Board Thread:Game Discussion/@comment-28169398-20180524195008/@comment-28169398-20180525172358

Mugwumps wrote: Amrosa wrote: They have actually gone back to easier bots after trouncing auralpod... I can't figure out how they wrote this code. Partly there is a +/- factor associate with given tracks (e.g. you always get slower bots at melbourne), but the main driver of bot difficulty isn't your margin of victory, it is your total time and where it falls in the distribution of all recorded total times. EK 10-lappers are a good place to observe this. You can destroy the field (lapping most/all) without cutting, and you will never draw the top bots. You need to cut heavily to bring them out even though they are incrementally faster than the bots you're already lapping. However, there are so many fast times set with cutting for that race on FM's servers, it shifts the time distribution. This is the same reason you can bot slow while winning. All you do is block the bots and increase you total time to be greater than what the top bot would have run without interference, and you'll get slower bots. Beyond that, FM seems to employ a sort of moving average of a couple of events to adjust bot speed. Certainly nothing interesting like a Bayesian approach. I think that this is exactly the reason why there should be User-Selectable Difficulty Settings. FM still controls your upgrade tree with Cup requirements and Trophies to unlock, so difficulty selectablity in Career, Bonus, and Limited-Time Series would in no way effect their revenue potiential.

As for EK, there is some degree of randomness as well. I don't cut on the ten-lappers, or any other race for that matter... But I had 1 race, after I messed up the Wait at the Line bot management, where I drew Dr. Perriwinkle. She wasn't the fastest bot either... I had to help her get into the lead by taking out a TSM bot, then I proceeded to build a 5km lead on her... never saw her after.