Chaik Space

This blog moving to www.mishmap.com

Brian Chaikelson's recent blog postings are at www.mishmap.com. Update your RSS feeds accordingly, please!

Twitter and Location Potential

Like the rest of tech world, I've been racking my brain around killer Twitter apps if the company further opens up their API as rumored. Twitter allows you to send web or SMS (text message) snippets to your "profile" which can be tracked by your friends on the web or with their cell phones.

My understanding is that the web service has potential for services like this:

1. Techie with too much time on their hands compiles a list of radio stations in major cities.
2. Same techie creates a Twitter profile named "radioWTF"
3. User (with less time on their hands) sends a text message to Twitter user radioWTF with a city name. Since the prefix of "d" sends a direct message, it would look something like "d radioWTF Seattle"
4. While User was frustratingly trying to compose her cryptic message using her Razr's T9 language support, Techie wrote a server side program to parse the city in the message (in this case, Seattle) and respond with the city's radio call signs, descriptions and frequencies.
5. Techie's program retrieves the message (this is the new part of the API), parses it and sends the  message back.
6. User scans the received message, identifies the frequency of the Neo-con talk show station she has been looking for, and sets her radio accordingly.

Where does location fit in? It turns out that there is already a clever hack used by the Twittervision Google Map mashup to plot people's locations on a map. Location is specified by a prefix of "L:" so I can send Twitter a message like "L:77.31, 161.5574 I'm watching penguins marching in Antarctica".  For the less geo-geek inclined, zip codes and city names can be used instead of latitude and longitude.

So here is how location can get interesting once Twitter opens up its API to enable web services. A crafty techie might create a profile named "geolookup", with an associated website (www.geolookup.com) as well. A Twitter user can create a separate (non-Twitter) account on www.geolookup.com and submit their Twitter credentials. They might select from a list of options - radio stations, weather, events tonight, coupons, etc, and also store their default location (eg. San Francisco). They save their settings and later that evening send an SMS message to the geolookup Twitter user. The geolookup web service would look up the user's account, aggregate the requested information, and send it back to the user who can make use of the information.

mobilocative markets (the new day trader)

I'm three sips committed to a double shot of espresso at a hip, tech-savvy cafe in San Francisco's Mission district, when a catchy tune comes up on the stereo system. I access the cafe's neighbor-net, which identifies the song from the playlist, referring me to Amazon or iTunes for a purchase.

Everyone wins, as the affiliate program fee covers the cafe's wireless access, providing incentive for high music quality to boot. Recognizing the song, a girl a few tables over posts a message to the cafe Wiki, listing her own Amazon cross-sells and also taking advantage of a referral bonus while plugging a similar band playing a local venue later that week.

Let's extend this idea to the "neighborprise". If I'm taking the sun on the beach at nearby pet-friendly Dolores Park, I might want to advertise my dog-walking service, encouraging people to stop by ("look for the guy with the yellow balloon") and talk to me in person to determine my level of trustworthiness. On a hot day, the corner cafe could post a bulletin declaring a lemonade happy hour to quench our thirsts. Anyone want to toss around a Frisbee?

Our just-in-time neighborhood Wiki is a transient workspace, discoverable only within a fixed proximity. Information exchange is highly targeted, and thus relevant to the users' physical context. One might access resources like (don't click on this) www.neighborhoodwiki.com?place=caferitual&key=435643534. This full url is a composition of:

• a high level url to access the web site
     
  
• a GPS coordinate or other location key (The Corner Cafe) to provide spatial context
   
• an optional key to discourage outside access if desired (perhaps a randomly generated number exposed on the wall)

Localized mobile markets emerge as individuals monetize their insight into their immediate surroundings. The content pusher is the new day trader.

