This is a special post for quick takes by Pablo. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Protesting at leading AI labs may be significantly more effective than most protests, even ignoring the object-level arguments for the importance of AI safety as a cause area. The impact per protester is likely unusually big, since early protests involve only a handful of people and impact probably scales sublinearly with size. And very early protests are unprecedented and hence more likely (for their size) to attract attention, shape future protests, and have other effects that boost their impact.
Off the top of my head, I think it could be especially useful to:
figure out how not to alienate the employees at these places (based on protesting different values, or using humour, and generally refining the message)
Refine a list of asks, other than just demanding a pause,
Evaluate whether the protests are even net useful in their current form, and
Figure out whether the protests are so small as to be ineffective, and if so, what to do about that.
Scott Aaronson just published a post announcing that he has won the ACM Prize in Computing and the $250k that come with it, and is asking for donation recommendations. He is particularly interested "in weird [charities] that I wouldn’t have heard of otherwise. If I support their values, I’ll make a small donation from my prize winnings. Or a larger donation, especially if you donate yourself and challenge me to match." An extremely rough and oversimplified back-of-the-envelope calculation suggests that a charity recommendation will cause, in expectation, ~$500 in donations to the recommended charity (~$70–2800 90% CI).
I don't know if there is a designated place to leave comments about the EA Forum, so for the time being I'm posting them here. I think the current homepage has a number of problems:
The 'Community Favorites' section keeps listing the same posts over and over again. I don't see the point of having a prominent list of favorite posts in the home page that changes so little. I suggest expanding the list considerably so that regular visitors can still expect to see novel posts every time they visit the homepage.
[Note: in light of Oli's comment below, I'm retracting this bullet point.] The 'Latest Posts' section sorts posts neither by karma nor by date; rather, it seems to rely on a hybrid sorting algorithm. I don't think this is useful: as someone who checks the home page regularly, I want to be able to easily see what the latest posts are, so that when I go down the list and eventually come across a post I have already seen, I can conclude that I have seen all posts after it as well. With the current sorting algorithm, there's no way for me to insure that my browsing session has exhausted all the posts seen since the previous session.
I find it hard to understand the meaning of the 'Community' category. The description says that it consists of "posts with topical content or which relate to the EA community itself". But that description also draws a contrast to "Frontpage posts, which are selected by moderators as especially interesting or useful to people with interest in doing good effectively." This contrast suggests that 'community' posts are simply posts that haven't been curated, as opposed to posts with a focus on the EA community. In other words, there are two separate distinctions here: that between curated vs. non-curated posts, and that between posts with a community focus vs. posts with a focus on other aspects of EA; and it is unclear in terms of which of these the 'Community' category is defined. To make things even more confusing, the 'Community Favorites' section doesn't appear to employ the term 'Community' in either of those senses; indeed, the term seems to be used with the opposite meaning of "non-curated", since the "Community Favorites" consists of a list of "all-time greatest posts".
I think Pablo is right about points (1) and (3). Community Favorites is quite net-negative for my experience of the forum (because it repeatedly shows the same old content), and probably likewise for users on average. "Community" seems to needlessly complicate the posting experience, whose simplicity should be valued highly.
The 'Community Favorites' section keeps listing the same posts over and over again. I don't see the point of having a prominent list of favorite posts in the home page that changes so little. I suggest expanding the list considerably so that regular visitors can still expect to see novel posts every time they visit the homepage.
I don't know what settings the EA Forum uses, but on LessWrong we filter this list to only show users posts that they have not clicked on. I expect eventually we will also add additional functionality to stop showing posts when a user has seen a post a number of times and has repeatedly decided to not click on it.
The 'Latest Posts' section sorts posts neither by karma nor by date; rather, it seems to rely on a hybrid sorting algorithm. I don't think this is useful: as someone who checks the home page regularly, I want to be able to easily see what the latest posts are, so that when I go down the list and eventually come across a post I have already seen, I can conclude that I have seen all posts after it as well.
I would use the All-Posts page for this use-case. I check the forum only about once a week, and I like only seeing the best posts from the last few days on the frontpage, as opposed to all of them. There are also a variety of other problems with sorting things strictly by recency, one of the biggest ones is that it basically fails completely as an attention allocation mechanism, and causes people to be a lot more aggressive with downvotes, because everything is competing in a much more direct way for frontpage space (i.e. any user creating any post will take up frontpage space, independently of whether the post is well-received or is of broad interest).
Ah, I hadn't noticed the 'All-posts page'. That addresses my needs, thanks. And point taken about the drawbacks of recency sorting. I retract that part of my comment.
