ozziegooen

I'm currently working as a Research Scholar at the Future of Humanity Institute. I've previously co-created the application Guesstimate. Opinions are typically my own.

Sequences

Prediction-Driven Collaborative Reasoning Systems

Comments

EA Relationship Status

Good to know, I wasn't aware. 

Maybe there's some small benefit there, but I'd still be surprised if the gender imbalance on net increased the percentage of marriages. 

EA Relationship Status

First, mostly; it is the speculation. Second, in the Bay Area, there's also a gender imbalance outside of EA.

EA Relationship Status

Quick thoughts:
1) I found this interesting, thanks!
2) There's a significant gender imbalance in the Bay Area scene at least.
3) Lots of EAs I know are in Academia; either getting advanced degrees or doing research. Academia presents a lot of challenges to relationships, I would be curious how our statistics compare to those in Academia.
4) Many of my non-EA friends seem to place great value on relationships, my EA friends less so.
5) I would be surprised if poly actually made that big of a difference. My guess would be that it's not actually that popular in EA. I get the impression that poly sounds radical and has been discussed by some of the key people, so seems like a much bigger deal than it really is.

Expansive translations: considerations and possibilities

Thanks so much Filipe, and I'm excited to see your thoughts on the topic. I think this kind of imagining is highly valuable. 

I don't have much context about you personally, but from my engineering and entrepreneurial experience, my main piece of feedback would be that I get the sense that you think this might be a whole lot easier than I think it would be. Something like what you propose sounds very interesting, but I think this initial proposal would be challenging to do well without tons of money and time. I've seen my fair share of people start far overambitious projects, totally (though predictably) fail, and be heartbroken as a result.

I think it's worthwhile to do the following, but think about them in distinct buckets:
1) Imagine what great systems would be like with near unbounded resources.
2) Figure out what pragmatic steps we can take in the short term to get started.

Both of these are valuable. All of my post was in the former camp, and I would suggest that your post mostly is as well.

Some thoughts on the comment, in the vein of category (1):
 


Translators in the platform could give a score (from 0 to 10) of how good that translation looked for different translation formats

This is a minor point, but I would suggest a system where people rank who good the translation is for individual people (with many defined attributes), instead of trying to bucket things into different categories. Defining the categories is a really messy process that will leave artifacts. This is kind of a classic ML prediction sort of problem.

 

Thus, we could create a market for expansive translations focused on people of different styles.

I think that the current infrastructure for setting up markets in the regular ways are quite mediocre. Another option would be to hire a team of translators working full-time, but monitor and optimize their performance.

---

On the topic of obtaining source data, using new content generation would be very expensive, and I could imagine it being difficult to do well. I think the word for "expansive translators" isn't "translator", but "communicator", for instance, so the people to learn from are the popular communicators, not people with translation experience. 

I think there's already a lot of content out there if you're a bit creative. There are probably tens of thousands of "What is Bitcoin" posts on YouTube and other platforms aimed at a wide variety of audiences, combined with metrics for how popular these are. If you could find ways of learning from those, I would be more optimistic.
 

Our new expansive-translations dot com, ou our new chrome extension.
 

Arbital had features kind of like what I'm suggesting. They identified a need, but found it very challenging to get people to actually do the writing. I suggest checking out the comments from that thread to learn about their experiences.

I'd be enthusiastic about making browser extensions to augment LessWrong in some key ways. It's possible translation could start small; like with the replacement (hopefully with hovers that demonstrate this) of some key words with words one may better know.

Expansive translations: considerations and possibilities

You'd have to have a very clear goal in mind when constructing your professional-context postmodern punk musical Hamlet, and the choice of that goal would make a huge difference to the end product.

 

Agreed. This is a radical definition. 

As translation gets more and more expansive, it becomes more difficult to ensure consistency and quality. But it also leads to a lot of value generation, so can often be worth it.

Hamilton, the Musical, was arguably a retelling / "expansive translation" of the book, which itself was a summary of the original documents. I think most people who originally heard about the idea of Hamilton thought it could never work because of how weird (and expansive) it was. Not only was it presented for people who liked musicals, but it was sort of optimized to appeal specifically to communities of color. It doesn't only translate the older dialects into modern English, but it converts it specifically to the vernacular and musical preferences of parts of Hip Hop culture.

I'm a big fan of that. I'm sure a lot of information was lost along the way, but the value proposition of this dramatic reinterpretation is clear to many viewers. 

Now, not every potential translator may be as talented as Lin-Manuel Miranda now, but the potential is still clear, and in the future we'll have AI to help us.

