Hide table of contents

A snap vote on features that people would like to see. 

Upvote based on some sense of what the community should want, agreevote (the tick) to adjust for you personally.

There is a feature request thread, but I sense that old requests are getting lost. Hence, this. 

Edit, I'm shifting my suggested voting rules. Agreevotes should now just be about the reasoning of the comment, as usual. They were basically entirely correlated. It was a trial and I was wrong.

66

0
0

Reactions

0
0
New Answer
New Comment

52 Answers sorted by

Ability to see search results in chronological order

This is not quite what you're asking for, but see the related feature in this timely PR!

6
Lorenzo Buonanno
2y
THANK YOU, this is life changing!
6
Habryka
2y
Man, I am excited to finally see a decent search page. It's been sad for so long.
2
Nathan Young
2y
Is there a way that users could vote on issues on the forum and then the could be ranked according to that on the github?
2
Habryka
2y
Alas, I don't think so. The Github API here has historically been kind of a pain to work with. Not sure whether it's impossible, but I can't think of a way of doing this.
2
Nathan Young
2y
What if there was a post that was automatically populated with all open github issues and people could up and downvote them with karma? Do you think that would create value for you?
4
Charles He
2y
In a deep sense, the features and projects developers work on is both socially/individually determined, as well as in the direct control of a project manager or lead. It seems really implausible that people could directly vote on them.  A major issue is the quality of information (internet crowd). If we ignored this and the upvotes somehow perfectly reflected the omniscient, ideal preferences of the community (and things along the lines of Arrow's theorem weren't an issue), it would still be different than what would be worked on for idiosyncratic/personal/planning reasons.  If this would be done, it would be by the leads/manager getting this would-be feedback...and that's sort of what Lizka/Ben West/JPA/Habryka are doing, in this very thread.
2
Habryka
2y
Oh god, no. We have like 400 open Github issues, most of them random backend bugs that nobody should ever really engage, and that are quite unclear to people without context of software engineering experience.
2
Nathan Young
2y
What if any with the label "frontend" went here? Orgs I've been a product manager for want this information. And the EA community is higher signal than most. So I'm curious what your posture towards it is if it was easy to obtain.
3
Rockwell
2y
@JP, I'll confess I don't know what PR stands for in this context and I'm confused from the linked site if this is something I can install in my browser!
8
JP Addison
2y
Sorry about that! When a developer opens a Pull Request on Github it means that they have ~completed work on a feature or a bugfix, and are submitting their code to be pulled to the main branch of the code. So what you're seeing is a feature that EA Forum engineer Sarah Cheng has built for the Forum, to be released soon.
1
Rockwell
2y
Awesome, thank you for the explanation! I learned something new today :)
2
Nathan Young
2y
I like linking pull requests to things! I think the community could be much more aware of how this works and try and support it.
2
Yonatan Cale
2y
[This might be totally wrong and please remember I have much less of an intuition about UX than you, but] This seems to be focusing on visual components like "tabs" instead of the thing which would be intuitive to me, which I'd call "search capabilities". I'd personally be happy to have all the searchable content of the forum pushed into some DB that works well with search, and use some open source search-UI that someone else built for that DB. This PR seems like rebuilding a UI like that which probably exists already, no? [remember my disclaimer!] [I can find a specific UI that seems nice to me if that would help]
2
Yonatan Cale
2y
Tried adding a Google Custom Search, you can try it here and see what you think: https://cse.google.com/cse?cx=21fe57e08e3c5467c (spoiler: I don't really like it)
1
Evan R. Murphy
2y
TIL that the code for LessWrong/EA Forum is open source.

@mentions

Any time you do this it sends a notification to the person. 

A huge part of my model of the value of the forum is about increasing knowledge density. Saying "look here" or "talk to this person". Ideally we'd be tagging each other all over the place, so we can know who is working on what and avoid duplication.

Thanks Nathan! This one's on our radar.

Nudge people to write shorter posts or at least to summarize them. 

For example: Have the default text of a new post be "TL;DR:"

How about all posts have a summary box where anyone can write it and then the most upvoted summary gets shown.

