There have been a number of posts recently making the case that EA Forum activity is dropping, and that this may be due to structural problems with the site. See GHD discussion here is slowly dying, Keeping the EA Forum alive, and this quick take by RyanCarey.
I think the diagnosis these posts offer is accurate, but there has not yet been consensus on the treatment. This is in large part because people have different models of what the EA Forum is for.
Here is a crude stereotype of the spectrum of views:
These vary by a few dimensions:
I have my own perspective on what ideals we should reach for and why.[1] But let's skip further theorizing and make some polls.
I was convinced by abrahamrowe's claim in Keeping the EA Forum alive that too many low-quality posts, often low-effort link posts from elsewhere, are competing for attention.
The disagreement I have with him is that I think bringing more EA content onto the Forum is essential. I have often found myself shocked by the apparent lack of discussion about absolutely massive EA news events happening elsewhere. We don't need to talk about every viral twitter post. But when the Anthropic-DOW dispute kicked off earlier this year, barely any EA Forum activity ensued!
As I commented at the time, this is a big issue if we are ever in a real time crunch. One of these days, we might suddenly find ourselves faced with the fact that the best thing we can do is buy plane tickets for thousands of EA university students and unemployed San Franciscan software engineers, and fly them out to DC to wave placards in front of the White House. If it comes to deciding that, we can't rely on a handful of people messaging in disjointed Slack channels. All parts of the movement need to be represented.
There's a lot happening out there and it's not just AI news. Farmed animal legislation often goes unremarked, as do new Precision Livestock Farming technologies. Perhaps most notably, many EA organizations routinely publish high-quality reports that never appear on the EA Forum. Here are two examples:
Both would have been among the most relevant and informative posts on the front page, had they been posted at the time. Not doing so defeats a core function of the Forum, that being to centralize content. I'm not faulting EA organizations for not crossposting - they can hardly be blamed for not taking the time if they know they will barely get any views. The engagement problem is self-perpetuating.
This brings us to the first two solutions I want to propose. Linkposts do what they are intended for poorly. They give unearned karma to people sharing others content, and they incentivize personal advertising over Forum-native posting. Crossposts are much better, but underutilized.
So. Let's replace linkposts with two things:
As a result: posts doing original thinking now have dedicated space. At the same time, a dedicated news tab should incentivize more posting on current events.[4]
The infinite scroll interface is that thing basically all platforms (Twitter, Reddit, Substack, etc.) have now where instead of immediately needing to click "load more", you can scroll and scroll and scroll. Yes, these are somewhat addicting, but I think it's a fairly innocuous form of attention seizing. As the biggest issue with the Forum is engagement, I think it's time to consider well-evidenced strategies like this.
To do this, we must figure out what to do with everything currently below the main feed:
What to do with the Community topic section and Quick Takes is the hard part. The easy part is the "Popular Comments" and "Recent Discussion" sections.
The idea for this also builds on the recent quick take by RyanCarey.
My concern is simple: where is the list of the latest posts? The screen is so overwhelmed by "featured posts" and "comments" and "quick takes" that the centerpiece of the whole thing is completely gone.
Visual clutter does seem to be a problem.
I understand that the aim of these is to boost discussion and commenting - if they achieved that, I'd be all for it. But I basically never interact with these sections. They are far too out-of-context.
But even if we agree on that, we will need to remove more to get that captivating Infinite Scroll UI in. The community topic was divorced from the regular feed for very good reasons in 2023. Even after being banished halfway down the front page, the topic still routinely gets significantly more engagement than any other.
But what if instead of segmenting the topic off, we just nuked the topic karma?
I discovered while writing this post that there's already an option to opt-in to including community posts, but with "Reduced" karma, where they are "shown as if they have half as much karma". But it's currently a fairly pointless adjustment with the community tab existing.
Note that this was the case for some time, before the separate Community tab section was created. The one thought for improvement I have over the historical precedent is that the internal page ranking weight should not be the only thing affected by nuking the karma. The karma itself, the visual upvotes, should also be halved as well.[5] It's a finding of experimental social science that herding effects occur for online voting mechanisms. Even if the community posts are low ranked, people will get FOMO and jump into the fray if that's where everyone else is perceived to be.
Now that more room has been cleared, I think it may just be a matter of moving the quick takes to the side:
A LessWrong feature I like is that when you scroll down, the menu on the left (here, in green) stays put. Adopting this feature would make ample room for the main feed to continue when scrolling down, allowing for a narrower quick takes section parallel to it.
I want to close with two last points. A warning: It's quite possible those of us not doing the actual implementation are ignorant of good reasons not to make these adjustments. So I don't want the polls to be perceived as a voting mechanism, just as a way to collect Forum user feedback.
And a note of appreciation: the Forum is already a wonderful platform, hosting most of the best EA writing on the internet. It deserves more readers.
Roughly somewhere in the middle. It should be a serious place; but I think that EA is such a broad movement and research agenda that going too far left on these axes would by default (i.e. not by intent) drastically curtail ideological diversity. Moral catastrophes would go unnoticed because advocates neglected to cite a 2013 Yudkowsky post.
Toby Tremlett has told me before that he'd like to hear new ideas for what to crosspost, which I appreciate; but I think constantly DMing random suggestions is just too awkward to be a sustainable way to source new crossposts.
Let me anticipate a concern: my guess is that more community sourcing like this would actually lessen rather than increase the Forum team's workload; even if the suggestion box got 50+ links daily, the initial link screening can be AI automated.
I think Matrice Jacobine has been almost single-handedly responsible for keeping the Forum up to date on current events, and I'd love to see more of that.
Or dropped by whatever percent is best, with the displayed karma rounded to the nearest whole number. Trial and error here is probably needed.
Thanks for posting Dylan! I like the creative ideas on how to separate out cross- and link-posts. I'll consider them if/when we work on that feature.
In case people are wondering what the Forum Team's response is to recent concern about the site - I've been drafting a long (too long) post today. Will edit, make sure I'm synced up with the rest of CEA etc... before posting. Shouldn't be too long.
PS- Kind of you to check with me first as well!