Expansive translations: considerations and possibilities

Why are modern translations so narrow? What level of nuance would you like them to capture?

By narrow I mean they are aiming to provide language-language translation, but they could hypothetically done on a much more granular level. For instance, a translation that matches the very specific vernacular of some shared Dutch & Jamaican family with its own code words. And there’s no reason the semantics can’t be considerably changed. Maybe Hamlet could be adjusted to take place in whichever professional context a small community would be most likely to understand, and then presented as a post modern punk musical because that community really likes post modern punk musicals. Whatever works.

One could argue that "liberal translations could never improve on the source, and therefore we need to force everyone to only use the source." I disagree. 
 

In translations of poetry - something I have amateur experience with - you have a lot of decisions to make.

Very true! There's actually a lot of discussion of this around Harry Potter, which needed a lot of translations very quickly, and does have a fair bit of wordplay and the like. See here:
https://en.wikipedia.org/wiki/Harry_Potter_in_translation 

I'm sure there must be a far greater deal of similar discussion around Biblical translations. See the entire field of Hermeneutics, for instance. 

That said, I'd note I'm personally interested in this for collective epistemic reasons. I think that the value of "an large cluster of people can better understand each other and thus do much better research and epistemic and moral thinking" is a bigger priority than doing this for artistic reasons, though perhaps it's less interesting.

Expansive translations: considerations and possibilities

No Fear Shakespeare 'translates' the original plays into modern English, which I admit is a helpful idea, but there's a problem with these beyond just the feeling of being juvenile: the 'translations' are often wrong, sometimes blatantly so.

Agreed that translations are often wrong, but I don't think this is reason to give up on them! Translations between languages often fail, but I'm thankful we have them. 

The alternative to translation that I was taught in school about Shakespeare was to just give us the source and have us figure it out. I'm absolutely sure we did a terrible job at it, even worse than that bad translation. I don't remember ever having a lesson on how to translate Early Modern English to Modern English. I think I barely understood how large the difference was, let alone interpreted it correctly.

My knowledge on this topic comes from the Great Courses course "The Story of Language" by John McWhorter. Lecture 7 is great and goes into detail on the topic. 

Some quotes, transcribed here:

"We don't process Shakespeare as readily as we often suppose. With all humility I think there is a kind of mythology - a bit of a hoax - surrounding our reception of Shakespeare as educated people. And I will openly admit that, except when I have read a Shakespeare play - and this is particularly the tragedies - when I go and hear it, cold, at normal speed, I don't understand enough to make the evening worth it. 

"I don't like to admit it - I learned long ago that you're not supposed to say so - but it's true. And even as somebody who loves languages and is familiar with English and all its historical layers, I have seen The Tempest not once, not twice, but three times, never having gotten down to reading that particular play, I have never known what in the world was going on in that play. 

"And I seriously doubt if I am alone. And it's not that the language is poetry. Poetry's fine. It's because Shakespeare in many ways was not writing in the language that I am familiar with. It's been many many centuries and the language has changed. 

"One friend of mine said that the only time he had gone to Shakespeare and really genuinely understood it the way we understand a play by O'Neal or by Tony Kushner is when he saw Hamlet in France because it was in relatively modern French and he was very good at French." 
 

Expansive translations: considerations and possibilities

Thanks for the comment! 

In regards to being able to read "the same thing" as other people; I would of course agree this is one benefit of the current system. Any novel system will have downsides, this is a downside for sure. I think the upsides are far more significant than this one downside at least. Generally we don't mind tutors or educational YouTube courses that are made to be particularly useful for small groups of people, even though these things do decrease the amount of standardization.

we don't have a great track record of using technology like this wisely and not overusing it

Agreed. With great power comes great responsibility, and often we don't use that responsibility that well. But two things: 
1) The upsides are really significant. If "being really good" at teaching people generic information is too powerful to be scary, that doesn't leave us much hope for other tech advancements.
2) Even if it comes out to be net-negative, it could be useful to investigate further (like investigating if it is net-negative).

Multivariate estimation & the Squiggly language

Here's a link to a staging version, though one without explanation. The editor uses vim, so you need to type "i" to begin adding text.

https://5f3973242bc7c000071f50a7--distracted-mirzakhani-b102f4.netlify.app/dist-builder

Multivariate estimation & the Squiggly language

Thanks!

Have you looked into other probabilistic programming languages? While Squiggle could help in the use case you are describing, Stan and PyMC3 may be usable for the case you are describing.

Squiggle would treat that similar to Guesstimate, though also make it possible to write simple functions. 

 

Load More