As Elon Musk’s X continues to explore the most “truth-seeking” system to ensure accuracy in online discussions, it is reconsidering the use of upvotes and downvotes on replies to posts as a way to better rank engagement and surface the most relevant comments within each thread.
Well, downvote, anyway.
As can be seen from this example, X is experimenting with introducing a dislike option to replies to posts.This gives users another way to share their feedback on user comments.
The option was discovered by researchers in the app’s backend code. Aaron PerisX Engineer and Jay Baxter Then check the internal tests.
It’s probably not quite what you think it is.
According to Baxter:
“Community notes wouldn’t work well without negative rating signals. But you have to be careful how you use them. If you simply add them all up, it becomes a Reddit-like hive mind. One way to do it is to only lower the rank if there are negative ratings from people you normally disagree with..”
Thus, the downvote option for replies is weighted based on political viewpoint, preventing users with a particular ideology or viewpoint from attacking replies and downvoting and ignoring everything they don’t agree with.
So the ultimate ideal would be a system of de-ranking replies without political bias, where junk or spam comments, or statements that are obviously misleading or false, would be slid lower in the list or even hidden, rather than silencing opposing views.
It’s an interesting concept, similar to how previous Twitter teams have experimented with downvoting in replies, not as a way to voice an opinion on the comment itself, but as a way to push down spam and junk.
In 2021, Twitter tried a largely identical process, without the political leaning weighting.
In essence, the idea was that the downvote arrows would indicate the relevance and value of a post’s replies, rather than the personal opinion of the comment expressed, in order to improve ranking quality.
As you can see from this image, Twitter has actually tried out a few variations on the UI for upvoting and downvoting, but part of the problem at the time was that users didn’t fully understand what these buttons meant.
As a result, many people have used them, much like they are used on Reddit, as an indicator of disagreement, which was not their intended purpose.
X takes this into account by making the process more transparent and actually down-ranking comments they don’t like. X is effectively acknowledging that people on one side of the political divide will undoubtedly down-vote opinions they disagree with, but by adding weighting, they take this into account and ensure that consequences are only applied where there is agreement.
Ultimately, opinions themselves should not be punished, but only the worst examples of spam and junk. Which approach is better will depend heavily on how X determines political leanings, and whether that process is accurate enough to maximize relevance.
Either way, it’s an interesting experiment that could help maximize engagement within the app by highlighting the most valuable post replies.
This means that prioritizing political preferences is a dangerous path in itself, as some topics lack universal agreement and certain types of misinformation may also be sustained within the app.
But at least in this context it does make sense as a way to filter out junk comments.
X hasn’t said it will be putting this to live testing, but it’s still in the early experimental stages.
