Normative Uncertainty and Information Value
Riley has a Bachelors of Economics and a Research Masters in Philosophy from the University of Adelaide, focusing on decision making under severe uncertainty. He also supports research about how to do as much good as possible at the Global Priorities Institute, Oxford University.
What was your thesis topic?
My thesis is about normative uncertainty, an approach to decision making that takes seriously uncertainty about which moral theories are best (moral uncertainty) and ways of making decisions are best (decision theoretic uncertainty). It’s related to MacAskill, Bykvist and Ord’s book on moral uncertainty.
I attempt to find out when information is valuable for agents who face decision theoretic uncertainty. I do in fact find some mathematical conditions, though I’m pretty sceptical that these are really applicable to actual decision making.
I give a possible framework for evaluating information under decision theoretic uncertainty, based on an earlier model by Philip Trammell. I really like Trammell’s model and think it’s an amazing contribution, but I end up arguing that one of the principles it’s based on is suspicious.
One of the key things that Trammell is doing is solving a problem of recursive uncertainty (known as the “regress problem”). This problem comes from the many different things we could be uncertain of, consider the following thought process:
- We initially face level 1 uncertainty: we are uncertain about what we should do. We try to come up with a way of deciding under this uncertainty.
- We then face level 2 uncertainty: we are uncertain about the approach we came up with to deal with level 1 uncertainty. We try to come up with a way of deciding under level 2 uncertainty.
- We then face level 3 uncertainty: we are uncertain about the approach to making decisions that we came up with to deal with level 2 uncertainty. We try to come up with a way of deciding under level 3 uncertainty.
This could continue infinitely, unless we are really sure about what to do at some level (unlikely). I think our project is to decide in a way that is both plausible and respects every level of our uncertainty. I argue for my own solution, on which you should always weigh the costs and benefits of further deliberation, and only continue up to the next level if it seems worthwhile from your current perspective. I defend this solution as better than alternatives in the literature.
I also review the literature on whether or not we can compare value between different ethical views, or between different people. I think this literature shows that we can only sometimes compare value, and that our ability to compare is context dependent. I also give a novel argument for this position.
I argue that we can compare value between decision theories.
I would recommend reading this longer summary if you’re interested. Most of the thesis is quite technical, but the summary is pretty accessible.
What do you think the stronger and weaker parts of your thesis are?
I think my review of the comparability literature makes it quite clear that we can only sometimes compare moral theories. I think further research should explore which situations and combinations of theories are comparable.
The thesis is fairly original for a masters, but that was partly due to the structure of masters degrees in Australia and research funding I received that allowed me to extend my research time a bit. One claim that seems important and original is that we can compare decision theoretic evaluations. I don’t really understand the scepticism about this (but maybe you could write a reply article if you do!).
It is also interdisciplinary, drawing heavily from both economics and philosophy.
While original, it’s likely not the best work or the last word on many of these topics. In particular I am pretty pessimistic about my results related to information value. I think that it is likely that the way I’ve set things up isn’t the best way, and that someone could come up with something much better – or systematically compare the options. I also think that even if it really is a plausible approach, my results aren’t readily applicable, and it would take some work to make them useful to real decision makers. In general I think that spending more time on applications to real decisions would have been a good use of time.
In what ways do you think your topic improves the world?
To be very honest, I often find it hard to see the direct usefulness of any single piece of theoretical research. Perhaps the best case I could honestly make would be as part of the large overall contribution of research of this kind. It seems plausible that work in ethics and decision theory is impactful as a whole: for examples consider the influence of early utilitarianism on the British Empires decision to abolish slavery, or Peter Singers influence on effective altruism and the animal rights movement. It’s often hard to see which contributions are particularly useful immediately after publication, but it’s easy to see the general trend. (I don’t feel like I’ve dug into these cases enough to know if they ultimately hold water but it seems plausible that theoretical research in ethics and decision theory has created a lot of value overall.)
This argument could be strengthened by considering that the last few centuries have seen humanity’s power increase immensely, and we may one day be able to mould the very atoms of the universe into whatever we choose. We might also hand off most of our decision making to autonomous systems. Either way, making our decision about how to use our cosmic endowment wiser, and improving the values we ultimately align our decision-making systems with (hopefully we can align them), will be extremely valuable, much more valuable than previous work seems to us now. My work, then, could be seen as a tiny unit of progress that contributes to this broader contribution of value.
In what ways have you changed your mind since writing it?
I’m not sure if much has changed, though I’m not sure all of my arguments will carry when fully fleshed out (especially the original argument for weak comparability and my solution to the regress problem).
What recommendations would you make to others interested in taking a similar direction with their research?
- Find out how you work best – for example, I find it super helpful to have a paper copy and go to a cafe (leaving my electronics behind) to study.
- Australia has pretty good masters programs. They usually let you research whatever you want for two years, often with funding for both course fees and scholarship. It’s like a mini PhD. Consider an MPhil (especially if you can get a scholarship). In philosophy, hiring is incredibly competitive, so most universities have good staff. ANU, ACU, Sydney, Adelaide, Wollongong, Melbourne and Monash are all worth considering depending on your research interests.
- Do the Effective Thesis program!
- Don’t neglect your mental and physical health, it’s essential.
- Try to find yourself someone who can check in on your progress and help you figure out how to solve your problems [as described in this podcast].