JSN / Mokini

Mokini

A Web 3.0 social platform to organize your world's information.

Produced and narrated by Jason Schmidt.

Background

After creating Visualplant as a solo founder, I wanted to join forces with co-founders on my next venture.

At a networking event I met two students finishing up Master's degrees in the MBA and Machine Learning programs at Stanford University. Together we debated ideas to pursue and presented to angel investors. We got accepted to a venture fellowship program sponsored by Lightspeed Venture Partners who provided us with funding, work space and mentorship at their Sandhill headquarters. Here are the highlights.

Launch pad

College campuses are a dense microcosm of society. Diverse interest groups vie for attention and compete to get their messages heard. In general, it's a great place to study how information flows and find opportunities to improve it.

My co-founders and I conducted research at Stanford University and found that students and faculty use a myriad of modalities to broadcast one-to-many messages: Web forums, Facebook, Twitter and to a great degree, old-school (pardon the pun) mailing lists. Such real-time communication is spontaneous and intimate, but sadly the messages are either profusely redundant or else missed by many intended recipients and unsearchable. I teamed up with a couple co-founders to tackle this problem. Stanford U. was a perfect place to conduct a pilot.

Cold start

How do you convince people to change the way they communicate? Offer better communication tools. But if no one uses the tools, then they have no reach and the tools are useless. Therein lies the problem of the cold start. So, we began from a premise: don't expect users to change their habits.

We first created a simple Web-based input to kickstart our data store with RSS feeds, Twitter accounts and email lists. We would promote the tool to students, but initially we used it ourselves to curate a collection of trusted sources, carefully seeking permission from the publishers when necessary. Users did not need to change their communication habits; they could continue to use their usual modalities above to broadcast and receive. But now, with the sources and content proxied to our platform, we began to capture the ephemeral chatter of exclusive interest groups.

Our platform captured the ephemeral chatter of
exclusive interest groups.

Fig 1: Users could add a myriad of source types—including RSS, Twitter and mailing lists—then vote to affect the trust index of each source. Users, themselves, were in fact sources with evolving influence.

Evaluate sources

Once captured, our application evaluated the sources in depth. It tracked multiple reputation factors and assigned a variable trust index to each. It then processed content from the sources and performed categorization, topic-modeling and ranking. Finally, it de-duped, prioritized and presented the content as trending topics and announcements — all in near real-time. The results were stunningly helpful.

Fig 2: People -> Sources -> Topics -> Announcements