Venture investor at RRE Ventures in NYC. Formerly Marketing at Drobo.
Thanks, tomloverro! Apologies on the video! We didn't have much time to complete that portion, with everything we had been writing.
From the implementation side, it's pretty simplistic: we're using sockets and running Impact.js (though we had to do a lot of monkey-patching to make it work outside of a browser environment). All keyboard commands are sent to the server, validated, run, and then the completed position and/or actions are sent to each client. This means that no physics are running on the client-side and everyone should always collide correctly.
Thanks for your vote and your Great feedback Tom!
If I had more time to put for the algorithm, I would have the users own dictionary of positive and negative dataset, those information can be learned from: - user following/like certain pages,people,brands - comparing the user status,+1,comments against a positive and negative dictionary - the interaction from the "social newspaper" which i plan to make a global comment section and +1's and maximizing posts/pictures and video, although this still would be experimental.
Would love to hear what you think of those initial functions release for a machine learning algorithm.