Run-Time Monitoring

Finding errors is easier at run time, as we only have to worry about a single trace instead of all possible ones. However, as in gradual typing, additional run-time checks cause overhead.

The goal of this project is to find useful ways in which one can monitor black-box programs efficiently, typically with the compromise that the monitoring is only a best effort. This can still be a cost-effective way to gain additional confidence in software created by untrusted vendors.