Author Archive


In our game, we observed the exchanges that occur in a barter economy of two groups of three players that are trying to fill their ITP sticker albums. The winner of the game would be the first player to fill the album. Both groups started the game apart, being able to trade only in their local markets. Later in the game, they were able to carry and exchange stickers with players from the other group in the global market. Each player acts at different stages of the game as a supplier, transporter, dealer and collector. The three boards (markets) not only serve as places to exchange stickers; but as places for the players to gather together, promoting social intelligence.

Read More

Alejandro Puentes, Rodrigo Derteano, Sabrina Osmany

We have run a couple of tests so far. We are reviewing the rules and solving some issues with the website.

Overall Objective

The objective of the experiment is to observe a barter market self-organize, keeping track of the flow of stock (stickers) during the process.

Mechanism – The Sticker Exchange

The Sticker Exchange: 1) keeps track of all the transactions in the market, 2) collects data from the experiment, and 3) receives and disburses stickers from and to the players.


  • 1 Cashier
  • 1 Computer
  • 1 Script
  • 1 Database
  • Internet

Link to the Sticker Exchange

The Game:

The Sticker Exchange Market


  • The Sticker Exchange Market
  • 6 Players
  • 6 ITP sticker albums
  • 108 Stickers – There are 18 types of stickers in the game, and 6 stickers out of each type.


Each player will receive 18 stickers. Grouped into 3 groups of 6 stickers of the same type. In other words, at the beginning of the game, each player must hold all the available stock of three unique types of stickers in the market.

Player’s objective

Each player has to collect all the stickers needed to fill the ITP Sticker Album, and finish such a task before other players.


The Sticker Exchange chooses player number one. Turn rotation is clockwise.


  1. When it is your turn, raise your hand showing the stickers that you are willing to trade; you can either pass or trade as much as three stickers per turn.
  2. The other players can bet on your offer holding up the stickers that they are willing to trade—all bets must contain the exact number of stickers than your original offer.
  3. Choose one from all the bets.
  4. Both you and your fellow trader must hand in your traded stickers to the Sticker Exchange.
  5. After the Sticker Exchange processes the transaction, both traders will receive the traded stickers.

End of the game

The game ends when one players collects all the stickers needed to fill an album.

The stickers



Read More


Read More

Alejandro, Rodrigo, Sabrina



1. Objects:

12 bicycles (cards)

4 stations (players)

2. Objective:

To eliminate your opponent—the person in front of you, before anyone else does.

There are two ways to accomplish this:

  1. By getting their station full – 6 bikes (5 bikes)
  2. By getting their station depleted

3. Moves:

In each turn, you will have five seconds to prepare your move, at the end of which you must send one of your bikes to any other station.


Read More

This visualization shows the activity of the start (yellow) and end (blue) stations during the Feb 03 – 09 2014 week.

Read More

Alejandro Puentes – Rodrigo Derteano

We did a prototype of a system with five vehicles and two stations–each one with five docks. A vehicle has to check in and out of the stations, and the LEDs in each stations let us know which vehicles are in or out. It is a simple way to track specific vehicles and to know where they are in the system.


Read More

Screen Shot 2014-10-06 at 3.36.13 AMScreen Shot 2014-10-06 at 3.25.32 AM Screen Shot 2014-10-06 at 3.26.40 AM Screen Shot 2014-10-06 at 3.26.57 AM Screen Shot 2014-10-06 at 3.27.12 AM Screen Shot 2014-10-06 at 3.27.25 AM Screen Shot 2014-10-06 at 3.27.37 AMScreen Shot 2014-10-06 at 3.27.53 AMScreen Shot 2014-10-06 at 3.28.58 AMScreen Shot 2014-10-06 at 3.29.07 AM Screen Shot 2014-10-06 at 3.29.19 AM Screen Shot 2014-10-06 at 3.29.33 AM


The tests we did, varying the proportions of short, medium, medium-long and long trips did not produce meaningful results, except in a tiny window between values, that was just enough to be statistically meaningful. In the class feedback and conversation after our presentation, new ideas flourished on how to proceed with a more interesting approach using our model as a starting point. Thus, our model could start to throw interesting results, if we vary the trip distance proportions between the stations, which could give us more clues about which proportions tend to stabilize the system, and which don’t.

If the results of such a test produces meaningful results, a next step could be to classify the stations according to the trip distance patterns found. A more in depth analysis of individual cases in the city, could be useful to determine if there are specific situations that allow for certain trip length patterns, and shed light on possible strategies to modify them.   \

Read More

Most of CitiBike users are subscribers who have up to 45min of included ride time per trip. What would happen if we reduce it?

