Run the tests and see what our best options are. Decisions on what tests to run are in T107163
Description
Description
Event Timeline
Comment Actions
David, Trey and I talked about this over an email thread for the last few days. We then had a meeting this morning to discuss where we are going with this test and future tests.
The summary is: suggester is already setup quite loosely with respect to how it generates suggestions. The remaining best option to tune is to reduce the confidence of the suggestions which changes the threshold at which a generated suggestion is discarded. David has instrumented the code base to make many suggestion variables, including these two, adjustable on demand for our tests.