Roaming Messenger for emergency dispatching

Roaming Messenger is a mobile messaging platform used to track down recipients and interact with them. David Stephenson's homeland security blog discusses a hybrid workflow/messenger application that uses the platform to organize just-in-time emergency teams (see ISERA Crew Scheduler demo video)

When an emergency needs to be addressed, the software attempts to contact team leaders using multiple devices (desktop, wireless notebook, PDA, cellphone) as defined in a priority queue. First responders are then contacted strategically, using rules such as closest location, as described in the press release:

..retrieve global positioning system (GPS) information for mobile devices, and also execute commands based on GPS location of users and devices...GPS location is no longer a separate process; it is now part of the messenger paradigm

As evidenced by Katrina, it is crucial to have capabilities for an ad-hoc response strategy that spans multiple organizations. This kind of location-aware rules-based IM technology could be used by average citizens for anything from neighborhood watch groups (alarm system goes off and neighbors are contacted to look out their windows) to picnic outings ("We forgot the mustard and you're closest to the deli!")

Book Summary: Ambient Findability

Ambient Findability emerges when "we can find anyone or anything from anywhere or anytime". Information Architect guru Peter Morville's new book (O'Reilly) explores the social impact and promise of this frontier, while also presenting the challenges to overcome in order to keep the increasing stream of information manageable.

Below, I provide a chapter by chapter summary of the book, in which chapters 5 and 6 are the real gems, in my opinion.

Chapter 1 (Lost and Found)

Morville laments the state of content, which is often un-findable by those who might value it most. Many websites take a top-down approach to search by assuming that people land on the home page and subsequently drill down. Often in practice however, our first experience with a site is in its deep-down content , and we might never even view the home page.  One strategy is making content findable is to search-optimize your site by using keywords to attract visitors looking for something more specific than the highest concept in the knowledge hierarchy.

Findability has bottom line implications. It powers the Long Tail, which describes the huge marketplace for niche products available most easily on-line, since brick and mortar stores can't stock massive inventory.

Chapter 2 (A Brief History of Wayfinding)

Through history, humans have learned to navigate environments of increasing complexity, creating wayfinding tools and vocabularies , all of which are ultimately adapted to the next mileu. At first, we found our path by referencing objects like mountain ranges in our natural environment.  Eventually, we formed urban mental models around man-made structures like temples and bridges in what can be considered our built environment. Today, we struggle to port these spatial metaphors to the web, where distance is poorly defined and "there is no there".

To promote better navigability on the web, the author introduces the idea of "findability", which is a bridge between our heritage of wayfinding in natural and built environments, with a focus on usability. This concept spans both physical and digital worlds, as the meaning  of an "object" becomes blurred.

Chapter 3 (Information Interaction)

Since Moore's Law implies that technology accelerates exponentially, it follows that we will be increasingly overwhelmed with available information. The lessor known law by Calvin Mooers' postulates that people will stay away from an information retrieval system if it is painful to use. The question must be asked: Will tomorrow's search engines keep pace with the flow of information to make people care about using it?

Because of the ambiguity of language and poor algorithms to determine "aboutness", traditional search engines have poorer  "precision" and "recall" as the network of information documents increases. Centralized assigning of metadata can help, but it is prohibitively expensive.

The paradigm of Human Information Interaction (HII) embraces social and psychological dimensions of information seeking behaviour. By studying people's behavior, perhaps we can enhance information findability!  Supporting this idea, innovators like Google, Flickr and del.icio.us have improved information retrieval by tapping into our nature for gossip and the power of popularity.

Chapter 4 (Interwingled)

Citing advances in locative technologies (RFID, GPS, etc), body/technology convergence, sensors, and other examples of ubiquitous computing, Morville convinces us of the interwingularity apparent in today's world. We move fluidly between topics and media, hyperlinking as appropriate to convey our ideas. True to Bruce Sterling's concept of "spimes" (located in space and time), objects are self-revealing and configurable. For example, an Amazon.com "book" is searchable, allowing intermingling between text, data about the author, rank, reviews, and related books.

Thus, the case for findability becomes more urgent  as our environment becomes more complex, with information about the real world being imported into cyberspace. We will strive to make good decisions on how to intermingle our lives with technology in order to make information manageable, viewing it with novel interfaces (orbs, digital paper, etc).

Chapter 5 (Push and Pull)

Ideally, we want to increase our signal-to-noise ratio to pull people, places, products and ideas into our attention, while reducing the push of unwanted messages and experiences. There should be a balance, involving a feedback mechanism - for instance, we opt in (pull) an RSS feed, which pushes information back to us. We enter search terms into Google (pull) and receive results and sponsored ads (push).

