iPhones for all!


I just wanted to let you know that your economics crusader is on the case to score each and everyone of you a free iPhone and to save the environment while doing it. The path became clear to me this weekend as we used our Tom Tom iPhone app to do its usual navigation around Boston. This time, it could tell us whether there were any traffic incidents on our route and advise us of alternatives. With this, I could see it all.

And what did I see? Imagine that we all had this in our cars. Imagine also (and I know this might be a stretch) that we all entered our route plans into the iPhone and those plans were uploaded into the cloud. Then all of this information could be aggregated and the ‘optimal’ route for each of us worked out so that traffic was minimised. We would then receive directions based on the centrally coordinated route and all be better off for it.

Now, for this to work, a sufficient number of people would have to have an iPhone or equivalent. I’m not sure what the sufficient number is (it is probably less than 100%). Now you wouldn’t be compelled to follow the instructions handed to you but if it was meaningful you would follow them anyway as it will likely make you at least weakly better off by doing so. That is, the routes handed down could be incentive compatible and also update in real time as others came into and out of the mix.

The point is that iPhone adoption (and consequent buying in to the centrally controlled traffic system) would involve positive externalities and hence, there would be a prima facie case for public subsidisation of them. I think it is time for a movement or at least an ‘iPhones can save the environment’ Facebook group.

16 Responses to "iPhones for all!"
  1. The possibilities are endless.  How about the ap tells me which showing to go see at the cinema? 

    I know I want to see the film as close to launch on Fri as possible, but the cinemas in town are busy. It books me tickets for the most efficient showing possible at a local place and gives me & my GF perfect driving routes to meet up there.

  2. I like the idea. But shouldn’t it be contingent on motorist first paying for all the negative externalities created by traffic congestion?

  3. Ahh, but may iPoint out that if, indeed, we all had iPhones, Apple would no doubt find a way to internalise the positive externalities, and therefore appropriate the social gains.
    Instead, the title of your post should be Nexus Ones’ for all. That way we can just view an ad or two before setting of on our ultra-quick conjestion free journey to work.

  4. This is a particular instance of the general principle that more widely shared information would increase efficiency.  My own favorite would be the immense increase in medical knowledge if all medication purchases and medical records were keyed to hospital records and death notices, providing us with unimpeachable epidemiological data for every combination of pills. I’d estimate that it could add twenty years to the lifespan.  It would also, of course, require that any notion of privacy be entirely eliminated; as would your suggestion, which is unworkable unless a central agency essentially knows where tou are at all times.  Can we agree to begin with an Australia Card before we add the electronic bells and whistles?

  5. You can pry that closed platform run by that ailing monomaniac into my cold, dead fingers 🙂
    Seriously, aside from your affection for one particular Turing-complete computing platform over dozens of equivalent ones, this is a neat idea, but if I recall correctly the mathematics of this are somewhat nastier than you might like.
    For one, I’m not sure that globally optimal routing is all that easy to compute (or even approximations); the second is that what might be globally optimal in terms of minimising total travel times may not be locally optimal in terms of minimising your travel time.  If enough people cheat, the system will probably break down.

  6. Robert,
    Yes, the “global optimal” would need to be a Nash equilibrium.  That sounds like a difficult problem to solve.

  7. This is an interesting piece. Those devices already exist – they are called Telematics units and they are cheaper and more robust than an iPhone.  It is intriguing to think that people who would not allow a Telematics unit onboard a vehicle (monitoring of location and speed etc.), yet will allow the same tracking capability from an iPhone.

    It’s all moot, however if nobody sticks to the trip’s waypoints. The traffic will end up just as chaotic as ever. Autonomous speed controls and driver assist: is there an app for that?

  8. hrgh,

    Have you seen the Melbourne PT trip planner? Any mobile service with GPRS or 3G will do the trick.  

  9. Welcome to the wonderful world of computer science, Joshua!
    The problem you are learning about is network flow algorithms. The bad news is that solving these problems on a central basis does not scale very well. The best algorithms scale linearly with the number of edges in the graph — the roads on the map. But then you’d have to run the algorithm again for every single car and constantly adjust the flow through each road for each car. That gives the problem at least O(EfV) complexity using the Ford-Fulkerson algorithm. ie a lot.
    One thing we do a lot in computer science, though, is cheat. Rather than asking for the ‘perfect’ result, we settle for a merely very good result. There are a number of ways to achieve this. In network flow problems we tend to break the network into smaller pieces and route locally. This is, give or take, why the internet works at all. And cars are conveniently like packets in that respect.
    Alternatively you could fiddle with the existing approach. Current navigation tools use various shortest-path algorithms to pick the route. That’s why sometimes you get sent down back streets instead of along the highway.
    Instead of using kilometres or miles as an input, estimated time would be more useful. This would be distance x current average speed on the road. Then each car plans its own route, taking account of reported traffic conditions.
    You may have noticed that this is a lot like a market: distributed agents making individual plans based on public information. That is not, in my view, a coincidence.

  10. @Jacques:  Better still, the government could set a congestion price (in $/hr) on each road, based on current traffic conditions.  Then motorists could use your computer algorithms to find a route that minimises the congestion charge – or perhaps a combination of the congestion charge and the driving time.
    Thus, all externalities would be addressed.  And effective navigators would not need to be given away free.  They would be worth their weight in gold.
    @Robert: does that fix your paradox?

  11. Robert M,
    Also very interesting. However it is not only the route that is important but the speed profile over sections of the route. Anybody travelling on the Monash Freeway is familar with this issue. The raised speed limit, post-construction, gets a lot of traffic to a congestion point quicker, rather than smoothing out the flow. Those risk-adverse drivers, using brakes on bends, have cascading effects on the traffic flow also. 

    Seems like the only solution is to take driving out of the hands of individuals and give it to a master controller – Skynet, if you will. It will have our interests at heart…

%d bloggers like this: