Hmm, I’m somewhat atypical in usually maintaining the heuristic that one should generally be transparent even in cases where doing so seems like it might hurt one’s popularity or receptiveness to one’s ideas. So I’m not that worried about journalists quoting out of context, and I think it’d be a big cost to hamper honest discussion just because of that concern. (The cost in terms of worse discussion probably exceeds any benefits in terms of not turning people off?)
Many other EAs disagree with me. They make an interesting point that even if one’s ultimate goal is to help people discover their own idealized preferences, bridging inferential gaps slower can make it more likely that people will eventually update to “weird” positions that they will then endorse on reflection.
Yeah, it is a real toss-up here. Well, I mean, you know I perceive lots of merit in your above position, but I also see merit in the arguments of EAs who disagree with you.
I used the phrase “maximally transparent” above, as distinct from “transparent”. I’ll unpack that so it’s clearer what I mean and why. Transparency in expressing ideas in writing, at least, can be thought of as following the spirit and/or letter of Grice’s maxims.
Being “maximally transparent” would be following all of Grice’s maxims as much as possible. The ‘compromise’ statement I gave above, describing the role of far-future animal altruism, follows the maxims of quality and relation, but doesn’t fully satisfy the maxims of quantity and clarity (“manner” at the link; I prefer “clarity”, as well as being taught it that way). We’re volunteering as much info as necessary (as an intro to the topic), but not all the info which would be appropriate (as an intro to the topic).
So, we’re satisficing instead of optimizing for transparent communication on the prediction this will prevent the costs of journalists or pundits using our own words against us at a later date.
Is this transparent “enough”? Is it better to optimize for transparent communication, because we’re overrating the real costs of blowback? Ought we optimize communication for maximum transparency at all times because journalists and pundits who want to hurt EA will find a way to do so regardless? How would we make the implied prediction both explicit and testable? I don’t know. Those are questions I’d love to see answers to. The above ‘compromise’ position is as far as I’ve gotten.
Hmm, I’m somewhat atypical in usually maintaining the heuristic that one should generally be transparent even in cases where doing so seems like it might hurt one’s popularity or receptiveness to one’s ideas. So I’m not that worried about journalists quoting out of context, and I think it’d be a big cost to hamper honest discussion just because of that concern. (The cost in terms of worse discussion probably exceeds any benefits in terms of not turning people off?)
Many other EAs disagree with me. They make an interesting point that even if one’s ultimate goal is to help people discover their own idealized preferences, bridging inferential gaps slower can make it more likely that people will eventually update to “weird” positions that they will then endorse on reflection.
Yeah, it is a real toss-up here. Well, I mean, you know I perceive lots of merit in your above position, but I also see merit in the arguments of EAs who disagree with you.
I used the phrase “maximally transparent” above, as distinct from “transparent”. I’ll unpack that so it’s clearer what I mean and why. Transparency in expressing ideas in writing, at least, can be thought of as following the spirit and/or letter of Grice’s maxims.
Being “maximally transparent” would be following all of Grice’s maxims as much as possible. The ‘compromise’ statement I gave above, describing the role of far-future animal altruism, follows the maxims of quality and relation, but doesn’t fully satisfy the maxims of quantity and clarity (“manner” at the link; I prefer “clarity”, as well as being taught it that way). We’re volunteering as much info as necessary (as an intro to the topic), but not all the info which would be appropriate (as an intro to the topic).
So, we’re satisficing instead of optimizing for transparent communication on the prediction this will prevent the costs of journalists or pundits using our own words against us at a later date.
Is this transparent “enough”? Is it better to optimize for transparent communication, because we’re overrating the real costs of blowback? Ought we optimize communication for maximum transparency at all times because journalists and pundits who want to hurt EA will find a way to do so regardless? How would we make the implied prediction both explicit and testable? I don’t know. Those are questions I’d love to see answers to. The above ‘compromise’ position is as far as I’ve gotten.