This isn't meant as negative criticism, just trying to figure out if I've understood how it works correctly? (Note, I appreciate that the source is available, but obviously I can't know that the source you share is the source you run on your back-end. So we're back to trust, which is fine in my book).
[edit: re other clients -- the most straightforward thing to do would be to make an iframe(or js)-based proxy service (a la stumble upon) in order to enable any browser, I suppose?
I've thought about something similar, and I wonder if an XMPP back-end with one "room" pr url and the server-side log-extension might work? That should then make it trivial (unnecessary!) to enable better clients than web-only -- ie use a command line xmpp client to participate in the discussion, rather than forcing the use of the web browser. I realise many people think using the web browser is an advantage -- I generally find GUIs built on top of browsers to be slow and not very efficient. Extensions less so -- but the problem is of course that you'd have to port to Firefox, Opera, Webkit and IE (along with mobile variants) -- to achieve parity with systems like disqus etc]
Perhaps XMPP+Server Log extension+rate limiting?
The inability to edit messages might be a worse strike against using (a not too heavily modified) XMPP server, though.
Seems like this might be a good business for someone to start, matching a github repository to a running server; third party verification that said code is actually what's running on a server.
If it helps, I made an "about me" page, so I'm not quite a total stranger: http://www.words4chrome.com/team/
Cold-start: convince a site with an existing audience to promote it.
As far as overcoming the cold start, I'm looking not so much for marketing ideas, but for product or feature enhancements that would be valuable to users immediately, without a huge number of other users.
Google search - useful for 1 user Facebook - theoretically useful for a handful of friends Twitter - useful for a slightly larger network
Words doesn't have friend/follower mechanisms -- it's just not the nature of the product. The intention is for users to interact with each other as they happen upon the same content. It's a dynamically-changing social graph. Unfortunately, this means the critical mass of users is definitely north of 100k, probably higher.
These "second screen" or "third voice" apps require people to be (a) trained on the tool, (b) have a common set of target objects that they want to talk about. You could start off simple by looking for a small community of people who:
(a) use chrome
(b) regularly visit multiple (5-10) sites related to their primary site
(c) currently talk about the secondary sites using their primary site
(d) see value in having their existing conversations appear in-context on the secondary sites
The plugin may have business applications for proof-reading.
Note: Re-posted with permission from Dan. Last week's post got dinged by a semi-false-positive algorithmic penalty after only a few minutes.