Watch the video


Original system (maximum distance trip = 32)

  • Unorganized trip patterns
  • Imbalance between busy and quiet stations
  • Few trips
  • Variable distance trips – unpredictable destination

Screen Shot 2014-09-24 at 10.03.29 AM

Maximum distance between trip = 25

  • Organized trip patterns
  • Balance between busy and quiet stations
  • More trips
  • Three possible distances – 6 out of 10 possible destinations

Screen Shot 2014-09-24 at 10.02.54 AM

Maximum distance trip = 17

  • Very organized trip patterns
  • Imbalance between busy and quiet stations
  • A lot of trips
  • One possible distance – 2 out of 10 possible destinations

Screen Shot 2014-09-24 at 10.07.12 AM

I expanded the Move Towards Target example, to try to visualize how reducing the maximum trip time would affect the model. In the original example (max distance = 32), where trips can go to any destination, there are times when stations stay quiet (without traffic) for a long period, while others are very busy. Also, the bike trips are unpredictable.  In the second example (max distance = 25), trips are more restricted and can go to 6 out of 10 stations.   Unlike the first example, there seems to be balance between busy and quiet stations, and the model is in general more predictable. Finally, in the third example (max distance = 17), there seems to be imbalance between busy and quiet stations with very predictable trips (only to two possible stations). I would say that there are good reasons to believe that changing the maximum trip time can be a powerful tool to improve the balance of the system. 


still working…

Read More

Monday, Feb 03


Tuesday, Feb 04


Wednesday, Feb 05


Thursday, Feb 06


Friday, Feb 07


Saturday, Feb 08 – People from New Jersey and East New York visiting the city? Start clusters (blue) are near the Path and NEC trains.


Sunday, Feb 09 – It seems like the system is “artificially balanced on Sundays”


These visualizations attempt to show a macro view of Citi Bike imbalances in the Feb 03-09 2014 week. I am overlapping the data of start and end stations in two layers to have a macro view of the system behavior: blue showing the clusters of stations where trips start, red showing the clusters of stations where trips end, and purple showing where there are intersections of red and blue (balanced clusters?). It is interesting to realize that the center of the system seems balanced, while areas in the periphery, such as Brooklyn, show more variation throughout the week. Also, Saturday and Sunday show patterns completely unrelated to those of weekdays. Both in Saturday and Sunday, the largest blue clusters (start stations) are in the areas of Manhattan where people from New Jersey (Path and NEC train) and Brooklyn would start their trips. On Saturday, the red clusters (end stations) are distributed throughout the city with two large clusters near Path stations, and other two large clusters near the East River. For me, the most interesting pattern from these visualizations, is the red layer from Sunday end stations, which shows a balanced system, ready to start a new week. I believe that this suggests that Sundays are tough days for Citi Bike workers, because they have to set the system ready for the next week. Zooming in and looking at the system from a micro viewpoint is also interesting, however, I believe that the most interested hypothesis from this visualization can be extracted from a macro viewpoint of the system. These visualizations were made using CartoDB.

Read More

Tappsi – Intelligent urban system example

Tappsi was the first mobile app for ordering taxi services in Colombia. The app pairs up nearby passengers and taxi drivers using the GPS of mobile phones and the mobile network. What started as a free app for a double-sided market in Bogota, soon became very popular and expanded to every major city in the country. One year after its debut, it started charging monthly fees to drivers, charging passengers for priority services, and offering the possibility to pay the fare using credit cards. In a city such as Bogota, where finding a cab can be a nightmare, Tappsi is a great example of an app that found leverage points in the system, improving the experience for both drivers and customers. The system is controlled by Tappsi, a private company that—according to them—acts as a filter, accepting or rejecting drivers in their system after carefully reviewing their profiles, cars, and documentation. The app is very active in social networks, always looking for ways to improve its services, occasionally sending surveys to its subscribers.

Do you think in the future we will need better centralized control, more self-governance, or a combination of the two?

I believe that if we are to improve the next generations of bike sharing, there has to be a strategy that harmonically combines the two of them. For me, the real challenge is to find leverage points in the system, where little changes can trigger profound transformations. Both centralized control and self-governance should focus primordialy on finding such leverage points. Developing adequate incentives for users to generate a self-organizing systems–such as the extra 15 minutes offered by Vélib’ to access uphill stations, can generate generate profound changes in the system with little effort. In our mid-term proposal, our goal is to generate a methodology to classify stations, depending on their typical inventory stocks, hoping to find some patterns that will allow us to identify some leverage points.

Read More