PROJECT BRIGHTLINE

Basic User Instructions

Brightline was written in Java and has been tested on various Wintel machines and Macs, as well as several Unix workstations. Though optimized for Netscape, the Brightline applets will also run just fine under Internet Explorer. The interface is not as clean, but the workings of the program are identical.

Click on the "Load Brightline Simulation" button to start the program. It may take several seconds to load. Once the main Brightline window comes up you can then Run, Pause, Step-cycle or Reset the simulation using the buttons at the bottom of the screen. Before you begin to experiment with the various Custom settings available under the Parameters menu you might want to Run the program several times in its default state just to become familiar with how it works.

The five large colored ovals are the Corporation agents. The five boxes above each oval represent the Shareholders of that Corporation, and the small black boxes represent Customers, with the Government agent represented by the large horizontal oval at the top. As the run progresses the five Corporate agents compete for Customers by lowering prices - which they can only do by increasingly externalizing costs. This concept of externalized costs is the key behavioral element that Brightline is intended to demonstrate. In future versions of Brightline we hope to further discriminate between varying types and degrees of externalization, i. e. accounting fraud, environmental crime, negative customer health consequences, excessive impact on government and "capture" of public goods, insider trading profits, excessive contingent lawyers fees and unreasonable levels of executive compensation, etc.

Both the Shareholder agents and the Government agent monitor this process and react accordingly. Shareholders are looking to maximize their investment return, and therefore will be happiest when their Corporations performance is high. When the Shareholder agents are green this indicates contentment with their Corporation's performance; they will turn yellow if market share goes down. The Government watches both each Corporation's individual externalization level AND the cumulative total of externalization by all Corporations agents. It will become potentially active - and turn red - when the cumulative total reaches a certain level, reflecting an overall negative public response to industry practices as a whole. Once active the Government agent will then begin to penalize the most highly externalizing Corporation in a given cycle, which will momentarily turn red in indication.

A normal run of the program is 300 cycles. Corporations who penalized for over externalizing become less competitive as a result, accounting for the shifting cycles of Customer migration. During the run you can monitor the main performance indicators in a number of separate, real-time graphic display windows by using the DataViews menu - if you have a large enough screen you can arrange nearly all of them to be visible at the same time, and pause them at any time to more closely examine the dynamics of a given cycle. You can also get more information about the relative states of each of the agents by simply pointing the mouse at them while paused.

The key action of the run occurs when three or more of a given Corporate agent's Shareholders become "Active" - indicated by those Shareholders turning red. In each run one of the Corporations is seeded with this potential - to become in effect a LENS-style "focus company". These Active Shareholders then have the effect of preventing their Corporation from henceforth breaking the law - from exceeding the Government agent's legal limit for individual corporate externalization. In the default mode the action of the simulation is fairly predictable, although the seeding of the specific focus company and the actual timing of the various possible interactions among the Corporate agents is set at the beginning of the run on a random basis, so that each run will be a little bit different.

What makes the program more interesting - and truly "complex" - is that you can alter essentially ALL of the initial variables of a given run by using the Custom Parameter Settings screen (accessed under the Parameters menu). Here you can change the overall length of the run, lock in the random number generator to a specific seed, and set the behavioral ranges and limits of each of the agents. While the language here is pretty straightforward and explicative, the variables used in this early iteration of the program are NOT particularly intuitive, and you will need to play around with them quite a bit in order to see just how varied the results can be.

There is at preset one known "bug" in the program - when you "Apply" new Custom Parameter Settings you should then Reset the main simulation window before beginning the new Run - and when you do the "Earnings" window, if you have it displayed, will close. You can then reopen the Earnings window, but at present (January 1998) this remains an irritating little quirk we have yet to repair.

In time we plan to add a database/spreadsheet backend to the program so that the output of a given run can be captured and stored, along with the initial settings. Then we will be able to do hundreds of subtly different runs and compare the results obtained analytically in order to study the actual cycles and trends of the model in a much more sophisticated way. We will eventually also add in additional interactive variables, and thus steadily increase the overall complexity of the simulation.

RETURN TO MAIN PAGE