Geo Services

There’s no lack of geo-aware tools out there: fireeagle, brightkite, latitude, shizzow, foursquare, loopt, more I don’t remember trying and tossing, and I’m sure a bunch more I’ve never heard of. I tried most of them just this year, as most are pretty new.  But they generally lack users, at least compared to Twitter, who’s new to the geo game. Just try importing your friends from gmail, twitter, and facebook with some of these services, and you’ll see what I mean. Latitude will be interesting to watch — given android’s impending ubiquity — but Google’s may be too big to be the real innovator in this space (despite all their leverage).

Mobility

I wonder where we’re going with mobility.  I gather that the majority of Twits are still using non-browser-based Twitter Desktop Clients (TDCs).  Maybe the sea-change to smartphones is underway.  But I’m not sure. Granted, this luddite got a MyTouch a few months ago, and I’ve got latitude and foursquare (and other random android apps crashing my phone) constantly sending my location around (after giving up on brightkite and shizzow clients).  But I know a lot of (predominantly AIR?) desktop apps are still overwhelmingly popular; they’re just lacking something…

The Problem With Desktop Geo-awareness

Quick background. Smartphones now have GPS, and it tends to be readily available (and now easy enough to turn off to save batteries) and incredibly accurate, often pinning you to the residence/office/coffee shop/place you shouldn’t be, you’re actually in. That constant information stream can really open some doors to innovation (e.g., geo-time-plot-animating your week). It’s the impetus for all those geo-services. Also, browsers like Firefox 3.5 can pinpoint you with similar accuracy with a little magic that I haven’t completely figured out.  It seems to have to do with IP lookup combined with unique MAC address locations and a nice relationship with Google who’s probably got MACs/SSIDs for most routers in your neighborhood. Probably collected while taking pictures of your street. So at least one bleeding-edge browser has it. Some others will figure it out too over the coming years.

But what about all those TDCs? When will they really become geo-aware? I tried out Python Twitter Tools today and it actually supported the geo-tag features just introduced into the Twitter API. I’m expecting it’s one of the few language libs with support as of today.  It’s really simplistic and naturally adapts to API changes (for better or worse), unlike the de facto python-twitter whose recent patches we’re insufficient and thus far ignored (by all but me?). So pretending to be a TDC I wanted to figure out how to send accurate geo-info alongside a tweet.

The TDC scenario is that we have no GPS and we’re outside a browser. That leaves us with a local IP address, like 192.168.1.X, and some mapping APIs. You can determine your un-NAT-ed IP most easily with somthing like:

curl www.whatismyip.com/automation/n09230945.asp

Then you can make use of GeoIP, which (along with its database) is installable via apt-get and some path-tweaking (or a bunch of other commercial offerings) to determine your approximate location, for widely varying degrees of approximate. I understand this approach can be way off outside the US, and my testing shows it’s not too great here either — something like “in the vicinity” 80% of the time.

For serious geo-tracking (illnesses, events, friends, advertising, etc) that’s usually not good enough. So I’m perplexed — what are all the TDCs going to do to make use of this compelling feature? Seriously, leave a comment if you’ve got some insights.

Redundant and Confusing

Since Day 1 Twitter has offered a “Location” field, to represent the city you live in (I guess). It’s been bastardized by various mobile clients to reflect a user’s current location.  Through the API a client can mess with a user’s profile to be constantly updating the Location field when she tweets. It apparently wasn’t designed to be used this way (or else why are they rolling out this shiny coord-only geo stuff?), but it has worked. In fact it works just about as well as the new-geo-facilities. But now we have two ways to do it.

The difference is that the new formal mechanism seems to enforce (or just encourage?) GeoRSS/GeoJSON (which have their own odd inconsistencies, like disagreeing on lat/lon and lon/lat). And it’s on a tweet-by-tweet basis, so the location is associated with the tweet itself; not the user (picture the schema). That’s really cool to be able to see where a user has been as he sends each tweet. But it would be a lot better IMHO if Twitter enabled geo updating without the tweets themselves (no, @devnull is not available for constant dumping of DMs with geo-coords). Then it might supplant any of the other new geo-services. I find it a little annoying to see floods of “I’m here!” tweets from the geo-aware-ati(?) I follow. So new-geo is for tweets, and I guess old-geo (Location field) is for users, and should be kept pretty static, maybe conventionally indicating where the user lives permanently.

I wish the intended uses of the fields had been made more clear. For now, I’m just hoping that the TDCs figure out how to get accurate geo-info, and start making use of new-geo. I’ll try to be tweeting from a mobile client like Twidroid that supports new-geo. And I’ll continue to let Latitude and FourSquare track me automatically, in hopes that their gathered data becomes useful. Eventually I hope to see a massive adoption of geo-aware clients so we can start to do real things with all that data (assuming we can continue to mine it en masse). I expect that Facebook will again be playing fast-follower once this really catches on.

