July 31, 2006

EXTRA! EXTRA! Newspaper sites to start linking to Web sites

Filed under: media, newspapers — Rakesh Agrawal @ 9:59 pm

The New York Times reports today on the launch of a service by that enables news sites to automatically link to related stories.

Newspapers have long been reluctant to link to other sites for fear of losing traffic. Never mind that Google built a $117B business primarily by linking to other sites.

It irks me whenever a news story refers to online content without a reference.

The wire services, which provide much of the content on newspapers and news sites, do a terrible job. Take this AFP story about a woman kicked off a Cathay Pacific flight for refusing to stow a Gucci purse. The last paragraph says:

“The incident, captured on camera by another passenger and published on the Internet, resulted in a delay of the flight of more than an hour.”

There is no URL anywhere in the story. I’ve searched flickr and YouTube with no luck.

About these ads

Personal mashups – letting users re-draw the map

Filed under: google, local search, maps, mashups, web 2.0 — Rakesh Agrawal @ 4:39 pm

The Google Maps API has been a runaway success, allowing developers around the world to create mashups marrying Google maps with data on a variety of topics, including craigslist real estate, taco trucks in seattle, flickr pictures, cellphone reception and crime in Chicago.

Unfortunately, this model has a few problems:

  • It requires a developer’s skillset. A user without technical skills can’t create their own mashups.
  • It isn’t easily discoverable. You can do a search for “google mashup”, but that’s hit or miss.
  • You can’t overlay multiple data sets. If you’re looking for the burrito bandit and want to overlay taco trucks and crime, you’re out of luck.
  • It scales at the rate the developer’s hardware can scale. Some of the more interesting applications I have seen have buckled under load.
  • Users have little incentive to contribute data because the site developer may disappear. I spent a lot of time geocoding pictures in flickr for use with a site called, which in a previous incarnation allowed you to plot flickr pictures on Google Maps. Unfortunately, the site died when the developer was hired by Yahoo!

I’d like to see a point-and-click interface that allows users to create their own points of interest on a map. These POIs would automatically appear when the user generates a map. The obvious POIs are “Home” and “Work”. Some of the other things I’d use this for:

  • Restaurants I’ve visited
  • Road trips I’ve taken
  • Pictures of places I’ve traveled
  • Homes of friends and family
  • My favorite places for tubing

Personal POIs would also be used for defining origins and destinations when looking for driving directions, e.g. route from “Home” to “Grandma’s house”.The POIs would be syndicatable, with access controls. I could choose to keep the homes of friends and family private, share my favorite places for tubing with friends and share restaurants I’ve visited with anyone who wants to see them.

You could use the collective knowledge of the Web to create maps of things like:

  • Speed traps (imagine getting a route that highlighted potential speed traps)
  • Subway stations
  • Traffic and red light cameras
  • Fastest, best roller coasters
  • Mel Gibson arrests
  • Giant balls of twine near the interstate

When on a map, I should be able to see a list of the most popular data feeds available for that map, with the ability to search others. I would also be able to subscribe to specific users or tags to automatically see their content.

I should also be able to have feeds from external sources appear on my map. For example, I maintain restaurant reviews on and an address book on Plaxo. The most up-to-date data from those sources should appear on my maps.

You can do some of these things with and Google Earth’s KML files, but I have yet to see an integrated solution in a user-friendly package, with the scale of users necessary to have compelling data sets.

July 28, 2006

Data entry on mobile phones

Filed under: google, mobile, wireless, wireless data — Rakesh Agrawal @ 6:09 am

There’s no question that data entry on mobile phones with the standard 12-key keypad is tedious. Technologies like Tegic’s T9, which is available on many phones makes this a lot easier. The technology predicts what word you are trying to enter by comparing the possible combinations against a dictionary.

On a standard phone, if you wanted to type the word “pizza” you have to enter 7-4-4-4-9-9-9-9-9-9-9-9-2. With T9, you enter 7-4-9-9-2. That’s a savings of 8 key presses, less than half the number of keys.

Yet many applications make you do all the hard work. On my Samsung A900’s address book, to get to Doris, I have to enter 3-6-6-6-7-7-7… until the phone’s typedown feature gets me to Dor.

Likewise with Gmail’s mobile application. I can’t search against the address book by simply entering 36747.

