Abstract
The software numerically tests the eFines mechanism, provided the players follow a strategy learned by machine learning. The testing is carried on model data. eFines mechanism:
We assume that a Sustainable Mobility Fund is in place; call it CH. The game consists of repeating the following steps, e.g. once a week: a. The traffic offenders waiting in the queue for the administrative procedure to be completed will be informed of their queue ranking. b. They can contribute to CH up to a hypothetical fine (F). c. Based on contributions, the queue changes as follows:
(i) If someone has collectively paid their entire hypothetical fine F, they are omitted from the queue and the Administrative Procedure is resolved with 'payment of the fine completed'; F is transferred from CH to the Municipality account as a fine paid.
(ii) If someone has remained in the queue for the maximum possible duration of the administrative procedure, they will leave the queue due to the expired deadline. Their aggregate contribution to CH remains to the Fund.
(iii) The order in the queue will change according to contributions to CH. Administrators are advised to make priority selections to process offences from an initial segment of the queue.
(iv) Contributions to the Sustainable Mobility Fund of offences selected by the officers for processing remain in the Fund and do not affect the administrative procedure in any way.
(v) New offenders arrive at the back of the queue.
The analysis of the eFines mechanism is non-trivial, using methods of algorithmic game theory, statistical modelling, machine learning. But the basic idea is simple, based on creation of an avalanche effect: the offenders in the initial segment of the queue pay something to avoid being processed, i.e., they pay in order to leave the initial segment of the queue. Those in the segment following the initial one realize that the rational for the initial segment is to pay something and so they have to pay something as well...and so on.