2
MichaelDickens
2y
Could work. My intuition is that there would be no good way to integrate the list of summaries into the UI, and almost nobody would want to read the list of summaries. The work of writing summaries seems analogous to the work of editing Wikipedia, so perhaps the UI for viewing and writing summaries could be similar to Wikipedia talk pages.
2
ChanaMessinger
2y
ooooo
2
ChanaMessinger
2y
I really really like this
1
Noah Scales
2y
Mmm, that might work when everybody understands what's written, but when most people don't get it, it won't help to have 10 bad summaries all voted on.

Having a TL;DR box at the beginning of the posts sounds amazing

Inline comments

Some comments should be written at the end of a piece. But some make sense embedded in context. I'd like both, with inline comments initially hidden. I'd turn them on in my own settings.

This sounds conceptually useful, but also like a UX nightmare.

We[1] have an exploratory approach under development. No promises for delivery however. It is in fact pretty hard to get right.

  1. ^

    "We" = LW

Annecdotally, a lot of busy people report visiting the Forum many times per week. Some report habitually over-consuming it. One of the biggest costs of the EA Forum might be that it reduces focussed work hours of the most talented people in the community.

So, I propose "EA Forum When Ready".

Features would be something like:

  1. Block schedule: block forum from loading during a period you schedule.

  2. Hide the post list on homepage by default. You have to press "Show posts" to see them. (You can still search or browse subpages, just not see the homepage list).

  3. Time limit: you can spend X mins per day on Forum, after that it is blocked.

HackerNews famously has a similar feature called "no procrast(inate)" mode.

Screenshot-20221006-031905.png

Haha, I just suggested this to the forum team today

You can do this easily enough with external tools. I use the stayfocused plugin on Chrome for this.

4
peterhartree
2y
Very few people will actually do this unless prompted, and it can be setup in 1-3 clicks. I'd guess less than 10% of those who should.

Lizka mentioned she goes to https://forum.effectivealtruism.org/search by default which does some of this, which I'm going to try to start doing

I have to admit I'm also addicted in this way, though I'm not sure I'd enjoy any of these features.

Polls in comments.

Any comment can have an 8-answer poll. Users click an answer and then get to see the results. 

A button to allow for multiple votes instead. 

Is this currentl possible in posts?

2
Nathan Young
2y
Technically, but people don't like it when I do it. 
2
ChanaMessinger
2y
How do you do it? Elicit embed?
7
Nathan Young
2y
I am a poll answer
4
ChanaMessinger
2y
This isn't "in posts"
2
Nathan Young
2y
I am a different poll answer

Thanks, added to our list.

2
Nathan Young
2y
Is your list github issues? Or is it stored somewhere else?
1
Sarah Cheng
2y
Yup, it's listed in our github issues - feel free to comment there with more detail on what you'd like to see!
2
Nathan Young
2y
Is there any way we can up or downvote these issues?

Matchmaking tools

Make connections with other forum users.

Calendly-like booking where I can request to talk to people and it automatically books calls with them in their next "free for EAs" slot.

3
Agustín Covarrubias
2y
There's precedent to this, particularly the EA Pen-pals project, which ended up being dropped along with EA Hub.

Everyone is anon mode: browse Forum without seeing any usernames in posts, comments, etc.

Could you explain how this feature would be valuable to you?

3
Imma
2y
I have the tendency to decide to read an article based on the name of the person who posted it. If you have a good reputation, you are more likely to get read. See a previous comment of mine.
1
Sharang Phadke
2y
Thanks, we've recorded this in our backlog!

Search among a specific user's posts/comments

Filter on regular search would suffice e.g. "author:Name"

2
Guy Raveh
2y
Agreed

Get more wiki to happen:

Wiki additions can get upvoted and karma: update -- this is already the case

New and upvoted wiki entries show up in feeds. update -- this can happen, to some extent .

Revising my comment: consider ways of making the wiki more prominent and better integrating it into the forum (I realize this is now kind of vague)

This exists. Wiki additions can get karma. Wiki additions above a certain size get shown in Recent Discussion on the home page.

2
Nathan Young
2y
But never in the main ranking, right? Which is presumably where most people look?
2
david_reinstein
2y
Thanks, I will revise my suggestion.

As in stackexchange: as you are posting it suggests posts that have similar content (based on ML I guess). … suggesting you make your post a comment on their post instead. (Or an edit/addition to their post if this feature is also enabled.