This despite the fact that Google has very smartly purchased

July 27, 2006

Google mobile maps adds live traffic data

Filed under: google, gps, maps, mobile, satellite navigation, wireless, wireless data — Rakesh Agrawal @ 9:04 pm

Google’s mobile map application added live traffic data earlier this week.

It is the best mobile application I’ve seen to date. With all the terrible WAP sites out there, it’s great to see a wireless data application that works and works very well.

I’ve been using Google mobile maps (sans traffic) on my Samsung A900 for about four months. You can quickly find businesses and get directions. You can pan almost as smoothly as you can on the Web version of Google Maps. You can even pull up satellite images.

Now you can also see traffic conditions on some roads in major metro areas. Roads are color-coded red, yellow or green based on congestion. (Unfortunately, the traffic overlay also covers up the road names in my tests.)

It’s a neat trick, though I have to wonder how many traffic accidents will be caused by people trying to check out traffic conditions on their mobile phone.

This release also adds the ability to store favorite places and routes, a must in the mobile world where data entry is tedious.

Routing is the best I’ve seen on a non-GPS enabled application. You can step through each segment of a route so that only the most pertinent information is presented. It will also estimate the amount of traffic delay. The one major flaw in the routing is that it doesn’t automatically adjust the scale of the map when you have a series of maneuvers that are close together.

July 25, 2006

Time of day and online maps

Filed under: cars, google, gps, maps, satellite navigation — Rakesh Agrawal @ 4:45 pm

If you follow the online driving directions from Mapquest, Google Maps or Yahoo! from Reston, Va. to Washington, D.C., in the morning, there’s a very good chance you’ll get a ticket.

Why? All three route you on I-66, the eastbound lanes of which are restricted to HOV-2 traffic in the morning.

My car’s navigation system gets around this problem by ignoring I-66 for most routings.

According to NAVTEQ, the company that provides map information used by the online providers and Acura:

“The NAVTEQ map data we supply quarterly to our navigation system vendors represents the direction of travel changes in the I-66 High Occupancy Vehicle (HOV) lanes to accommodate the morning and evening commute. While the timed carpool coding in the map data is coded to reflect reality, it appears that some navigation systems avoid these roads altogether, thus causing non-optimized routing.” (I give NAVTEQ kudos for providing a detailed response to my question.)

Another road in the DC area, the Rock Creek Parkway, is ignored by the big three online sites. The road runs in one direction during the morning and another in the evening and is two way the rest of the time. On weekends, the online routing turns a pleasant drive through the park into a convoluted route through many traffic lights and DC’s confusing traffic circles.

Because none of the systems factor time of day into the routings, you end up with illegal or suboptimal routes.

Ideally, the online providers would include time of day in calculating routing. In addition to HOV and other restrictions, they could utilize historical traffic data to provide more optimal routes during rush hour. e.g. avoid the Capital Beltway at all costs.

At a minimum, the online providers should put specific, prominent cautions on routes that use (or avoid) time-restricted roads and offer the user the ability to request routes that avoid (or use) these roads.

July 24, 2006

Reinventing satellite navigation

Filed under: gps, maps, satellite navigation — Rakesh Agrawal @ 3:28 pm

Satellite navigation is one of the most expensive options on new cars today, costing around $2,000. But it is also one of the least flexible and most quickly outdated items in the car.

Although your position is determined by triangulating signals sent by satellites, the information used to generate maps, routes, business listings, etc. sits on a DVD in the trunk of the car.

Manufacturers typically release one DVD a year that captures the updates. (Acura charges $180 for each update.)

This approach has a number of problems:

  • Businesses close and new businesses open continuously.
  • New roads are built.
  • Roads are closed.
  • Mistakes in the database can’t be corrected until the next release.
  • The system has no ability to learn.

Another way to do this would be to store the information on a hard drive or flash memory and continuously update the data via satellite. XM currently offers NavTraffic, which streams live traffic information to the car’s display, on cars like the Acura RL. A similar system could be used to send data updates that are then written to local storage.

This would make it possible to update on a much more frequent basis. If you had the data, you could theoretically include temporary road closures for special events.

The system could also learn from the user’s behavior. For example, they’ve put in a number of new interchanges around my office. As I drive on these new roads, the system could add them to available routings. If I frequently ignore certain routings, the system could adjust future routes.

