Flora is a testbed for simulating what might happen if a proposed sociotechnical ssytem, such as a reputation or currency system, is implemented in a community.
Example Usage | Run Request | Technical Details
Using Flora, we compared the relative effectiveness of different approaches to reputation systems. We generated data for the following types of screening scenarios, where screening means a transaction party will refuse to transact with another party that fails certain evaluation criteria:
The reputation system used accounting-based and complaints-based metrics to evaluate whether a buyer or seller should transact with the other party. Complaints are related to how a resource producer, consumer, or waste cleaner is affected by the waste generated by a 'brand' - a person or group that serves market needs.
We see in the chart below that the fitness loss for disreputable entities are clearly different between
buyer-screening and no-screening. On the other hand, there is not as much difference
between seller-screening and no-screening.
In this chart, we see the fitness loss go up as the number of rejected transactions
increases for buyer-screening. For seller screening, the fitness loss remained around
the same value for different numbers of rejected transactions
From these analyses, we conclude that buyer-screening is potentially more effective than seller- screening with regards to pressuring entities to become more accountable and responsible. More detailed and technical explanation is available as a preprint at webcitations.org.
Please visit the user interface for submitting run requests. (Demonstration video to be uploaded soon.)
The code is maintained in bitbucket. It is currently in a private repository, but the intent is to make it public once the code is inspected by more reviewers or collaborators. Please send me a message at siosonel -at- gmail.com if you want to access it sooner rather than later (would require bitbucket membership).
The codebase is pretty stable, but would need to be updated to incorporate smarter decision engines and more diverse types of information systems such as different proposals for decentralized currency systems. For the most part, the user interface needs to be improved a lot more as well as coding how to handle runs on a live web server. One idea is for the live web server to just store run requests while another, 'hidden' machine pulls and runs those saved requests.