Blog Post

Using the iPhone to Mine for Gold & Sense

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Our growing ability to use the Internet as a giant database, apply that information in a creative way to build interesting mash-up applications, and then apply them to markets — stock, real estate or fantasy — is an area that holds a lot of fascination for me. But while a few efforts have produced rudimentary, data-based mashups that are good, so far none have been truly game-changing.

We’ve already showcased two startups — Skygrid and Placebase — that have impressed us with their ability to offer pointers that can be translated into actions on the real-world stock and real estate markets. Trulia and Zillow fall in that category as well, though I think they’re both eons away from where they should be. And today we’re adding New York-based Sense Networks to the interesting and growing list of intelligent mashup companies.

By combining historical and real-time location data acquired through either GPS and Wi-Fi, along with other real-world information, the company has come up with a “social navigation and nightlife discovery application” called Citysense. The mobile app, which runs on Blackberry and the iPhone and is currently limited to San Francisco, shows nightlife hot spots on a map in real time. You can then drill down to find information on, say, local bars and restaurants. But Citysense itself is actually a small part of a bigger story.

It runs on Sense Networks’ platform, called Macrosense, which has the ability to take geo-location data sent out by phones and vehicles, such as taxis, and map it to historical data, such as old traffic patterns, local restaurants and other geographical information. I would describe Macrosense as a machine-to-machine platform that can mash up many inputs to create real-time “heat maps.”

This is where it gets interesting: The company doesn’t want to take any advertising or charge people for the application. Instead, it wants to take the trend information it’s gathering and sell it to investors who want to trade based on that information — which is understandable, given that the company has been seed-funded by money from a hedge fund. And I like this idea, even though I have some concerns about privacy. The company says their system is based on “anonymous, aggregate location data.”

“Citysense demonstrates the power of combining anonymous, aggregate location data for social navigation,” said Sandy Pentland, chief privacy officer, co-founder of Sense Networks, and director of human dynamics research at MIT. “The idea is similar to automobile GPS systems sharing and pooling current road speed conditions so that everyone can avoid congestion.”

I’m still not entirely convinced. But if we put privacy concerns aside for a minute, the possibilities of this are mind-boggling. Imagine mapping foot traffic to, say, Gap or Apple stores. While it would never tell you if people were shopping or not, it would be a great indication of how hot (or not) the store was, enabling you to trade on the information. Take it one step further and mash it with web-based data or Twitter feeds: You could build a highly complex and near real-time view into what’s happening on the innerwebs.

Which reminds me: It’s time to call my buddies Paul Kedrosky and Tim O’Reilly so I can pick their brains about these trading mashups.

15 Responses to “Using the iPhone to Mine for Gold & Sense”

  1. Recently I browsed around about the location-based services (LBS). Most people believe it would be the next big thing or killer app. Quite a few others have different opinion. e.g.,

    Here I can possibly present one opinion from the consumer/end-user perspective, which I have posted in some other places too.

    Do we need LBS so badly?

    Before I really go to the details. Let me give a review of one simple concept and theory here, which are called “Home Range Concept” and “Traffic Pattern Theory”.

    Home Range Concept. It is a concept that can be traced back to a publication in 1943 by W. H. Burt, who constructed maps delineating the spatial extent or outside boundary of an animal’s movement during the course of its everyday activities.

    Traffic Pattern Theory. A people’s daily activity pattern is pretty regular, which comprises of several major events, such as school, work, home, shopping.

    As I remember, a technical explanation of traffic pattern theory can be found in a report by Stefan Schonfelder, STRC 2001.

    What happened here is if you are looking at the traffic pattern of a person, saying a full-time employed, 45 years, car, 3-person-household, one child, the regular activity route is so LIMITED. So, does this mean …

    A more detailed explanation of LBS for mobiles can be found by

  2. George

    How is the issue of privacy being addressed? is this service running in the background of the phones that enable it?

    It can be extrapolated out of actual cell phones to track other things. Like RFID on steroids, using the cellular networks as a way to report back to a server

  3. Andrew

    The idea is basically sound in the abstract and quite a few hedge funds do have an interest in this kind of thing, but it usually does not scale in practice. Conventional spatial data platforms are deeply inadequate for managing the volume of data involved, which can easily run into the petabytes per day. Consequently, most of these companies will fail to really scale up in an interesting way, as the limits are surprisingly low (mere terabytes for most platforms). Making this concept work essentially requires solving an old algorithm problem in theoretical computer science, which would allow someone to build a Google-like cluster generalized to spatial data (which is a superset of text data — it would be a potent research breakthrough). Anyone who has had to wrangle modest amounts of spatial data for analytics know just how difficult this scale-up issue is, and there is a bit of naivete among some location-based startups with respect to this issue.

    As far as I know, the only startup venture involved in the spatial aggregation area that is interesting is SpaceCurve, and then only because their spatial platform is supposed to be based on a very general solution to the theoretical computer science problem mentioned above, giving them some unusual leverage. There is a lot of high-value mining to be done on this data but you have to be able to absorb it first, which if it was easy Google would have already done it years ago. That Google apparently can’t do it, despite the desire to do so, should raise some red flags with respect to many spatial data startups that are entering the market without a clear idea of how they are going to address this issue over the long term — it is not a case of being solvable by throwing money at the problem.

  4. A different approach to this is to use a social media monitoring and analysis tool like ours (sorry but it’s very relevant to this post) to track conversations about a potential investment across social media: blogs, Twitter, forums, UGC, social networks, etc. You enter keywords in categories including competitors, sectors, brands etc. The results are real time so trends and memes including sentiment, demographics and geo-location can be spotted far before they hit ‘traditional’ web media. These tools are subject-agnostic rather than being focused on investments, real estate or other vertical social media tools- and the data is anyone’s to track, keep and analyze.

  5. Trading is the tip of the iceberg: the real power comes in being able to use this data to make mobile and localized advertising across old and new media almost as measurable (and real-time responsive) as AdSense. This leads potentially to conducting localized, statistically valid advertising experiments with old media like billboards, print, radio, and local cable.