Optimizing 7 comments

24 May 2005 @ 17:44, by Flemming Funch Robert Cringely had a couple of articles recently, DayJet May Be the First Peer-to-Peer Airline and Jet Me to Work DayJet Software Could Revolutionize More Than Just Air Travel. Essentially about a private jet company called DayJet, which has made a piece of software that makes it economical to provide relatively cheap business flights from just about anywhere to just about anywhere in the U.S. on cheap jets that have seats for just 3 passengers. They plan on having between 40 and 300 planes like that. The big deal is their optimization software. All airlines have to do a big optimization calculation, to figure out how best to utilize their aircraft, and transport the most people to places they want to go, for the lowest cost. But even for a major airline, the number of planes and routes and destinations is relatively small, and they don't change their routes very often. What DayJet has in mind is a much huger set of optimization calculations, done not every month, but every time somebody wants to go somewhere. The use of the planes will change all the time, and any little airport might be used, as long as the jets can land there. So, the whole calculation needs to be done basically for each customer, in relation to whoever else wants to fly around that time. You know, it isn't economical to fly just one person, so the optimization will figure out how best to serve everybody and fill up the seats as much as possible. I don't think I'll ever fly in one of those seats, so that's not why it is interesting. What's interesting is the general possibilities of optimizing a whole bunch of things, if you have the software and the computing power for it. Our society is a rather inefficient machine in many ways exactly because we don't have very efficient means of coordinating what is available with what is needed. If I have a book I've read and I don't need it any longer, why don't I give it to somebody else who'd like it, particularly if they live just around the corner, and they can come and pick it up this afternoon? Because we don't know of each other, that's why. And there's no good means for us to do so. I'm not going to go around and put up notices on the lamp posts for my book, that's too expensive in effort expended. And if everybody did that, there'd be too much information to wade through. But software could do it. It is a matter of representing what is available and what is needed in some kind of useful manner, and then it is an optimization problem. Why don't everybody on a street share a small number of lawn mowers, instead of having to maintain their own? It could be much cheaper to share. The reason is that it is too much trouble to coordinate the sharing, and figuring how the right number of lawn mowers and the scheduling. In the absence of some kind of optimization program, it is just much easier to each have our own. Why do I need to have my own bicycle or car? If I could always find an available vehicle standing close by, and it was reasonably cheap and somebody took care of the maintenance, it would probably be more economical for everybody that we shared. And, indeed, there are now companies that deliver that service in various major cities. Cars or bicycles with GPS systems, coordinated by optimization software. Really, our whole society is not much more than one big optimization problem. We've got these resources, we've got these needs - how do we bring them together in the most optimum manner? On a micro level. In a way the solution to that could be a synthesis of free market capitalism and centralized communism. Better than either of them. Both of them bog down because it is just too centralized and there's no good capability to give most people what they need, or to use their resources well. So the solution has been to just make up some product that a lot of people ought to want, and make a one-size-fits-all version of it, and market the hell out of it, and not worry about if it isn't perfect for everybody. Or, in the communist version, to decide what kind of appartment everybody ought to have, or what kind of bicycle, and then just manufacture the same for everybody, in a big dull centralized factory. If we had the infrastructure for it, it could both open up a great many more business opportunities, and at the same time it can cheaply solve many social issues. As to the book I no longer need, I might either be happy giving it away for free, or I might participate in a business that gives me an appropriate amount of money for it. If I have a restaurant that throws away all the left-over food every day, it would be no extra trouble for me to let somebody come and pick it up and feed it to poor people. All based on that the communication infrastructure is very cheap and easy to use. And that it optimizes both the use of resources and the result or profit gained from their use. I might either want to get the most possible dollars for my book, or I might want to give it to whoever will have the biggest benefit for it. It is merely an optimization problem either way. The way it will happen is probably that comparnies carve out a niche, delivering more personalized service more efficiently, and we start getting used to expecting custom solutions, rather than general solutions that don't quite fit. We'll start expecting that there will be public transportation in front of our door, rather than half a mile away. It would be better if it were a transparent open-source system that did it, but competing business solutions would be a good start. The more transparent, the better. It can potentially be a huge shift in how things are done. See, most traditional businesses are based on hiding from you what the optimum solution would be. In part because they might not know, but also because their economics are based on giving you roughly what they give everybody else, and making you believe that it is the best that is available. Imagine that instead you had your own optimizer, which was plugged into a vast and finegrained infrastructure, which pretty instantly could point out to you the shortest route between what you have and what you want.

7 comments