# — ITPG-GT 2621 001 CLOUDCOMMUTING

Archive
Author Archive

This is a first attempt to model Lima’s transport system and how it regulates itself. The goal of the system, from the point of view of the bus drivers, is maximise the occupied bus capacity over time. This in turn drives them to regulate their speed according to different factors: number of passengers on the street, how full or empty the bus is, the proximity to the bus that already passed. The dateros, are like sensors that enable the dirvers to regulate their speed according to the other buses. This is a work in progress…

Read More

As I previously explained, the dateros are known for standing at the corners tracking the frequency of different bus lines. When the driver of a bus wants to know how much time passed since the last bus of the same line drove by, they ask a datero. Dateros sell their information, so they receive some dimes each time they provide it. It’s a self-organizing system, that regulates the velocity of the buses and their frequencies, while generating jobs and sustaining economic transactions between more or less independent agents.

Allright, so going a little further into system analysis, I’ll try to lay out the elements of the feedback loop in a basic way. The following components come to mind:

• buses (drivers)
• passengers
• dateros

The basic feedback loop we’re looking at is this: Drivers of the buses regulate (speed up or down) their velocity according to information given by dateros. But there’s more elements that influence the behaviour of the drivers and the whole system, and thus can throw the system off balance or make it’s behavior more complex:

• number of passengers on the bus (available seats or space): usually when the bus is full, drivers go faster, because they want to lay off passengers and gain time to start collecting new passengers. When the bus is empty, they start to drive slower, trying to get the passengers that the next bus would get (until the two buses meet and a crazy race starts!), poking around asking dateros for information in order to infer when the next bus should be coming.
• number of passengers on the street: if there’s a lot of passengers waiting to take the bus, buses don’t have to regulate their speed, they just drive. And when there’s no passengers, they start to drive very slow.
• number of dateros: if there are no dateros, buses are not able to regulate their speed, and the system looses efficiency (more dateros -> should lead to more efficiency if there’s competition). The number of dateros depends on how profitable it is to do this work. There’s a maximum amount of dateros per corner (usually one, max two), and if there’s not enough buses and competition between them, dateros start to disappear.
• bus frequency: if there’s little number of buses, then they don’t need to regulate their speed in relation to other buses, thus the number of dateros decreases, because they are not needed (and the other way around).

Thus, in an environment as Lima, where there’s plenty of passengers but also a lot of buses competing, the appearance of the datero was just a matter of time. Buses need to regulate their speed according to each other in order for the system to be more efficient. Since they don’t follow a schedule and their speed also abides by other factors, dateros are the ones that provide an essential feedback loop to balance the system.

Still, there are a lot of things that throw the system off balance. To name a few:

• spikes in passengers looking for buses
• Full buses going faster than they should
• Empty buses going too slow
• Bus frequency interruptions or mechanical problems

But at least in theory, this system is able to regulate itself more smoothly than one with fixed timetables. Still, when two buses or more buses met and start to compete for passengers, it can get quite dangerous very quickly.

Read More

http://code.rodrigoderteano.com/ccom-viz1/

So this is my first visualization of the citybike trips for a day in New York (Fri May 2nd 2014). The idea behind it was to see if trip patterns emerge between the stations that are more strongly connected towards each other, looking at stations as if there were nodes of a network. Trips as the links that connect them. The basic principle was to draw thicker lines for stations that are connected by more than one trip at a time and to do this in a time lapse. It was done in javascript using the leaflet library and mapbox, data is by city bike nyc.

I would’t go as far as to label this attempt a failure, but obviously the sheer number of stations make the result pretty difficult to read. There’s still some room for improvement though, because the trips could stay a little longer on screen to see the patterns and by using colors to mark, for example, longer trips, other patterns could start to emerge. But in any case, my inspiration to do this came more from wanting to see the visual result, than from analytical thinking, which is probably the root of the problem.

Still, looking at the result, we’re able to see:

• most station get used quite consistently during the heavy duty cicles
• Trip patterns suggest that, besides some specific cases, trips are quite distributed, meaning there’s not too much people doing the same trip at the same moment.

I a second attempt, what I did is to stop removing the trips from the screen, and only draw connections between stations with more than one trip at the time. In this case, it’s possible to see a bit more patterns, specially for connections to brooklyn.

http://code.rodrigoderteano.com/ccom-viz2/

Read More

1. I can think of an intelligent urban system related to transport, but in this case without the electronic components. Back in Perú in the early 90s, when the economy crashed completely and an economic shock treatment was put in place, there were massive layoffs, both in public in private payrolls. A lot of the jobless decided to invest in the newly deregulated market of public transport, buying cars, vans and small buses. This was the beginning of public transport nightmare of Lima, that is actually private.

A new character emerged inside this system, called “datero” (someone who provides useful data or tips). The dateros are known for standing at the corners tracking the frequency of different bus lines. When the driver team of a bus, normally consisting of a driver and a collector (who sells the tickets), wants to know how much time passed since the last bus of the same line drove by, they ask the datero. Dateros actually sell this information, so they receive some dimes each time they provide it.

If you look at this from a large scale point of view, you could see a self-organizing system, that not only regulates the velocity of the buses and their frequencies, while generating jobs and sustaining economic transactions between more or less independent agents. All of that with no central authority in the mix (the government is still quite absent).  That doesn’t make the experience of riding any better, though :)

2. Suppose that in few years we have driverless automobiles that can be used in shared schemes without the need for redistribution. Is this a plausible solution/future for you?

I can’t really imagine this as completely plausible, specially in emerging countries, because there will be a lot of issues involved in dealing with obsolete existing cars (even social unrest), and the system having to rely on so much information, connectivity and precision in order to be save. In that case, the cities would probably have to instate specific timeframes or zones where only self-driving cars can function, and they would have to have sophisticated safety-checks. How independent are these cars from each other? Do they rely on sending information to each other, or do they rely on a central information system? If this can be completely sorted out, for which 2016 seems a little early to me, then maybe I could see it happening in my lifetime. But if it ends being backed up on a central information system, it would be interesting to see how a system blackout would look like. In this case I agree with the author’s conclusions, that this is still in the early phases and that maybe our efforts could find better rewards in other ideas.

Read More