Marketing is a double edged sword, which unfortunately today tends to push information out of context. Designers and marketers should cooperate to make information and products more findable (perhaps based on users' personalization profile), so people are more inclined to appreciate a product and buy it.

In general, companies would be better served if they promoted interdisciplinary collaboration amongst different skill sets in an organization. Engineers should be cautious with technology decisions like using dynamic urls which might not be spiderable. Designers should make efficient use of image size for mobile browsing. Marketers should not push out of context upgrades or risk alienating the user. Information architects need to work with brand architects to map marketing jargon to user vocabulary. Search Engine Advertising (SEA) and Search Engine Optimization (SEO) should be owned collectively for increased findability, leading to a better bottom line.

In this chapter, Morville also describes the design qualities that shape the user experience, arguing that findability is of primary importance.

Chapter 6 (The Sociosemantic Web)

The Semantic Web promises era where search and navigation systems (i.e. agents)  bring us the information we need. Whether this vision is attainable is hotly debated by social software advocates, who argue about whether properly assigned metadata can bring us there. The author is optimistic that metadata can serve as a "boundary object", which will bring opposing camps together to build a shared understanding and encourage social progress.

Morville walks through the history of metadata, discussing taxonomies, ontologies and folksonomies. A taxonomy is a categorization, typically with a root node, to facilitate understanding. Ontologies go a step further, adding a set of inference rules. For example, RDF, an W3C standard for describing metadata, assigns properties such as "is a member of" or "is related to". So far, this promising classification technology has not met expectations. Social software on the web led to a bottom-up model, a "folksonomy", where "tagging" allows. users associate objects with keywords. The tags are shared and become pivots for social navigation, as well as a great tool for trend spotting.

Here's a brilliant excerpt (p141), which ties these compatible classification schemes together:

"For quite some time, I have believed this concept of pace layering holds great promise within the narrower domain of web design. In this discussion of metadata, the potential for unifying architecture is self-evident. Semantic Web tools and standards create a powerful, enduring foundation. Taxonomies and ontologies provide a solid semantic network that connects interface to infrastructure. And the fast-moving, fashionable folksonomies sit on top: flexible, adaptable, and responsive to user feedback.

And over time, the lessons learned at the top are passed down, embedded into the more enduring layers of social and semantic infrastructure. This is the future of findability and sociosemantic navigation: a rich tapestry of words and code that builds upon the strange connections between people and content and metadata"

Morville argues with the assertion that the "document" is becoming irrelevant, as syndicated snippets of information become more common. He introduces the abstract concept of "genre" (the combination of form, content, and purpose) and insists that it naturally follows that documents are inherently findable, and thus valuable. In fact, we are constantly creating new kinds of documents (FAQs, sitemaps, and blogs) and we will continually invent new forms of documents as new categories of objects around us become findable (eg. people, places) and mobile devices gain popularity:

"In an age of location-awareness, when metadata can be attached to people, possessions, and places, the findability and value of our documents and objects will be shaped by strange new forms of sociosemantic aboutness"

Chapter 7 (Inspired Decisions)

I found this chapter to be somewhat non-cohesive, with the author discussing ideas in AI (mentioning Jeff Hawkins' fascinating book, On Intelligence), irrational human behavior, and information overload. I think he is illustrating the difficult path ahead for us in attaining our goal of ambient findability, as  obstacles in human nature make us resistant to search for the best, objective information.

His idea of graffiti theory suggests that we are unconsciously shaped by the information we digest, and this feeds back into the information we seek. The web is both a tool for making informed decisions, but also has the power to propagate ignorance.

Map Mobs

Map Mobs will occur when people have an incentive to use mobile technology to easily SMS an information payload, along with their GPS coordinate.

"Easily" means that there is minimal or no action by the end-user.

"Incentive" implies that one's current context (location, proximity to a "micro-event", etc) is a valuable commodity in a gestalt kind of way.

Map Mob Application #1: Dodging Traffic

1-A query is received from a traffic service: "What is your speed".

2-Your car's computer interfaces with a mobile device (built-in or bluetooth), which responds to the query with your GPS coordinate and current speed. One can comprehend the "mob" when multiplying this micro-response by thousands of cars all over the city.

