https://github.com/cptanalatriste/techdebt-game-model

A game-theoretic model of technical debt.
https://github.com/cptanalatriste/techdebt-game-model

game-theory simulation technical-debt

Last synced: about 1 month ago
JSON representation

A game-theoretic model of technical debt.

• Host: GitHub
• URL: https://github.com/cptanalatriste/techdebt-game-model
• Owner: cptanalatriste
• License: mit
• Created: 2018-08-16T15:01:15.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2020-09-08T06:51:49.000Z (over 4 years ago)
• Last Synced: 2024-11-13T14:32:34.258Z (3 months ago)
• Topics: game-theory, simulation, technical-debt
• Language: Python
• Homepage: https://www.sciencedirect.com/science/article/pii/S0164121219301980
• Size: 377 KB
• Stars: 3
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Equilibrium Analysis for Technical Debt

*For an in-depht discussion of our empirical game-theoretic model, you can read our paper in the Journal of Systems and Software: https://www.sciencedirect.com/science/article/pii/S0164121219301980*

This model can generate the payoff values for a payoff matrix representing technical debt. After triggering the script `payoffbuilder.py`, the payoff information per strategy profile will be stored in the file `payoff_table_builder.log`.

This values can be later ported to a game solver like [Gambit](http://www.gambit-project.org/) to obtain its Nash Equilibrium.

Some relevant parameters of `payoffbuilder.py` are:

* `simulation_episodes` at line 44 controls the number of simulation iterations to execute.

The payoff values generated by the script is the average number of fixes per developer over these 

iterations.

* `sloppy_rework_factor` at line 48 controls the increased probability that a kludge commit will later generate rework, like a bug or a change request due to code review.

The payoff values are generated via simulation. Simulation-specific parameters are exposed via the script `trainingdriver.py`. 

Some of them are:

* `SCENARIO_TIME_UNITS` at line 13 is the time units simulated during an iteration.

* `SCENARIO_AVG_RESOLUTION_TIME` at line 14 is the average resolution time of a programming task.

* `SCENARIO_PROB_REWORK` at line 15 is the base probability of rework for the project.