Feedback requested: EA Forum reactions

The EA Forum team is experimenting with design options for reacting to posts or comments, and we’d like your feedback over the next week. This post presents some reasoning and a few design directions we are considering.

Motivation and goals

We’ve been working on a feature to allow users to react to content in more granular ways than upvoting without needing to write a full comment. This project is similar to what our friends at LessWrong are working on, with some differences in motivation and design constraints that we’ll lay out below.

Motivation

We’ve heard from a number of users that after putting in many hours of work into Forum posts, sometimes the only (positive) feedback they get is in the form of karma. Karma can sometimes feel opaque and sterile rather than motivating. While some users do write short positive comments, we’ve heard that it can feel like the bar for commenting is high, requiring a nuanced take or useful feedback. We think reactions can provide a middle ground between an upvote and a comment.

Why restrict reactions to positive feedback?

In our first iteration of this feature, we tested several reactions with neutral or negative emotional connotations like 🤔 (confused). We found that authors were motivated to follow up with users reacting in such a way and clarify confusion or feedback, but negative reaction emojis didn’t have enough information to be actionable in the way that a full comment is. Therefore, we’ve decided to limit ourselves to positive reactions. One counterargument we’ve heard is that focusing only on positive reactions could make it harder to give constructive and critical feedback. We suspect this wouldn’t deter most readers who have good points to make, but it is something we want to hear from you about and track over time.

Keep it simple

As we approach this feature, we want to keep the user experience simple and intuitive.

If we add another form of voting /​ reacting, it must be very easy to understand how it relates to karma and agree/​disagree voting. Our first iteration of the feature added reactions at the comment level, and most users we interviewed found it confusing to deal with karma, agree/​disagree, and reactions on the same piece of content. Based on that feedback, we’re quite confident we need to prioritize simplicity. For instance, we’re unlikely to add a third dimension of voting like LessWrong are considering without some way of maintaining simplicity, and we’re unlikely to add a very large set of possible reactions.

Design questions we have

1. Reactions on posts, comments, or both?

Importantly, if we add reactions to comments, we would likely change agree/​disagree voting significantly (removing strong votes and possibly anonymity) to maintain simplicity. We’re curious how users feel about this tradeoff.

We hypothesize that the biggest gap in terms of content-interaction and feedback is at the post level, since comments are generally shorter and more specific, and already have agree/​disagree voting. On the other hand, we recognize that a lot of interesting back-and-forth discussion happens in comments, where reactions can also be useful.

2. Anonymous or non-anonymous?

We’ve heard that seeing positive feedback from identified individuals can feel more meaningful. On the other hand, we’ve heard that non-anonymous reactions may deter some users from using the feature, particularly if we combine agree/​disagree voting into reactions (which we would only do for comments, since posts don’t have agree/​disagree voting).

3. What set of reactions?

As mentioned above, we plan to introduce a fairly restricted set of reactions, but we’re interested in what reactions are most compelling for users and communicate the most important concepts. Do you think we’re missing any key reactions? Are any of these confusing?

4. (minor) Forum-styled or common emojis?

Forum-styled icons would fit better with the aesthetic of the site, but common emojis are more recognizable and add a bit more energy and color to reactions.

We’re really looking forward to feedback and arguments from the community, so please share your thoughts!