I suggest many stack exchange features would be good 

Users can set posts to be editable by any logged-in user.

Or it's easy to request edits, or to submit them without them being seen until approved.

2
Nathan Young
2y
I think putting edits as branches that appear as comments and can be approved by the original author.

Emoji replies, like Slack

(Nathan, you can use this for polls!)

Nothing counts mode: browse the forum without seeing karma score and comment count.

I feel like this already exists on LW? Maybe we'll see it soon.

1
Evan R. Murphy
2y
LessWrong has no such feature built-in that I'm aware of. Though there are external tools people have created to hide karma.

You can currently do this on greaterwrong.com and also there is a browser plugin that does this, I think.

1
Evan R. Murphy
2y
That's true but hiding karma is an important thing for reducing bias (e.g. people being inclined to upvote something because it's already upvoted) and mental health. So it would be a great thing to have has a built-in feature (and perhaps even the default) so that it's accessible for many more users. Karma scores are still important when it's hidden for sorting content by quality/popularity, it's just more of a behind-the-scenes thing. (Thanks for all your great work on these forums.)

I personally actually think that voting towards the karma score you think a comment should have, instead of voting independently, produces a better aggregate judgement (I.e. I downvote 200 karma posts if I think they should only have 150 karma, and I upvote comments at -5 if I think they should only be at -2). This of course requires seeing the current vote, and so am a bit hesitant to have too large of a population to a different thing instead (and to have a UI that makes it feel like people are supposed to vote independently).

That said, we don't have any super clear guidance on this topic, and there is a decent amount of disagreement in the user base on what style of voting (independent or relative) is better.

Link articles in text by typing [[ and then it brings up a search box of all forum articles. 

Turns out this already exists using "@"

And tags / wiki entries.

2
Pablo
2y
+1. Maybe a different prefix could be used, e.g. '##' for Wiki entries and '#' for posts.

This already exists. But it's # instead of //

2
Nathan Young
2y
Wild. 

"Listen to this post" button which links to an automatically generated AI narration, and a human narration if available.

I use this a lot. Google assistant does this for me, and the Nonlinear Library podcast has AI narrations of all posts above some karma threshold.

A human "epistemic spot checker" who tries to find flaws in verifiable claims made in EA Forum posts

This would be awesome!

I could imagine some people not liking this as it might make the forum a more intimidating place to post to. I imagine that the kind of person who says this would have less of an issue with:
* people opting in by making the post a non-personal post
* people opting in by adding a "check-me" tag

Another mechanism could be for the forum team to pay out $25 bounties when people falsify claims (as a way to incentivise this kind of checking), and maybe take some of the authors karma.

One worry is that if there's negative feedback for making clearly falsifiable claims, people will stop making clear claims. Another worry is that the service is inaccurate, like sometimes happened with Facebook fact-checkers.

It exists on LessWrong so I guess it will come soon.

Polls in shortforms 

Anyone can summarise: anyone can write a summary of anyone else's post, and the original author can choose whether to feature one of them at the top, visible by default.

The other summaries are accessible at the top of post but with a click required to view them. The summaries accumulate karma in usual way, low karma ones are hidden.

There's also some mechanism by which the author is prompted to write their own summary, either inline on the post or in a separate box, to compete with other people's summaries.

(Spitballing here, can easily imagine coming to think this is a pretty bad idea all things considered)

More powerful "customise the homepage" options.

In particular: follow a list of topics then see a list that contains only new posts (and classic posts) on those topics.

A way of sharing and upvoting external content along the lines of HackerNews.

There are linkposts already but people don't post linkposts in the way people do on HN.

Linkposts could be improved by differentiating them more from normal posts and allowing some form of embedding (or otherwise not copying their content).

3
Vaidehi Agarwalla
2y
Yeah some kind of nudging or possibly a requirement to add more context would be good.

YouTube and TikTok.

The literal boss of CEA online suggested a video format and there's basically an EA TikTok account with 50K followers

Let's go!

4
Charles He
2y
Hmm, yes that post contains a video, you're right. I can't seem to replicate this, from limited effort. I (very quickly) tried drafting a post—it wasn't obvious how to embed a YouTube video.  Also, I'm thinking about comments and I can't find an easy way to do this in comments. It's possible that this embedding can be achieved by a different editing mode (selected by the EA Forum settings). If you can show this, that would be great.   You're right, Youtube Links turn into videos (my links weren't working, maybe they were too long)

I'll also add that I think we should take a stance against TikTok in particular because (I assume) it makes our data accessible to the Chinese government, who I think are probably the most dangerous actor on the world stage.

