Time to pick up the Information Flow story again. The next leg for me is iCurrent. Over the last 2+ years, I’ve been working with a fabulous team with the full support of the VC I wished for to build a product for personalized news delivery for the mainstream audience. Two years quietly building a product. Are you crazy? I can hear some asking. In this day and age, when you can ship on the Internet in 6 weeks with no money from VCs, and when the audience can help you discover what they want?
Our case is rather simple:
1) the vision of what is wanted is old
2) achieving it is tremendously hard
3) we had earned a viewpoint on what was required
4) we were ready to play a round at a time to eliminate risks
Visions of what is wanted are quite old
… and they still genuinely appeal to broad audiences. Different complementary visions emphasize different angles. Consider:
- Negroponte’s Daily Me: a virtual newspaper tailored to each individual’s interests. This vision promises individualized delivery with the simplicity and value of traditional media packaging.
- Intelligent Agents: mechanical intelligence harnessed to enable a smart flow of information. This requires representing your needs and interests so they can compute with these structures to do work for you.
- Daily Briefing Service in a Box: imagine having a Fortune 50-sized organization whose primary mission was to gather intelligence so you could do your job. One person in America has that. This angle says that along with collecting and filtering, we want prioritization and distilling i.e. REAP-ing of value.
Yet, for all the bounty the Internet has produced, nothing has achieved these visions in a form plausible for mainstream adoption.
Because it’s hard …
… what was before is mighty good and these visions speak only metaphorically. Just as the vision of Artificial Intelligence in 1959 mislead the field’s founders on how long it’d take (they thought 25 years), these visions in their simplicity and sexiness create a false optimism.
The shallowness of first-gen delivery or discovery or personalization becomes clear in retrospect. In 1997, PointCast was cast as the new concept of push, but isn’t push what we had in the old form? As in pushy and broadcast? Jumping to now, Digg—even if we call it a success pending a bigger success—is only about what’s hot as filtered by people of a type, but how about the range of interests? And start pages: they leave the actual work of selecting feeds (plumbing) and widgets (paneling) to the person.
Now consider the functionality to simplicity ratio of the Great American Newspaper. What it offered did not happen overnight, though at a historical scale it did. We marvel at the nice simple package that robustly fit in our lives as they once were. It’s not mystical, almost anybody who has experienced it can explain what all newspapers did/do so well.
Now as we move to personalized papers—each with a circulation of one—we still want all those values preserved. We’re adding requirements, not subtracting. We are just 15 years into the mainstream Internet, and user behaviors, infrastructure, and capabilities are ripe for the full experience to be achieved.
We started with a simple observation …
… there is no way to give somebody what she wants without her being meaningfully involved. Meaningfully to her and to the system. Personalization requires the person not just to “consume” the product but to participate in its very production. The key to creating this active participation is an experience in which the user can elaborate their interest incrementally in a grounded way.
An analogy: because of the roaring success of Google, people have until recently focused too much on the magic in responding to the query in, and not enough to the result out as a canvas for ongoing dialog. For personalized delivery, we saw that it was all the more important to ground the dialog in the right specificities to capture something so elusive as interests. This all comes together at what I use to call reified interest objects. Now more correctly to its intention and to everybody’s relief, we call it a channel in our system.
The second significant bet was that a hybrid blend of user participation, curatorial leverage, and algorithmic methods is necessary so that the mechanism of channels could support the broad range of interests we wanted. The additional challenges of prioritizing and packaging many channels into a simple experience only made this all the more clear.
And there the path began …
… we knew it would be hard. We saw many moving parts in the system and knew we would need sophisticated tools for the curatorial and design processes. And despite years with the tinkertoy suite and the design ideas, I wasn’t convinced we could make it all come together in the right experience for the broad audience. Still it was worth the shot, so we raised a $1M seed in Aug 2007.
We spent the first 6 months building a conceptually complete architecture. Not architecture in the sense of a scalable Internet architecture, but as in a framework with places for the ideas. We implemented enough at coarse 90% levels so that we could assess proceeding. Beyond the airflow simulation, it was wind tunnel tests, and then enough of an airplane to test in real world conditions.
Throughout we interviewed people all over the US (Idaho, Kentucky, Ohio, Florida), a typical qualifier being not knowing what “TechCrunch” was. We saw that the audience had bounced off the best of the silver bullet ideas, that they did actually want to stay current across their interests in a simplified way, but that what was required was subtle and rich. Also felt confirmed was that the “wow” would have to be that “it works” and that this would be the lever into building an audience.
At the 8 month point, based on what we had, we decided to move forward. We set out to extend the seed into a full A round and in parallel we proceeded to a private alpha. We enrolled a batch of 24 users, and have added 20-30 each month, releasing on average two times a month over 18 months. We’ve constantly added functionality, got past quality thresholds that obscured value, and refined the whole in the constant pursuit of the right “simplicity on the other side of complexity.”
Now the next leg begins. We need many more users to take the product to the next level. And we need the attention of those that get what we are doing to take the company to its next level. Thus we are ready to engage in the bigger-than-us discussion of the Great Internet News experience to come.
If you are interested, request a beta invite or drop me a line.
(Cross-posted @ Ramana Rao's Information Flow)
