Every language. One network.

Most of the internet has a default language.

A handful of dominant languages run the visible internet. Users of everything else, ranging from large regional languages to indigenous languages, from ancient languages to constructed languages, even most sign languages, are pushed towards using English on digital platforms. The result is a flattened internet which pretends linguistic diversity is a localization problem to be solved instead of the actual shape of how humans communicate.

There are more than 7,000 living languages, more than 20,000 dialects (per the Multitree project's linguistic genealogy), hundreds of sign languages, dozens of liturgical languages with active scholarly communities, and entire writing systems that mainstream platforms don't render correctly. None of this is fringe. It represents the past and present for the majority of Earth.

Plaza is built on the opposite assumption to other platforms: every language is a first-class citizen, the structure of the network reflects how languages actually relate to each other, and the technical work to provide universal language support is not just feasible but necessary.

How Plaza approaches it

Historical text in its original form.

A detailed and granular representation of all the world's languages, both living and dead, presents the perfect opportunity to showcase historical works as they were originally written. Simple bots post excerpts from major texts of cultural significance such as the Iliad and this can very easily be extended to a huge corpus of lesser known works. For languages without machine translation support a public domain translation is cached with each post. This makes these posts comprehensible to the entire website, usually via an English language pivot.

In the future more complex bots constructed in the same vein as Talkie-1930 (a 13-billion-parameter language model trained on pre-1931 text) can have knowledge confined to a single language in a single time period or civilization. They will offer an invaluable resource for scholars and learners of these languages with Plaza serving as a rare place on the internet providing this kind of full immersion.

The sovereign option.

The communities whose languages are least represented online, notably indigenous communities, are often the ones with the most reason to distrust platforms. Words, songs, sacred knowledge, and everyday speech have all been extracted and appropriated by outside institutions before, and a platform that asks these communities to upload everything and trust its privacy promises is asking too much.

Plaza is being built so that privacy won't rest on promises. The architecture allows for chosen sensitive posts to live on hardware the user owns, instead of our central database. Plaza handles only the metadata needed to index and route content, metadata Plaza does hold, and is transparent about; the content itself is served directly to permissioned users. This is guaranteed. Plaza will be structurally unable to read what sits on a server it doesn't host.

Sovereignty doesn't depend on Plaza being well-intentioned or well-run in any given year, it depends on architecture in which Plaza never possesses the content at all. Extraction is prevented by design rather than merely against the rules. It is highly important to communities whose work has been mishandled before, and for anyone else who'd rather avoid their most personal content doubling as a company's asset.

Where things stand

Stay involved

Whatever your reason for being here — language speaker, signer, learner, researcher, builder, or just curious — there's room for you. Sign up to follow Plaza as it grows. You'll get the occasional honest update on our development progress and which language communities are joining. If you want to be considered for an alpha invitation, you will be. If you just want to know what happens, that's fine too.