Lighting the way to the throne

Filed under: hotels, travel — Rakesh Agrawal @ 3:05 pm

I stayed at the New York Hilton this weekend in one of their renovated guest rooms. One piece of technology in the room caught my eye: a sensor-activated light. You step out of bed and LEDs at floor level automatically shine a light across the room. It doesn’t light up the whole room – it’s just enough light to help you find your way to the bathroom in the middle of the night. The light also has a light sensor so that it doesn’t light up when the room has enough ambient light.

It’s a very thoughtful touch.

July 21, 2006

Unleashing the power of the buddy list

Filed under: facebook, im, instant messaging, social networking, web 2, web 2.0 — Rakesh Agrawal @ 12:28 pm

I’ve been using a number of different social networking sites recently – LinkedIn, MySpace, Flickr, AIM Pages, Digg, Netscape, YouTube, Yelp, Netflix. With every one of them, I face the same problem. When I join, I’m all by lonesome. (Except for my good buddy Tom.) I’m staring at a big, blank “Friends” page that screams “you’re a loser”.

For every application, I have to re-create my list of friends. If I forget to invite someone (even if they’re a user of that service), I don’t get access to their content. When I meet someone new, I have to add them to various sites. And if I have a falling out with someone, I have to remember to get rid of them everywhere.

This is despite the fact that I maintain a master list of friends that every one of these applications could leverage – my buddy list.

When I sign up for a new application, say YouTube, I should be able to tell YouTube that my AIM Screename is johndoe32. YouTube can then find out from AIM who else on my buddy list is a YouTube user and automatically show me their content based on the permission settings they’ve applied to the content.

YouTube could also incorporate their AIM presence as I’m browsing through content. If I see a video from a friend that I like, I can IM them right from that page.

If I want, YouTube could also use AIM to alert me of new videos posted by my friends.

This is a huge win, win, win:

  • Users benefit from streamlined friend management and easier access to communities.
  • Applications benefit by lowering the barrier to entry, making it much easier to adopt new products. Startups can get to market faster because they can offload the work of managing user relationships. Applications also benefit from the alerts on the IM platform.
  • The IM platform benefits with increased distribution and visibility on the application sites.

July 20, 2006

The remote control lockbox

Filed under: audio, consumer electronics, remote controls, web 2.0 — Rakesh Agrawal @ 5:06 pm

I recently purchased a Harmony 880 universal remote. It’s one of those gadgets for those who have a lot of gadgets. The promise is that you can consolidate remotes for a lot of devices into one. In many ways, the remote is a model of good design:

  • It feels well balanced in the hand
  • When you pick it up, a tilt sensor automatically turns on a backlight to make it easy to see in the dark
  • For most people and most devices, the Web based set up process should be relatively easy

The Harmony’s key flaw is that the code database – the instructions that tell the remote what commands to send – is controlled entirely by Logitech.

When I tried to set up my other new purchase, the Squeezebox 3, I got a very limited feature set for the music server. The selection of commands that the remote offered were most likely made by someone who had never seen a Squeezebox, much less lived with it and used it for a few weeks.

A couple of weeks later, I’ve been able to manually configure my remote with the key features of the Squeezebox. Yet this data is locked in my remote. There’s no way for me to export that data and post it for the world to use, despite the fact that the configuration is entirely Web-based.

So who cares? A Harmony 880 and Squeezebox 3 is serving a niche of a niche.

But that’s exactly the point. By leveraging the knowledge of A/V enthusiasts on the Web, Logitech could provide a much better experience to a lot of niches, like Philips does.

July 18, 2006

Show me the good stuff

Filed under: email, spam, web 2.0 — Rakesh Agrawal @ 6:11 pm

When I check my email, I scan the list of new messages for names of people I know.

This is something that machines can do much faster and better. With the volume of spam and bulk mail these days, some of the vast effort spent on reducing spam should be spent on showing users the good stuff.

You can do this the hard way today with many mail services by creating filters for each person you want to highlight (or move to a different folder). This is usually a multi-step process; few people will create and update these filters.

There’s a simple, high value way to accomplish this: Check the messages against the email addresses the user has sent email to and the user’s address book, then highlight the ones that match.

If the user has categorized the address book, you could also color code the highlighting to indicate whether the message is from friends, family, co-workers, etc.

Older Posts »

The Silver is the New Black Theme. Blog at


Get every new post delivered to your Inbox.

Join 227 other followers

%d bloggers like this: