Expansive translations: considerations and possibilities
By Ozzie Gooen @ 2020-09-18T21:38 (+12)
This is a linkpost to https://www.lesswrong.com/posts/PC4yowA2TiRne69iD/expansive-translations-considerations-and-possibilities
TLDR: Language translation is a decent first step for written works, but the ideal looks more like an empathetic personal tutor. There’s a lot to do in-between, both in the near term with human labor, and the longer term with Machine Learning.
(EA Forum Relevance)
I think "futuristic translation" may be of similar epistemic importance as forecasting. Also, the other discussion of translation may be interesting for consideration of the writing and proliferation of EA related content
EdoArad @ 2020-09-19T06:28 (+2)
This is a very interesting idea! I am reminded of Distill's concept of Research Debt. This sounds potentially promising, but I'm not sure I understand exactly why you think of this as being of perhaps similar epistemic importance as forecasting.
First, just to clarify, by "futuristic translation" did you mean any form of expansive translation as is written in your post (which would be developed using future tech or innovations) or something like a specific type of translation that is orientated towards understanding the future? (I assume that it is the former)
The case I see for its importance is basically that it increases our capacity for sharing ideas more efficiently, which can improve general reasoning about complex issues and hasten progress. Is this mostly how you think of it?
One interesting point regarding how promising this is, is that either there would be an economic incentive for someone to create such an innovation or that there won't be enough public interest. I think that, perhaps similar to forecasting, most of the added value we can bring would come if we live in the second world where it would take effort to show how this tech can be used well and when it can be of public interest.
Ozzie Gooen @ 2020-09-19T09:19 (+4)
Thanks! Some responses:
I'm not sure I understand exactly why you think of this as being of perhaps similar epistemic importance as forecasting.
I plan to get to this more in future posts. The TLDR is something like,
"Jugemental forecasting has a lot of room to grow in both research and technology. If it gets really great, that could be really useful for our shared epistemics. It would help us be more accurate about the world. Expansive translations have similar properties."
by "futuristic translation" did you mean any form of expansive translation as is written in your post
Correct. I think that these definitions will require a lot of technology and research to do well, so I'm labeling them as "futuristic".
The case I see for its importance is basically that it increases our capacity for sharing ideas more efficiently, which can improve general reasoning about complex issues and hasten progress. Is this mostly how you think of it?
Yep, that's a good way of putting it.
One interesting point regarding how promising this is, is that either there would be an economic incentive for someone to create such an innovation or that there won't be enough public interest.
It's a common point around EA circles, but I think things are more complicated. Having worked in the tech sector for a while, and read a fair bit around the edges, I think the idea that "technology progress that's useful for industry is an efficient market" has large gaps in it. A lot of really ambitious technological development takes decades to develop and begins in academic institutions long before corporate ones. I think doing great work in this area could require long-term systematic efforts, and the way things are right now, those seem to be very haphazard and spotty to me.
I think it's possible that much of "effective general scientific, academic, and technological progress" is a highly neglected area, even though it seems on the surface that things possibly can't be that bad.
EdoArad @ 2020-09-19T11:20 (+4)
cool, thanks! Looking forward to reading future posts on the matter 😊