3-Information is received by a GIS server, which crunches the data and serves up compiled data to interested consumers. The interface might be a color coded map, or more likely an application which interprets the current traffic results and suggests the most efficient route. In "web 2.0" fashion, the information can be syndicated to relevant sources throughout the Internet.

Map Mob Application #2: The Price of Milk

1-While shopping, you SMS the price of a product of interest (with your GPS coordinate, of course) and send the information to  a GIS server.

2-The data is crunched, likely associating the pricing information with store names using the GPS coordinates as a lookup.

3-Consumers are thus more informed about the best values and increased competition promotes better prices.

This model can also publicize product inventory - no more frustrating Christmas eve trips to the toy store to buy your kid the latest gizmo!

Map Mob Application #3: Get Out the Vote

1-Political canvassers ring doorbells, identifying potential supporters who need a ride to the poll on election day.

2-Their location is sent to a central server, which uses a traffic routing algorithm to determine the most efficient way to get these people to the polls.

3-More votes are cast.

Map Mob Application #4: Psychic Pizza (or kozmo.com on steroids)

1-A market for JIT (Just In Time) services is born. Roaming your neighborhood, one can find pizza pies, geek squads (computer/appliance installation experts), doctors, "escorts", knife sharpeners, shoe shiners, psychologists, taxis, graphic designers, fresh produce grocers, babysitters, florists, make-up artists, dog walkers!

2-Their whereabouts are sent to your desktop/cell phone mapping application. As a whole, the individual vendors constitute a mob, plotted on a map according to your subscription preferences. Trusted networks will play an important role in this new economy.

3-You bid or enlist for the desired service.

Map Mob Application #5: Drinks on the House

1-You decide to take off the work edge with a last minute after-work cocktail. You SMS your location to your trusted social network (a la Dodgeball.com, but using GPS instead of predefined locations).

2-Your colleagues, who notice your plea for company on their cell phone mapping application, flock to meet up with you.

3-Your popularity qualifies your party for a group discount.

Locative Search Query Aggregator

I've mocked up the example described at the end of my last post (Prototype: Bounded Search Area), using Adobe InDesign to demonstrate a Web 2.0 interface that allows us to query multiple data sources to fulfill a location based search.

This application, available as a jpg here, consists of widgets that are assembled in a plug and play manner and use an area of interest drawn on a map to filter the search results.

Some observations:

• Useful search query widget combinations become a commodity.


• A market develops for access to subject matter experts' knowledge bases.


• Algorithms are created to "intersect" search results from multiple data sources, many of which have poor data quality. For example, if the street name of a japanese restaurant returned from zagat differs slightly from that of a traffic server, software must determine whether they  are truly the same location.

Prototype: Bounded Search Area

With all the google map mashups being developed, I started thinking about a compelling interface to specify search areas on a map - especially on a mobile device. To date, I've seen zip codes used using google local, but this paradigm is limited if you are unfamiliar with the surrounding area.

Using the google map API, I've put together a javascript prototype (tested with Firefox) that shows how one might draw an enclosing polygon to extract information.

Still a work in progress, the application echos  the descriptions of all the predefined points on the map that have been surrounded by the drawn enclosure. Using a PDA and stylus, these polygons could be drawn quite easily with a similar model.

One might envision a map (google maps/earth, etc) as a visual tool to select an area of interest. The coordinates, which specify the surrounding polygon, could be sent to various data sources, each one responsive to a unique search criteria. By using set operations (union, intersection, etc) with the polygon coordinates as common input, the user could extract useful information.

For example, borrowing from a dining example, I might be interested in "Sushi restaurants" (Zagat data server) in "Safe neighborhoods" (police beat server) accessible via "Public Transportation" (MTA server). I would probably expect the resulting result set to be plotted on a map.

Prototype: Location Aware Podcasting

I've put together a prototype which allows you to use your mobile phone to listen to GPS encoded podcasts based on your current location.  Please read System Requirements before you download the application

How it works:

1. Subscribe to podcast channels (currently hardcoded in the software) whose RSS documents have been extended with a GPS tag. For example, subscribe to the "San Francisco places" podcast channel.


