One of the ways in which Bluesky and ATProto are complete game changers for social media is the way they allow even a lone developer to offer an improved social media experience in an ecosystem with millions of users. Ideas that were never going to happen at other places are within reach enough to be worth discussion.
This is hopefully the first post in a series of them on some ideas I have that stem from the same premise: The incentives created by viral posts and algorithmic feeds rewarding engagement have made things worse. However, by altering the incentives we can similarly make things better or at least less bad.
We need to direct people's attention away from the bad posts and towards the good ones. Likes, replies, quotes and other forms of engagement only do a marginal job at best of indicating which content is good and have only gotten worse the more people set out to game the system. Better recommendations require better data.
It has become apparent, that even though there is a dedicated group of power users who strongly prefer reverse chronological feeds, an overwhelming majority of users are going to end up consuming content via algorithmic recommendation feeds. Bluesky giving users a choice in algorithms is a great first step, but even the best possible algorithm is limited by the data to which it has access.
One of the Problems
Since the very early days of the internet, people have encountered the issue that creating good, original content is really hard and copying stuff other people made without acknowledgement is very easy.
The example that inspired this post was this video from @noturtlesoup17.bsky.social:
She explains what happened here and illustrates what can happen:
This is a problem as old as the internet that inspired a now classic Nedroid comic by Anthony Clark:
There are so many variations on this where it's less obvious what happened and why it matters than a video of a video without any modifications that includes the original post handle and an independent journalist trying to cover important national events who needs direct financial support from their audience and posted the original content directly on the same platform.
On one hand, this is the way the internet has been since I can remember and it enables many fun aspects of internet culture like memes and copypastas.
However, assuming there are actually benefits to having a popular post, allowing those to accrue to engagement farming accounts rather than the people most capable of creating quality new and original content is decidedly bad.
The Solution
The AT Protocol label lexicon provides a blueprint that can potentially be copied almost exactly with one addition piece of information, a second AT URI to allow it to point from the reposted content to the original. Until someone comes up with a better name, I am going to call it a repost pointer.
Rather than the value being the name of the label, in a repost pointer, the value or equivalent could be used to further describe the relationship between the original content and the repost. Different use cases for the information contained in a repost pointer could handle cases of literal word for word reposts differently than ones that change something minor, such as adding a joke, or reposts that take only a small fraction of a video in order to make a new point about that specific segment.
The ability for algorithmic feeds to use this data in ways that could end up being incredibly powerful if done well. It not only provides a mechanism for redirecting exposure away from engagement farming accounts and towards original content creators, but also allows for grouping together the metrics from several posts that got good engagement to recognize what should have been one post that received great engagement.
There is also the potential for apps that show the post content to users to either show the original content alongside (or more dangerously, in place of) the reposted content or provide a link to allow users to easily navigate to the original and thus help direct more of the engagement back to the original post.