ThomasSittler comments on Concrete Ways to Reduce Risks of Value Drift - Effective Altruism Forum

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (23)

You are viewing a single comment's thread.

Comment author: ThomasSittler 06 May 2018 05:20:21PM *  7 points [-]

Thanks for the post. I'm sceptical of lock-in (or, more Homerically, tie-yourself-to-the-mast) strategies. It seems strange to override what your future self wants to do, if you expect your future self to be in an equally good epistemic position. If anything, future you is better informed and wiser...

I know you said your post just aims to provide ideas and tools for how you can avoid value drift if you want to do so. But even so, in the spirit of compromise between your time-slices, solutions that destroy less option value are preferable.

Comment author: Yannick_Muehlhaeuser 06 May 2018 05:48:51PM 4 points [-]

There's probably something to be gained by investigating this further, but i would guess that most cases of value drift are because a loss of willpower and motivation, rather that an update of one's opinion. I think the word value drift is a bit ambigious here, because i think the stuff you mention is something we don't really want to include in whatever term we use here. Now that i think about it, i think what really makes the difference here are deeply held intuitions about the range of our moral duty and so for which 'changing your mind' doesn't alway seem appropriate.

Comment author: Joey 06 May 2018 06:11:43PM 3 points [-]

Say a person could check a box and commit to being vegan for the rest of their lives, do you think that would be a ethical/good thing for someone to do? Given what we know about average recidivism in vegans?

Comment author: RandomEA 07 May 2018 11:03:07AM *  4 points [-]

It could turn out to be bad. For example, say she pledges in 2000 to "never eat meat, dairy, or eggs again." By 2030, clean meat, dairy, and eggs become near universal (something she did not anticipate in 2000). Her view in 2030 is that she should be willing to order non-vegan food at restaurants since asking for vegan food would make her seem weird while being unlikely to prevent animal suffering. If she takes her pledge seriously and literally, she is tied to a suboptimal position (despite only intending to prevent loss of motivation).

This could happen in a number of other ways:

  1. She takes the Giving What We Can Further Pledge* intending to prevent herself from buying unnecessary stuff but the result is that her future self (who is just as altruistic) cannot move to a higher cost of living location.

  2. She places her donation money into a donor-advised fund intending to prevent herself from spending it non-altruistically later but the result is that her future self (who is just as altruistic) cannot donate to promising projects that lack 501(c)(3) status.

  3. She chooses a direct work career path with little flexible career capital intending to prevent herself from switching to a high earning career and keeping all the money but the result is that her future self (who is just as altruistic) cannot easily switch to a new cause area where she would be able to have a much larger impact.

It seems to me that actions that bind you can constrain you in unexpected ways despite your intention being to only constrain yourself in case you lose motivation. Of course, it may still be good to constrain yourself because the expected benefit from preventing reduced altruism due to loss of motivation could outweigh the expected cost from the possibility of preventing yourself from becoming more impactful. However, the possibility of constraining actions ultimately being harmful makes me think that they are distinct from actions like surrounding yourself with like-minded people and regularly consuming EA content.

*Giving What We Can does not push people to take the Further Pledge.

Comment author: MichaelPlant 06 May 2018 08:43:29PM 3 points [-]

It seems strange to override what your future self wants to do,

I think you're just denying the possibility of value drift here. If you think it exists, then committment strategies could make sense. if you don't, they won't.

Comment author: ThomasSittler 07 May 2018 09:14:01AM 2 points [-]

Michael -- keen philosopher that you are, you're right ;)

The part you quote does ultimately deny that value drift is something we ought to combat (holding constant information, etc.). That would be my (weakly held) view on the philosophy of things.

In practise though, there may be large gains from compromise between time-slices, compared to the two extremes of always doing what your current self wants, or using drastic commitment devices. So we could aim to get those gains so long as we're unsure about the philosophy.

Comment author: Khorton 06 May 2018 09:48:00PM 2 points [-]

I disagree - I think you can believe "value drift" exists and also allow your future self autonomy.

My current "values" or priorities are different from my teenage values, because I've learned and because I have a different peer group now. In ten years, they will likely be different again.

Which "values" should I follow: 16-year-old me, 26-year-old me, or 36-year-old me? It's not obvious to me that the right answer is 26-year-old me (my current values).

Comment author: Darius_Meissner 10 May 2018 02:16:58PM *  1 point [-]

Thanks, Tom! I agree with with you that all else being equal

solutions that destroy less option value are preferable

though I still think that in some cases the benefits of hard-to-reverse decisions can outweigh the costs.

It seems strange to override what your future self wants to do, if you expect your future self to be in an equally good epistemic position. If anything, future you is better informed and wiser...

This seems to assume that our future selves will actually make important decisions purely (or mostly) based on their epistemic status. However, as CalebWithers points out in a comment:

I believe most people who appear to have value "drifted" will merely have drifted into situations where fulfilling a core drive (e.g. belonging, status) is less consistent with effective altruism than it was previously; as per The Elephant in the Brain, I believe these non-altruistic motives are more important than most people think.

If this is valid (as it seems to me) than many of the important decisions of our future selves are a result of some more or less conscious psychological drives rather than an all-things-considered, reflective and value-based judgment. It is very hard for me to imagine that my future self could ever decide to stop being altruistic or caring about effectiveness on the basis of being better informed and more rational. However, I find it much more plausible that other psychological drives could bring my future self to abandon these core values (and find a rationalization for it). To be frank, though I generally appreciate the idea of 'being loyal to and cooperating with my future self', it seems to me that I place a considerably lower trust in the driving motivations of my future self than many others. From my perspective now, it is my future self that might act disloyally with regards to my current values and that is what I want to find ways to prevent.

It is worth pointing out that in the whole article and this comment I mostly speak about high-level, abstract values such as a fundamental commitment to altruism and to effectiveness. This is what I don't want to lose and what I'd like to lock in for my future self. As illustrated by RandomEAs comment, I would be much more careful about attempting to tie-myself-to-the-mast with respect to very specific values such as discount rates between humans and non-human animals, specific cause area or intervention preferences etc.

Comment author: BenMillwood  (EA Profile) 13 May 2018 08:36:40AM 0 points [-]

It's not enough to place a low level of trust in your future self for commitment devices to be a bad idea. You also have to put a high level of trust in your current self :)

That is, if you believe in moral uncertainty, and believe you currently haven't done a good job of figuring out the "correct" way of thinking about ethics, you may think you're likely to make mistakes by committing and acting now, and so be willing to wait, even in the face of a strong chance your future self won't even be interested in those questions anymore.