We've tried to make the "All Posts" page fairly visible (it appears on the homepage sidebar, and can be found in the drop-down sidebar from any post on the Forum).
Do you have any suggestions for what might have made it more visible to you? You strike me as a careful reader, and someone who spends a lot of time viewing EA content, so if you hadn't noticed this page, I'd guess that many other readers aren't seeing it either.
I think in this case the fault lies entirely with me, given the number of different ways one can see a list of all the most recent posts.
(My original bullet point also mentioned that sorting by recency seemed like a preferable way to display posts anyway, and for this reason I concluded that this should be the default display. But in his reply Oli mentioned some important drawbacks that I had overlooked, so I no longer believe this.)
Pablo already replied, but FWIW I had the same irritation (and similarly had all posts pointed out to me by someone else after complaining to them about it). I think in my case the original assumption was that 'latest posts' meant what it sounds like, and on discovering that it wasn't I (lazily) assumed there wasn't a way to get what I wanted.
I don't have a constructive suggestion for a better name though.
I now realize I had already seen that post. Perhaps my memory is faulty, or perhaps the distinction between Frontpage and Community is not one that sticks. A couple of comments:
In general, I think it's not a good sign if a central feature of a website isn't self-explanatory, but instead requires the reading of a detailed explanation. Moreover, in this case the explanation is buried in a post that new users are unlikely to encounter (and at least some old users are apt to forget). But, more fundamentally, I just don't see a compelling reason for categorizing posts in this complicated manner to begin with. Why not just have a "curated" category to promote posts that stand out in the relevant dimensions, like LessWrong does? Or dispense with the idea of "promoted" posts altogether, and let the karma system do the work. Keep it simple, stupid.
Regarding the categories: We’ve been thinking for a while about whether they should remain on the Forum. We hoped early on that they would improve the reading experience for people who were primarily interested in research rather than community topics (or vice versa), but we’re unsure of the extent to which this has happened.
For now, these are internal conversations, but I wouldn't be surprised if we made a decision on this soon after an upcoming feature (tagging posts) becomes available to users (no date on this yet). It’s possible that using tags like “research”, “events”, or “community culture” will obsolete the broader categories we currently have, in which case the distinction could disappear; it's also possible that we'll find ways to make use of broader category pages that aren't covered by tags.
I find it hard to understand the meaning of the 'Community' category. The description says that it consists of "posts with topical content or which relate to the EA community itself". But that description also draws a contrast to "Frontpage posts, which are selected by moderators as especially interesting or useful to people with interest in doing good effectively." This contrast suggests that 'community' posts are simply posts that haven't been curated, as opposed to posts with a focus on the EA community.
I agree that these descriptions aren't very clear. Here are replacement descriptions that will go up in a future update (date still uncertain):
Frontpage: "Posts relevant to promising cause areas or cause prioritization, which may be useful even to people who aren't involved with the EA community”
Community: “Posts about the EA community (including jobs, events, and announcements)"
These descriptions aren’t perfectly mutually exclusive, because I'm not sure how to simplify the standards of the explanatory post to that extent, but we’ll continue to apply those standards as best we can for close calls.
To make things even more confusing, the 'Community Favorites' section doesn't appear to employ the term 'Community' in either of those senses; indeed, the term seems to be used with the opposite meaning of "non-curated", since the "Community Favorites" consists of a list of "all-time greatest posts".
We considered many titles for the “Community Favorites” section, including “Recommended”, “From the Archives”, “Vintage”, and “Posts you may have missed”. However, all of the titles had various issues:
Because we didn’t personally select which posts appear here (it’s just unread, high-karma posts), “Recommended” is inaccurate (no one is doing any recommending other than an algorithm)
“From the Archives” and “Vintage” imply that the posts are fairly old, even though a high-karma new post could easily appear in the list.
“Posts you may have missed” is too long to work on mobile
Even though “Community Favorites” overlaps with the “Community” category, we chose it because it accurately expressed that these posts were highly upvoted by the Forum’s community, and because it was the only option approved by all moderators.
After more thought, we’ve decided that we will change the name to “Forum Favorites” (this change won’t be visible immediately, but will be part of a future update).
I don't know if there is a designated place to leave comments about the EA Forum
Your shortform works if you like. Also:
• Use the Intercom box in the bottom right part of your screen ↘️
• Email Aaron or me (the CEA staff who work on the Forum)
• Send Aaron or me a private message on the Forum
The 'Community Favorites' section keeps listing the same posts over and over again. I don't see the point of having a prominent list of favorite posts in the home page that changes so little. I suggest expanding the list considerably so that regular visitors can still expect to see novel posts every time they visit the homepage.
