love these counterpoints! Point 3 is a bit like opportunity cost I guess, if we did A we couldn't do B or C for example? Those are very problem-solving approaches to a question rather than an ethical consideration. Really important distinction that is not always easy to see as we spend so much time problem-solving - it's hard to step back and stop doing that!
Coming late back to this as I review issues, but something that has come up whilst doing the JGI Seedcorn projects is that many of them are intended to address Data Hazards. While they still have their own Hazards, we have also allowed a section in the form for people to explain how the work they are doing is aiming for improvements.
I used to feel strongly that we shouldn't let people do that because it drew attention away from the Hazards, but now I feel like allowing people to say what they're addressing is a nice compromise on this area, because it may also allow for a collation of projects working to improve the status quo around the Data Hazard themes, which could be a powerful way of seeing gaps and opportunities for things we want to work on.
Also, in terms of the culture of research around this, I've been doing some work on this project about researcher well-being, and one of the helpful measures for people who work in distressing areas is hope that what you are doing makes a difference. In the current context of AI and data science (over)development, I think being able to say what we are doing to push back might be even more important.
The most important thing (I think) is that people working to address one or more of the labels don't fall into the assumption that because they are trying to improve things, their work can't have the same or other Hazards :D