XNoBot’s: designing a reversible account filter without calling every small account a bot
AI assistance disclosure: this text was prepared with ChatGPT assistance for the project initiator. I am exploring XNoBot’s, an independent early-stage project that would give people more control over which accounts appear in their experience on X. There is no public app or demonstrated bot-detection accuracy to announce. The initial rule would hide content only when an account has fewer than 10…
XNoBot is an emerging project aimed at providing users with greater control over the accounts they encounter on X. The project does not yet have a publicly available app or evidence of its effectiveness at detecting bots. The proposed rule to hide content would only apply if an account has fewer than 10 followers, fewer than 10 accounts it follows, and fewer than 10 posts.
These figures are merely hypotheses to be tested and do not constitute proof that an account is automated. Real newcomers and quiet users might be unintentionally hidden, whereas established bots could potentially bypass the filter. This should be treated as a personal preference with the ability to be easily reversed, rather than a way to make accusations against other users.
Before committing to the project, several questions need to be answered. First, can the required counts, such as the number of followers, accounts followed, and posts, be obtained reliably and cost-effectively given the platform's current access limitations? It's crucial that unknown or outdated values do not automatically default to zero.
User control is another key consideration. The project will need to evaluate thresholds, potential exceptions, a temporary disable option, and a straightforward method to restore any content that may have been hidden. Privacy is also a significant concern. The project should aim to process as much data locally as possible to minimize the need for any integration and reduce data collection to the bare minimum.
To assess the impact of the project, it's essential to measure the unintentional hiding of legitimate accounts and its usefulness to volunteer moderators, rather than simply claiming an unmeasured bot-detection rate. The project is currently exploring a browser extension as a potential first step, but it's important to note that the coverage across different aspects of X and across various devices has not yet been established.
For developers who have previously created personal content filters, a primary failure mode to investigate would be determining what explanation to display, such as "hidden by your account-metrics rule," and whether a single-click option to reveal the hidden content would be beneficial. The project uses the alias XNoBot as a privacy precaution because social-media manipulation is a politically sensitive topic.
This story is not making any claims about specific threats. The intended funding model for the project is voluntary personal contributions of any amount, without any institutional funding or donor control. If you're interested in collaborating with the project, you can reach out to the project team at xnobots@proton.me.
Rewritten by urgent.news; may contain errors — check the original.
Source: dev.to