I basically agree with this. For context, I view it as a way to discover posts that you missed if you took a break from the Forum for a while. If you have seen the same ones over and over, it's no longer serving that purpose. I like Oli's suggestion about hiding posts you've seen N times; you could also do a weighted randomization of the posts. This feature isn't a development priority for the Forum right now, but if LessWrong makes it we’ll port it. It also might make it to the docket after our current batch of improvements.
Here's a useful heuristic whose name isn't widely known: the principle of stylistic consistency. As Drexler writes,[1]
In judging people and bodies of work, one can use stylistic consistency as a rule of thumb, and start by checking the statements in one's field. The mere presence of correct material means little: it proves only that the author can read and paraphrase standard works. In contrast, a pattern of clearcut, major errors is important evidence: it shows a sloppy thinking style which may well flow through the author's work in many fields, from physics, to biology, to computation, to policy. A body of surprising but sound results may mean something, but in a new field lacking standard journals, it could merely represent plagiarism. More generally, one can watch for signs of intellectual care, such as the qualification of conclusions, the noting of open questions, the dear demarcation of speculation, and the presence of prior review.
The heuristic has been endorsed by some eminent thinkers.
When I was young, most teachers of philosophy in British and American universities were Hegelians, so that, until I read Hegel, I supposed there must be some truth to his system; I was cured, however, by discovering that everything he said on the philosophy of mathematics was plain nonsense.
Sir, it is not unreasonable; for when people see a man absurd in what they understand, they may conclude the same of him in what they do not understand. If a physician were to take to eating of horse-flesh, nobody would employ him; though one may eat horse-flesh, and be a very skilful physician.
Protesting at leading AI labs may be significantly more effective than most protests, even ignoring the object-level arguments for the importance of AI safety as a cause area. The impact per protester is likely unusually big, since early protests involve only a handful of people and impact probably scales sublinearly with size. And very early protests are unprecedented and hence more likely (for their size) to attract attention, shape future protests, and have other effects that boost their impact.
Off the top of my head, I think it could be especially useful to:
I thought the recent Hear This Idea podcast episode with Ben Garfinkel was excellent. If you are at all interested in AI governance (or AI safety generally), you probably want to check it out.
PSA: Only about 4% of the world's population uses the date format month/day/year.
Not to trigger you, but I think by now it's probably more than 4%. The reason being America's soft power :/
Scott Aaronson just published a post announcing that he has won the ACM Prize in Computing and the $250k that come with it, and is asking for donation recommendations. He is particularly interested "in weird [charities] that I wouldn’t have heard of otherwise. If I support their values, I’ll make a small donation from my prize winnings. Or a larger donation, especially if you donate yourself and challenge me to match." An extremely rough and oversimplified back-of-the-envelope calculation suggests that a charity recommendation will cause, in expectation, ~$500 in donations to the recommended charity (~$70–2800 90% CI).
I don't know if there is a designated place to leave comments about the EA Forum, so for the time being I'm posting them here. I think the current homepage has a number of problems:
I think Pablo is right about points (1) and (3). Community Favorites is quite net-negative for my experience of the forum (because it repeatedly shows the same old content), and probably likewise for users on average. "Community" seems to needlessly complicate the posting experience, whose simplicity should be valued highly.
I don't know what settings the EA Forum uses, but on LessWrong we filter this list to only show users posts that they have not clicked on. I expect eventually we will also add additional functionality to stop showing posts when a user has seen a post a number of times and has repeatedly decided to not click on it.
I would use the All-Posts page for this use-case. I check the forum only about once a week, and I like only seeing the best posts from the last few days on the frontpage, as opposed to all of them. There are also a variety of other problems with sorting things strictly by recency, one of the biggest ones is that it basically fails completely as an attention allocation mechanism, and causes people to be a lot more aggressive with downvotes, because everything is competing in a much more direct way for frontpage space (i.e. any user creating any post will take up frontpage space, independently of whether the post is well-received or is of broad interest).
Ah, I hadn't noticed the 'All-posts page'. That addresses my needs, thanks. And point taken about the drawbacks of recency sorting. I retract that part of my comment.
We've tried to make the "All Posts" page fairly visible (it appears on the homepage sidebar, and can be found in the drop-down sidebar from any post on the Forum).
