(These notes are collated from the StartHereSocial project, a prototype domain picker for choosing a Social Web provider)
People join social networking services for many reasons. In descending order, the main motivations are
- socialisation especially with friends and family,
- prosocial behaviour like sharing helpful and informative content,
- escapism to manage mood and be entertained, and
- self-presentation – expressing oneself and sharing personal states and situations.
Some individuals want to connect with people in their town or country. Others are drawn to spaces defined by a particular language or shared cultural background that may span across regions. Some (most) want to lurk, read news, enjoy memes, share content to their network, and look at photos of sheep. A small minority (around 1%) want as wide an audience as possible with a lot of global connections.
Not everyone, in fact not most people, want to be part of a global town square. While these large, open platforms receive the most attention and investment, the real work of connecting humans to each other is a global exercise in relatively unremarkable, everyday community-building.
A poll of 550 Fediverse accounts found that while 50% of new users would choose a service provider by topic, 37% prefer to choose by language or region, and 13% by a range of other decision points.
Joining the Fediverse is Hard – Helping People Choose is Even Harder
Unless a Fediverse app handles onboarding effectively, new users often face analysis paralysis when choosing a server. To address this, many developers and community members create guides to help people choose. These often rely on metadata from nodeinfo files, hosting data, or inferred geolocation to classify services.
If you’re building such a guide, especially one using nodeinfo data, please consider the following:
Common False Assumptions When Classifying Fediverse Servers
⚠️Falsehood: Language equals region
This is not always true. Someone seeking a Spanish-language community may be in South America, not Spain. Welsh learners worldwide may prefer a Welsh-language service. Chinese speakers live around the globe, and their language preference does not imply a geographic location.
⚠️Falsehood: Region equals language.
People in a certain country may not speak the dominant local language, like an English-language service in Japan for expats. Countries like Ireland, Canada, Belgium, or Switzerland have more than one official language. Others may serve border regions or international communities. Where possible, use the user’s browser language preferences from the Accept-Language header, this is a more accurate reflection of their needs.
⚠️Hosting country equals language or regional focus. Many Asia-Pacific communities are hosted in Australia but are not Australian in focus. 1,500 services hosted by popular managed host provider masto.host will show up as French, but most are not intended for French audiences. Sites using Cloudflare may show up as being US-based, even though they are hosted in other countries.
⚠️Falsehood: Language or regional focus means it’s not a general purpose service.
A Madagascar-based or Italian language service might still be a general purpose service for their audience. General purpose servers exist to support languages other than English.
⚠️Falsehood: All domains in Fediverse databases are safe or appropriate.
This is dangerously untrue. Some high-traffic Fediverse services and communities are known to host illegal content, coordinate harassment, or promote hate speech. Consider using a denylist to prevent your users from being exposed to unsafe environments. If you publish a data resource, consider gating certain domains behind a click-through warning or limiting access to full datasets.
Reality check: four of the ten largest Mastodon instances, according to many public Fediverse server lists, are also among the most-blocked domains. One is operated as a Russian news spam network and contains clearly falsified nodeinfo data. Misuse of public metrics and rankings is common.
If you’re using third-party data services to guide user recommendations, it is essential to review their inclusion and exclusion criteria. Displaying a domain can imply endorsement.
⚠️Falsehood: Nodeinfo data is honest and accurate.
GoToSocial servers have an optional “baffle” feature that reports spurious data (“Serve randomized, preposterous stats at instance and nodeinfo # endpoints”) to crawlers. NodeBB publishes nodeinfo whether it’s federated or not. Plenty of nodes purposefully report inaccurate data, for example one server reports 97 users with 97 million posts… (it has neither).
About This Work
I maintain a curated decision support tool that includes a hand-reviewed selection of Fediverse servers and carefully crafted language that can be understood by a newcomer to the space.
The services listed are open for registration, actively moderated by engaged administrators, and – where relevant – annotated with geographic or language-related information.
This curation cannot be automated. It requires careful, human assessment of moderation practices, community tone, and safety policies. While labour-intensive, it is essential for ensuring that users are directed toward trustworthy and welcoming spaces.
The dataset is freely available for anyone to use. A working example of how the data can be visualised to support user onboarding and server selection is available on the StartHereSocial pages.