2
Charles He
2y
The vibe with TikTok is bad, but the concrete risks and causal linkages with harm seems unclear, this article seems balanced?  https://www.wired.com/story/tiktok-nationa-security-threat-why/
2
Guy Raveh
2y
I read most of the article and skimmed the rest. I think it's short-sighted and misses a few points. 1. It's very US-centric. like the quote "In a series of responses this summer both to lawmakers and the public, TikTok has staunchly maintained that it does not and would never share US user data with the Chinese government.", which only refers to US data. So the Chinese authorities can still easily mine non-US data. Which is... a lot. 2. It believes the statements of Bytedance too much. They say they would never share US data with the Chinese authorities, but do you really buy it, when they could just share it and compel TikTok inc. to hide it someway? 3. It mentions that China is notorious for stealing data, but I think an app like TikTok is a whole new level. Imagine you were Robin Hood, and you discovered that instead of putting all this energy into assaulting rich guys in the forest, you can somehow open a bank and convince all the rich people's kids to open accounts and pay fees. You'd take that for sure? 4. This is actually my main point - it underestimates the risks of data mining, which I think mostly come from advanced AI. You could, for example, use vast amounts of data to create models of American society to plan stratgic moves, or to find outliers and recruit spies, or to use existing AI to find people who criticise the CCP, etc.
2
Charles He
2y
These considerations seem important. Some of them seem deep and general. Because I don't have depth and because some involve bigger views of the world, it's hard for me write a reply that would be really useful.  One reason I gave a US-centric article, is because that country is actively opposed to the other. So to me, a moderate take from a mainstream publication that seems clueful, carries more weight to me.

The literal boss of CEA online suggested a video format and there's basically an EA TikTok account with 50K followers

should be:

The literal boss of CEA online suggested a video format and made an EA TikTok account with 50K followers

Reminders / encouragement to do things like:

  • make takeaways concrete
  • put an epistemic status
  • quantify some claims

I guess I'd like to incentivise this, but I'm not sure telling people is the right approach.

I think the StackExchange sites have automatic reminders, or maybe even checks, of similar stuff. My last post on cross-validated (stack exchange for statistics) had hints about reproducible examples, I think.

Gwern has a writing checklist. Similar checklists could be forced on the author prior to submission.

What does this mean?

5
ChanaMessinger
2y
There's a lot the forum does that few people use. A weekly tip / explanation of how to use a feature.
2
Vaidehi Agarwalla
2y
Do you have a sense of specific features which might be being underused?
4
ChanaMessinger
2y
* Crossposting to LW * Embedding things: https://forum.effectivealtruism.org/posts/p3b9Hbsg9CKykKfC2/what-are-all-the-things-you-can-embed-into-forum-posts * Customizing the Frontpage with tags

Quiet comments

The ability to submit a comment without it showing up in "Recent Discussion". Among other things, this would allow discussion of controversial content without it stealing our collective attention from good content. Moderation already helps with this, but I would still have a use for quiet comments.

Hi, thank for this idea. Just so I understand, is your theory that people would self-identify comments which they think shouldn't show up in recent discussions? And that perhaps  people would put comments about controversial topics in particular into this category?

Allow commenting in "tag" pages, as if they were normal posts:

And have this be the default place for an AMA, or for discussing if an AI company is net-good or net-bad, or whatever, instead of having those conversations scattered (getting lost, repeating themselves) between different posts about the org

You can actually comment on tag pages, by clicking on "Discussion" under the tag name. I don't think that feature is really a good fit for your particular use case though - for that, I would suggest creating a post to start the discussion and tagging it with the organization name.

There's something here about collating all the relevant discussion around an org or tag, but I feel there's a lot of possible options - not sure what the best solution is.

2
ChanaMessinger
2y
Agreed. Kind of like bringing back arbital.

Embed timestamped Youtube videos

Agreevote and disagreevote on posts. Sometimes I'd like to upvote a post without implying I agree with it.