Do you have any suggestions for what might have made it more visible to you? You strike me as a careful reader, and someone who spends a lot of time viewing EA content, so if you hadn't noticed this page, I'd guess that many other readers aren't seeing it either.
I think in this case the fault lies entirely with me, given the number of different ways one can see a list of all the most recent posts.
(My original bullet point also mentioned that sorting by recency seemed like a preferable way to display posts anyway, and for this reason I concluded that this should be the default display. But in his reply Oli mentioned some important drawbacks that I had overlooked, so I no longer believe this.)
It's also linked from the frontpage with the "Advanced Sorting/Filtering" link.
Pablo already replied, but FWIW I had the same irritation (and similarly had all posts pointed out to me by someone else after complaining to them about it). I think in my case the original assumption was that 'latest posts' meant what it sounds like, and on discovering that it wasn't I (lazily) assumed there wasn't a way to get what I wanted.
I don't have a constructive suggestion for a better name though.
Note this post on the Community / Frontpage distinction.
I agree that the term 'Community Favorites' is confusing as well 😵
I now realize I had already seen that post. Perhaps my memory is faulty, or perhaps the distinction between Frontpage and Community is not one that sticks. A couple of comments:
In general, I think it's not a good sign if a central feature of a website isn't self-explanatory, but instead requires the reading of a detailed explanation. Moreover, in this case the explanation is buried in a post that new users are unlikely to encounter (and at least some old users are apt to forget). But, more fundamentally, I just don't see a compelling reason for categorizing posts in this complicated manner to begin with. Why not just have a "curated" category to promote posts that stand out in the relevant dimensions, like LessWrong does? Or dispense with the idea of "promoted" posts altogether, and let the karma system do the work. Keep it simple, stupid.
Regarding the categories: We’ve been thinking for a while about whether they should remain on the Forum. We hoped early on that they would improve the reading experience for people who were primarily interested in research rather than community topics (or vice versa), but we’re unsure of the extent to which this has happened.
For now, these are internal conversations, but I wouldn't be surprised if we made a decision on this soon after an upcoming feature (tagging posts) becomes available to users (no date on this yet). It’s possible that using tags like “research”, “events”, or “community culture” will obsolete the broader categories we currently have, in which case the distinction could disappear; it's also possible that we'll find ways to make use of broader category pages that aren't covered by tags.
Thank you for all of these suggestions, Pablo!
I agree that these descriptions aren't very clear. Here are replacement descriptions that will go up in a future update (date still uncertain):
Frontpage: "Posts relevant to promising cause areas or cause prioritization, which may be useful even to people who aren't involved with the EA community”
Community: “Posts about the EA community (including jobs, events, and announcements)"
These descriptions aren’t perfectly mutually exclusive, because I'm not sure how to simplify the standards of the explanatory post to that extent, but we’ll continue to apply those standards as best we can for close calls.
We considered many titles for the “Community Favorites” section, including “Recommended”, “From the Archives”, “Vintage”, and “Posts you may have missed”. However, all of the titles had various issues:
Even though “Community Favorites” overlaps with the “Community” category, we chose it because it accurately expressed that these posts were highly upvoted by the Forum’s community, and because it was the only option approved by all moderators.
After more thought, we’ve decided that we will change the name to “Forum Favorites” (this change won’t be visible immediately, but will be part of a future update).
Great, thank you!
Hi, thanks for the comments.
Your shortform works if you like. Also:
• Use the Intercom box in the bottom right part of your screen ↘️
• Email Aaron or me (the CEA staff who work on the Forum)
• Send Aaron or me a private message on the Forum
I basically agree with this. For context, I view it as a way to discover posts that you missed if you took a break from the Forum for a while. If you have seen the same ones over and over, it's no longer serving that purpose. I like Oli's suggestion about hiding posts you've seen N times; you could also do a weighted randomization of the posts. This feature isn't a development priority for the Forum right now, but if LessWrong makes it we’ll port it. It also might make it to the docket after our current batch of improvements.
Thanks for the reply. I think it's totally fine for you to deprioritize this suggestion—not very important.
Here's a useful heuristic whose name isn't widely known: the principle of stylistic consistency. As Drexler writes,[1]
The heuristic has been endorsed by some eminent thinkers.
Lord Russell:[2]
Dr Johnson:[3]
Eric Drexler, ‘Abrupt Change, Nonsense, Nobels, and Other Topics’, Foresight Institute, 1987.
Bertrand Russell, Unpopular Essays, New York, 1950, chap. 1
Boswell, Life of Samuel Johnson, London, 1791