Lightnion: Tor in the Browser

tor speed is an improvement, it impedes on social network addiction :p

It's a lightweight Tor client written in Javascript than runs in the end-user browser. The websocket-to-TCP (dumb) proxy acts in some ways as a Guard node from the live network (it's not about running a Tor client on a remote proxy). Hopefully, a pluggable transport will be implemented at some point and Guard nodes will be able to directly accept incoming connections, without proxy.

Note that it's not a replacement for the Tor browser: it is not build for the same usages. A good example is people implementing privacy-oriented tools with strong cryptography in Javascript to drives adoption. The goal is to improve the privacy of such partially honest services by bringing anonymity into the picture, with a better compromise than a "trust us to not log" policy.

You can even fantom models where a commercial service provider focused on privacy push its users to use the Tor Browser while shipping a Tor client by default to its user that didn't bothered installing one. Because at the end, it will be easier to keep your promises about not doing logs if you intentionally forced all of your users through the Tor network.

There's a consensus made on the global state of the Tor network made between several well-known parties: just as a regular Tor client, Lighnion downloads the consensus document, parses it, validates its signatures and perform a path selection algorithm on it. It's several megabytes to downloaded by the Javascript client (itself being few kilobytes) and is currently a limitation of the protocol.

If the "walking onions" proposal gets maintstream (i.e. private retrieval of a consensus subset), we can hope for improvements (see https://gitweb.torproject.org/torspec.git/tree/proposals/300...).