2. Download the podcasts (which are just audio files) from the channels to which you have subscribed.


3. When you are roaming around, click "generate playlist" which will select the podcast most relevant to your location. The software selects the podcast whose GPS coordinate is closest to your current location and creates a playlist. For the current protoype, you must enter your current location manually, but a future version will take this value automatically from a connected GPS device (some of the newer phones already have integrated GPS).


4. Open up an MP3 player (such as pTunes) on your mobile device and select the generated playlist. A future version will automatically launch the MP3 player with the appropriate podcast.

Real world examples:

• Subscribe to a podcast channel which provides information about important architecture in New York. After downloading the podcasts, walk around Manhattan and listen to podcasts about buildings in close proximity.

• Subscribe to a podcast channel about an art exhibit in a museum. After downloading the podcasts, walk around the exhibit and listen to podcasts about the artwork in front of you. Note that an indoor application requires a different technology than regular GPS (see future directions below).

 System requirements:

• PDA/cellphone loaded with java (J2ME with CLDC/MIDP 2.0 profile and JSR 75 for FileConnection).  

• The software has been tested on a Palm Treo 600 with IBM's J9 jvm, which you need if you have a treo. You can download it here.
• External memory card

Challenges:

1. There is no readily available GPS solution (without car mount/car lighter power) for the Palm Treo 600, since their are challenges interfacing with its serial port. Thus, for this prototype, the user must manually enter their GPS location. The Treo 650 has Bluetooth capability which would allow me to get around the serial port issue.


2. The J2ME MIDP 2.0 profile is limiting. The jvm cannot launch an external application, so the user must manually launch an MP3 player in order to listen to the selected podcast

Future Directions:

• Integrate a GPS device (solving challenge #1)

• Use a different J2ME profile (ie CDC), which would permit an external MP3 application to be automatically launched with the appropriate podcast (solving challenge #2)

• Extend the GPS tag to permit a "relative" coordinate, so that podcasts can be marked with an indoor location. This would be valuable for a museum type of application where the user can listen to information about various pieces in the exhibit.

• Extend the GPS tag with a "discoverability radius". Thus, a podcast would not be discoverable if the user is too far away, even if the podcast is closer than any other in the vicinity. This would be useful in a museum type of application where a podcast about a particular artwork should not be selected if the user is at the other end of the room.

Responsive interface

Take a peak at the top 20 tags on del.icio.us and you'll notice a number of topics related to computer interface.  Just below "mac" and "javascript", you'll find "ajax", which is web buzzword du jour.

The term Ajax was coined by Jesse James Garrett last February in an adaptive path essay.  Ajax is short for Asynchronous JavaScript + XML, and is not so much a new technology as a recognition that the web should be more responsive. Instead of the usual "click submit/wait for server to respond with updated web page" model, ajax web pages use a javascript xml call to interact with the server while the page is still active. The result is an internet experience which feels more like a desktop application.

The most visible ajax web site is clearly the much hyped Google Maps, google's locative play, which is even more exciting when combined with clever user hacks like housingmaps.com and chicagocrime.org. I would imagine that they have big plans for the PDA/cellphone market once connection speeds get faster and GPS becomes ubiquitous in these devices. An Ajax webpage model can also provide a more tactile online shopping experience - try adding a t-shirt to your shopping bag at panic's web store.

A more responsive interface is a useful paradigm for mobile devices, which in my own experience suffer from sluggish response making me more likely to keep them in my pocket than in my hand.

Social network capitalism

Jyri Engeström borrows from sociologist Karin Knorr Cetina's concept of "object-centered sociality" to explain the successes of certain social network communities and the failures of others to capture our imagination. He proposes that many of the first generation social networking websites (the previous version of Linked In is provided as an example) assumed incorrectly that people would maintain interest for the social ties themselves.

Expanding on his perceptive blog entry, one might argue that the more useful communities provide an "object incentive", whether it be a job (a revamped Linked In), photo (flickr), hookup (friendster, to some degree), event (tribe), destination (urls for del.icio.us; spaces for dodgeball) or even a creative fix with the added incentive of cold cash (second life).