I mean, I opposed the original idea, but if we have it for comments we might as well have it for posts.

1
sawyer
2y
I'd guess the reason this was done for comments first is that posts are much longer and more complicated, such that it's often not clear what "agreeing" with the post even means. I think it's plausibly a good feature for posts, but I think it makes a lot more sense for comments.

When I interact with someone on the Forum, encourage me to follow them on Twitter.

Ways you might do this:

  1. Everywhere a Forum username is shown, add a bird icon that links to their Twitter profile.

  2. When I upvote, reply or send a DM, prompt me to follow the person on Twitter.

  3. Periodically email me a list of Twitter profiles of the EA Forum users I interact with the most.

I imagine (1) would get a lot of the value, and it's cheap to test.

Generally I think it would be interesting to brainstorm on ways that EA Forum could interact with EA Twitter, and vice versa.

The forum could use a "save comment" feature.

Hi, do you mean save as in bookmark to view later?

2
Charles He
2y
Yes, exactly. Specific use cases I'm thinking about:  * Comments I want to reply to, but can't immediately type out. It would be nice to batch them up to reply to later, and saving them would help. * There are also good informative comments, like someone recommending some authors here, that I want to save and use for personal value: I can save these in another system I use, but it's heavier weight, and many people wouldn't have another system to save these in.
1
Sharang Phadke
2y
Thanks, that makes sense

Markdown Syntax Reference

My User Story: As proficient markdown author and new EA Forum user, I want a reference guide for the forum's markdown, so that I can more quickly determine the correct syntax for features that vary across markdown implementations. (E.g., footnotes took me a loooong time to figure out.)

(I suspect there's an identical user story for new users who are NOT already used to writing markdown, but I don't want to put words in anyone's mouth.)

I suspect that a reference guide already exists, but my google-fu is failing me--I tried searching here, the general reference docs, github, and the code itself. As evidence that I'm not the only one, I see that the previous feature request thread has mentions a markdown cheatsheet as part of larger request for general markdown improvements. I'm hoping that adding just the syntax reference part is an easy win.

Thanks for this suggestion, I've made a note on our backlog!

Make tagging mandatory.

I don't think this is a good idea. I think it's quite hard for a new user to know what tags might apply to their post. Better for experienced users to add the tags for them.

1
Imma
2y
Good point. There are so many different tags that probably nobody would assign the optimal tags to a post. (At least, I won't)

As in, you can't publish a post without at least one tag?

1
Imma
2y
yes

Hide complete feature requests in the feature request forum article. 

Or, create a proper feature request board to have more visibility on the status of feature requests (there are many plugins which are pretty easy to set up)

Any number in a post can be suggested by any viewer.

 

Readers can see the median of suggested numbers, or weighted by upvotes. 

Can this be implemented via the current forecast system?

A bulk export feature. Something that provides the entire content of my  writing (drafts, posts, comments, questions, answers) in a suitable format.

At the moment, I do manual back-ups, but this would be nice to have for browsing or mining my own content for later use.

I doubt it's the most important, and maybe it's not a good idea, but an option for users to hide post and comment karma from their own point of view could be interesting. 

I think my polls create value, but not this much. Writing polls on the forum is absolutely overpowered in terms of gaining karma.

Better analytics for both authors and readers:

  • Readers can highlight sections of an article. The forum might then show a "featured highlight" similarly to how this works in Medium.
  • The forum can also measure how much screen time each paragraph in an article gets, and show this to users (a bit like a heatmap of where readers look at). This could lead to improved writing, and incentivize shorter articles.
  • An article's engagement and read time can become factors used for ranking, as a complement to Karma.

Remove the ability to strong upvote/agree on your own comments/posts and remove all existing such votes. (I think you can on posts - I haven't written any so can't check.)

I disagree. I like being able to sometimes say I think something is important. And the whole point is that my large amounts of karma give me this power. 

I wish that some people had a lot more karma than they currently do though. I wish I could transfer some of my karma to other people.

Maybe we could just report how much people have upvoted their own comments.

I agree that personal upvoting shouldn't affect one's overall karma.