But then it started to click. LinkedIn did some things to make it fun and useful. Just like Facebook (which I’m only just starting to use, and don’t plan to use much), it encourages users to build up their connections. I started to find friends and colleagues, and it was kinda cool to see what they’d been doing with their careers. But that wasn’t quite enough to make it compelling. Then I noticed a little green-bar indicator telling me that my profile was only 55% complete. What does that mean? Well, I better fix that. Do I get some special status when I get to 100%? Let me just fill in the rest of the education section, and then add some other jobs. Okay, cool, now I’m at 70%. This silly gimick (the green bar) was making me want to keep adding information.

It kindly told me that I’d need to add some “recommendations” to increase my completion status. Okay, what’s a recommendation. I started browsing around some of my connections, and saw that my buddy, Son, had like 20 recommendations. Wow, people must really think he’s great! But I can’t just go asking people to say nice things about me. Guess I’ll try writing one about a former colleague. Hmm, it’s pretty hard to find people to write nice things about. But I chose someone, and wrote a short blurb about how nice Wendy was to work with. Whoa, that felt good! She was a little surprised to see it, but I explained that LinkedIn compelled me to do it.

After a couple more, I caught the bug and started writing nice things about more acquaintences. Now I’m really feeling great. This wonderful excercise caused the following to happen:

• Reflecting back on positive experiences in my career
• Remembering the people who were fun to work with
• Thinking about business relationships in a new way, focusing on tangible positive events
• Making others feel good and motivating their positive behavior
• Encouraging others to make something positive of each relationship

Maybe someone will write back something nice about me. Admittedly, that was part of the initial selfish motivation. But really, I don’t care much if I get the reciprocation. It just feels good to capture a positive relationship you’ve had at some point in your career.

There are still various LinkedIn features that are not obvious to me — why am I a member of these groups?? But at this point I have to say that LinkedIn is doing some things very well.

Business idea: build a social system (or add-on) to encourage sharing kind relational reflection.

But I’m glad I attended in person. I probably won’t make the trip next year since I now understand the whole experience: highly successful geeky speakers (that you can even talk to!), lots of ambitious young people, fancy setting, and pretty nice weather (compared to Portland).

The two biggest benefits of physically attending the show were: meeting people like me, and being inspired (by the vibe) to go out and take the plunge.

In chatting with fellow geeks, I learned that geeks have wild ideas. None of the ideas I heard resonated with me as things that I’d really want to team up to do. I wonder if people felt the same way about my ideas. I’ve been thinking about ways to organize groups in Portland to try and do some brainstorming. Looks like startupdrinks might be getting some traction. My other idea is to just start showing up and introducing myself at user interest groups, like those focused on Ruby, Python, and Linux.

There was not full concensus in the messages — I need to use my own judgement in rolling up my take-aways. So here is my view of key points from the talks in summary:

• Big companies make lots of acquisitions — lots of startups aim to be another.
• Raising money: as late and as little as possible. But when the time does come, get twice as much as you think you need.
• There are some business terms and concepts you should get familiar with, but don’t worry about this too early — hackers are much more important than business types. Focus on hacking.
• Get a good team of engineers (I think two is optimal for most cases).
• Make something people want.
• Build your business plan around benevolence — when times get tough, you’ll need a “good-will justification” for maintaining morale and pulling out of it. When facing adversity, “getting rich” is not a strong enough motivator for success.
• Do whatever is best for your users.
• Find a good wave to ride — it’s very hard to create your own.
• Tie your problem to money already being spent.
• Find a way for your technology to be disruptive.
• Target a market who has their hair on fire — they will be happy with whatever you build them.
• Setting a price comes between building a great app and profiting — don’t forget to have a price.
• Target the Fortune 5,000,000. Lots of businesses only have 1-5 employees — they are a huge market.
• You don’t have to be viral.
• Focus on solving real problems — ones that you face personally.
• Find an enthusiastic user.
• Focus on making 100 users happy before launching.
• Be lazy and selfish in implementing user-requested features.
• Iterate rapidly.
• Leverage scalability of Amazon and Google elastic offerings.
• Become a part of the community — start blogging (personal and business)
• Be part of important discussions.
• Embrace criticism.
• Data is more agile than code — find a way to generate and leverage it.

My impressions in summary:

• Talks are great, but they are wholly available online — if you don’t attend you really should spend a day giving them focused attention
• Lots of speaker messaging is repeated one year to the next
• Speakers are not all on the same page (Sequoia vs 37Signals)
• Young persons can be successful if they are motivated; but older fellas should not feel excluded from entering the race
• You can meet some friendly geeks and maybe even make some useful connections
• You need to break out of your shell and talk to people in order to meet your future partners
• People actually use Lisp for project code (I’ll be blogging about getting started with Lisp soon)
• There is a lot of startup activity in The Bay, but many believe that you don’t have to live there to get started (my naive self included)
• There needs to be better planning for pizza (my lunch turned out to be a Snickers and banana)
• You can be inspired to take the plunge

If you haven’t been to Startup School, you should get prepared and go.