People orient their online affiliations inline with their wants. I suppose that's a wide spectrum - although for some people, their desire may simply be who-knows-who popularity. Pioneering sites like friendster will find success with that limited audience, but the rest of us, who are looking for a more focused fix, likely fall in the "last login early 2003" category. To be fair, the allure of popularity can be very lucrative for some demographics - for instance, South Korea's (Cyworld is a commercial success with virtually every 20 something spending real currency to buy "acorns" to fill their virtual rooms (mini room) on their homepages (mini-hompies) with furniture and art.

Jyri suggests that by identifying new objects that society finds useful, we can predict other valuable social networks that full the social networking spectrum. In the spirit of this particular blog space, he mentions spatial annotations, while describing the technological limitations (limited GPS capable cameras) which will force us to wait awhile for services that will provide useful spatial context. Similarly, Jyri comments that people might find “proximity announcements” to be useful as NFC-esque (near field communication) tags become more easily discoverable.

Credibility may be another object which fills a void and around which communities form. For instance, eBay allows people to rate sellers, while other sites rate medical practitioners. Is street credibility truly the object which fills the void and provides incentive for a social network, or is it a proxy to the final product - the auctioned item or good health in the aforementioned examples?

micro-events

Why wait for the next generation of cell phones to integrate GPS? Justin Martenstein's notebook points to Psiloc's miniGPS, which uses a cell tower signal hack to trigger location based events with your mobile phone.

For example, you can set a micro-event to wake you up from your after-work snooze as your commuter train approaches your destination. By using location instead of clock time, you can gain an extra few minutes of shut eye if your train is delayed (and saves you from missing your stop if it arrives early!) . 

Another application might be to key in the location of your local movie theatre to automatically switch your phone to silent, preventing an embarrassing  mid-show jingle.

Alice Waters and RFID

Wouldn't an organic food company's marketing department be thrilled by a model in which each link in a product's life cycle is documented and available to the public? Perhaps this can be accomplished with a combination of RFID and location aware technology. Just reference the product barcode to hear the story of how your morning cup of coffee found its way to your kitchen table:

[Montevede, Costa Rica, May 29, 2004] Hola! My name is  Andres Guadamuz and I am a worker on a local coffee plantation, responsible for maintaining machinery used in gathering coffee beans. Read more about our operation at http://www.cafe-monteverde.com/default.htm.

[Transamerica trucking company, Northern route, June 3, 2004] I left San Jose, Costa Rica on Monday, and I am on schedule to arrive in Los Angeles on Saturday. Gasoline prices have recently risen by 12 cents a gallon, and we are trying to insulate you, our customer, from the increased prices by reducing the weight of the packaging of our cargo.

[Monteverde coffee packaging plant, Los Angeles, June 9, 2004] This bag of espresso coffee bean is best if used by September 3th, 2005.

[Whole Foods, San Francisco, July 19th, 2005] Hi! I'm Jennifer, your cashier, and I'd like to thank you for buying organic foods. By supporting the Monteverde Coffee Company in Costa Rica, you help provide jobs and support sustainable agriculture, while enjoying an exceptional product! Rest assured that this RFID chip will now deactivate itself in order to protect your privacy. If you've enjoy this product, Rainbow Foods will give you 5% off your next purchase.

Maybe "supply chain narratives" like these will make us more informed and compassionate consumers, as we recognize that we live in a global village. This model isn't limited to produce - it can be applied to other industries like textiles or services (what really happens when you drop off your dry cleaning? )

Measurability policy

For years, people have anticipated a wave of automated agents that monitor the Net, alerting you to topics of interest. To some degree, this concept is alive. Google Alerts keeps tabs on selected topics; Amazon and Tivo mine their huge databases with collaborative filtering technology to offer suggestions of products and broadcasts which you might enjoy.

Likewise, successful mobile locative platforms will understand the nuances of their host user. Software will observe the way we behave and will adapt to our lifestyles, potentially interacting with trusted services around us:

• When I go to the theatre, I would expect my cell phone to turn to silent mode (but only when the production begins)
• If I tend to stream Jazz radio, I would welcome a message that indicates that a show is scheduled this evening in my neighbourhood
• Is traffic at a snarl this morning? Would public transport get me to work quicker (or is someone trying to organize a last minute carpool so that I can use the carpool lane?

Software platforms will fail if they implement a "measurability policy" that feels intrusive. Like the Internet, people will provide a certain quality of information in exchange for a useful service.