2
Guy Raveh
2y
I don't think it does?
1
Holly Morgan
2y
Oh interesting. I was imagining that some people are just strong upvoting all of their own stuff, so it rewards dishonest behaviour. But maybe I'm just cynical :P
2
Nathan Young
2y
No, I have some of that sense that the current system is very gameable. (A better system might be that your strong upvoting is spread across everything you upvote every week so you can eitehr give a huge boost to a few things or a small boost to many)

You automatically strong-upvote your posts, and automatically upvote your comments. I think it should stay this way. But I have status quo bias.

Some kind of default categorisation filter that makes the frontpage more relevant to you - look to existing forum software for inspiration (currently this feels more like a zillion-author blog)

RSS feeds for tags (be surprised anyone else wants this but maybe?)

Nudge people to write shorter posts:

For example: Allow users to sort their feed by some algorithm that puts shorter posts higher up (not ignoring the karma, but also taking into account the length)

It might be tough to implement this in a way that doesn't boost linkposts (which  I think would be counter to your purpose).

2
Yonatan Cale
2y
1. I agree 2. No sorting algorithm is perfect. The relevant question, I think, is if this would be better than the current algorithm. (Would you prefer using it even though linkposts would be too high?) 3. With some extra effort, one could solve most of the link post problem. Specifically, I think the forum currently supports built-in link posts. Or one could search for "linkpost" or "link post" in the first line. But in practice I would just leave this problem as-is and see if anyone still uses this feed

Ability to collapse comments back after opening them on the front page

Easy way for users to blind authorship of posts. And a way to see a trending feed which is only affected by blinded karma

Thanks, I've recorded this, and I think it's a good idea!

I've shared this one before, but felt worth sharing again on this more recent thread: 

Co-authors on posts should get karma from that post. 

  • Allow authors to split karma how they wish (could default to 50/50).
  • Each gets some percentage of the whole which adds up to more than 100% e.g. if there's 100 karma each person gets 75 or something, so you're not discounting karma just because there's more people - presumably more authors doesn't necessarily mean a post is "worth" less per person or something, although there are ways people can game the system that are worth taking into account).

(originally posted here: https://forum.effectivealtruism.org/posts/NhSBgYq55BFs7t2cA/ea-forum-feature-suggestion-thread)

Hi, I believe the team recently shipped giving karma to multiple coauthors. Is your suggestion something more specific? If not, how much of the problem you see is addressed by this change?

2
Vaidehi Agarwalla
2y
Nope that's the same thing I think! Very cool :)

Sequences show up in search results

Would you want to see the result as "Sequence A, post X", or the opposite ordering, or just "sequence A"?

4
Thomas Kwa
2y
Probably just "sequence A", though anything would be fine to remove this gap in the search feature? I've searched the names of a sequence multiple times because that's just what I remember, it doesn't show up, and I have to Google it.

Ability to mute posts/comments from certain users.

(I don’t think this is done already but downvote or comment if so)

Some optional/additional way to weight karma as a percentage of total users (active users? readers?) so that sorting all posts by karma doesn’t show only newer posts at the top, and older popular posts way down with newer less-popular posts.

You can sort by inflation adjusted karma so this already exists I think

1
Jeremy
2y
Thank you for the correction!

Automatically crosspost from a variety of EA sources, with their permission. Eg LessWrong, SSC, alignment forum

I'd like a platform that does this, but not the ea forum.

2
calebp
2y
Can you make this for yourself using the RSS feeds and an RSS reader?
2
Vaidehi Agarwalla
2y
Probably, but I think it would be nice to have a discussion platform.
Comments4
Sorted by Click to highlight new comments since: Today at 4:24 PM

Moderate importance bug, probably applies to various situations where user uploaded images are presented:

This user's thumbnail is enormous, because an image inside of it is resized to the entire window.

Also, in comments/posts, I am having trouble controlling the size of images, they just resize horizontally to fit the screen, and this isn't ideal for many situations. This should be controllable.

Regarding Copy-pasting footnotes from a Google Document, I think it would be nice if after the copy-pasting:

  • Nested bullet points were not converted to non-nested bullet points.
  • The text in the cells of the headers of tables were not converted to sections which appear on the left navigation panel.
  • The footnote links were not broken.

For what it's worth I think my suggested voting pattern was dumb. I think I should have said agreevote for if you agree with the reasoning. 

Like 5 people have commented on this, why has noone upvoted it? 

Curated and popular this week
Relevant opportunities