Blue Horizons has grown quickly. Thousands of Bluesky users already use it to understand their audience, find interesting people and make the platform easier to navigate. For an independent project, that response has been enormously encouraging. Apparently, it has also been rather upsetting in certain quarters. Our investigation into the events of 26 October found connections between several hostile accounts and another Bluesky management and analytics tool. We suspect competitive resentment helped fuel what followed. Having users freely choose an application is a wonderful principle, right up until they choose somebody else’s.
We tend to think that situation calls for improving your product. Others appear to have taken a more experimental approach to customer acquisition. On 26 October 2026, Blue Horizons experienced what our investigation assessed as an attempted distributed denial-of-service attack, or DDoS, involving accounts that had already shown sustained hostility towards the project. The attempt failed to take the application down. Our defensive architecture held up against the activity we observed.
Some of those involved also appear to have attempted the disruption while their Bluesky accounts were still connected to Blue Horizons, providing direct leads for our investigation. A memorable contribution to the art of covert operations: arriving at the target, signing the visitor book, and then attempting to look mysterious.
Over the following hours, we examined publicly available information to understand the surrounding activity. We identified connections between accounts involved in the incident and independently operated labeling services and lists targeting Blue Horizons and its developer, Pudgy Cat. These services were separate from Bluesky’s official moderation team. Their labels reflected the judgments of their operators. Running a labeler does not come with a badge, a judicial appointment or ownership of the platform, although the accessories appear to be available in the imagination.
The accusations included “Spam,” “Bot,” “Grifting,” “Platform Abuse,” and the particularly ambitious “MAGA Propaganda.” For context, the accounts concerned publish updates about a Bluesky management tool and articles about cats. We were unaware that feline editorial content had become a geopolitical threat. The cat’s political programme remains limited to earlier dinners and unrestricted access to the sofa.
Alongside these labels, we found calls encouraging others to report our accounts under assorted allegations. The accusations varied considerably; the preferred outcome seemed remarkably consistent. Apparently, choosing a charge was the flexible part. The connections our review found to another Bluesky application raised questions about competitive motives. Those connections alone do not establish that its developer directed the attempted attack. We will leave the business of turning suspicion into a verdict to the people who apparently specialise in it.
Nevertheless, the apparent hostility towards an alternative tool deserves an answer. We are comfortable with other developers building useful applications. Users can try them, compare them and choose whichever works best. They can even use several. Nobody needs to swear allegiance to a dashboard. If another tool is better, people should use it. Trying to make an alternative disappear is a curious way to express confidence in your own. We had rather hoped “competition” would involve shipping features. A neighbourhood campaign to have the other shop closed suggests the product demonstration has gone badly.
As the harassment continued, we used Blue Horizons to block accounts we identified as directly involved, together with a wider set whose connections raised concerns about further targeting. This involved a large number of accounts, but still several orders of magnitude fewer than Bluesky’s total user base: a negligible share at platform scale. Some uninvolved people were caught in that wider selection. We are sorry about those mistaken blocks. Association alone does not establish involvement, and our selection needed greater precision.
For some of the accounts we blocked, however, the discovery that the block button also worked in our direction appears to have been deeply unsettling. Further complaints went to Bluesky, including from accounts our investigation had identified as directly involved. The allegations also expanded to independent accounts that collaborate with us to promote the tool. Those collaborators operate independently. Working with us does not make them responsible for every decision we take. Extending a grievance to anyone standing nearby does, however, make the underlying behaviour rather easier to recognise.
The application continued operating, and the situation subsequently settled. We cannot speak for Bluesky’s internal handling of individual reports. We can speak for our decision to stop entertaining persistent hostility. Being blocked is a normal possibility on a social network. It can be inconvenient, disappointing or occasionally overdue. It does not require an emergency session of the United Nations.
Which brings us to the self-appointed proprietors of Bluesky. Certain corners of the platform have developed an impressive sense of jurisdiction. They seem to believe that arriving earlier, accumulating followers or operating a moderation list entitles them to decide who may participate, which tools may exist, and whose presence requires an explanation. We must have missed the ceremony at which they were handed the keys.
A follow becomes a marriage contract. An unfollow becomes an act of betrayal. A block becomes an offence against the Crown. Meanwhile, their own blocks, labels and campaigns are understood to be essential community maintenance. It is an admirably convenient constitution.
Bluesky’s independent moderation tools can be useful. People should be able to choose filters, avoid abuse and organise their own experience. Those powers require judgment, including the ability to distinguish a genuine problem from someone doing something you personally dislike. A labeler is a tool. Its operator can be mistaken. A list can contain poor judgments. Adding more subscribers does not improve the quality of the evidence underneath an accusation.
Bluesky’s own rules explicitly prohibit abusing lists, labels and community moderation tools, along with reporting campaigns conducted in bad faith. There is no exemption for people who are very certain they are the good ones. Bluesky Community Guidelines. The platform’s Safety team has also publicly addressed dogpiling and harassment, announcing enforcement efforts against that behaviour. The problem is sufficiently real for Bluesky itself to acknowledge it. Bluesky Safety statement.
For newcomers, these territorial disputes must be especially appealing. Imagine joining a social network to talk about books, share photographs or find people with similar interests, then discovering that an unelected residents’ association would like to inspect your follows. Some people will argue back. Others will simply leave. Then everyone gets to wonder why attracting new users is difficult. A community grows when people can arrive, participate and find something worth returning to. Requiring them to navigate somebody else’s grudges is an unusual onboarding strategy.
Blue Horizons was built to make that experience easier to manage. It brings audience relationships, activity and discovery into one place, so people can understand their connections and find accounts they actually want to interact with. It also provides practical ways to organise those relationships: filters, private labels, protected accounts and blacklist controls, alongside follow, unfollow, mute and block actions. Those choices belong to the person managing the account.
The ability to inspect public labels, labeled posts, and known lists or starter packs containing an account gives users more context about how they are being categorised. Where information is incomplete, that limitation matters. A connection is something to examine; it is not an automatic verdict.
Activity insights, saved searches, post management and scheduling address the more ordinary side of using Bluesky: spending less time administering an account and more time having something worthwhile to say. We would much prefer that ordinary work to occupy everyone’s attention. It is, after all, why most people joined a social network in the first place. But when someone insists on making themselves a recurring problem, users should have practical ways to disengage. They should not need permission from the person causing it.
We will keep developing Blue Horizons and helping people make Bluesky a place they want to use. Anyone building something useful for the same community is welcome to do the same. There is plenty of room, even if a few people keep measuring it as though they inherited the property.
The attempted disruption left us with an application still running and an unusually clear demonstration of why its controls are useful. If you set out to undermine a tool and end up demonstrating its usefulness, you have not eliminated a competitor. You have volunteered for its product demo.
We appreciate the initiative. Next time, a feature request would involve considerably less embarrassment.