* The scheduler that ran the data collection step would get delayed and clogged sometimes in mid-2020, so data for 2+ days would be collected and allocated to one day. This is why some earlier dates have big spikes.

* Even in 2021, these spikes still sometimes appear (e.g. Vectis, MK1-Furis). I spot-checked a good amount of these and unless otherwise addressed, these actually appear to be legit.

* There were ~2 weeks during June 2021 where the data collector didn't run because I messed up my Java like a dumbass. This shouldn't have affected the total values too much though.

* Around July 15th - 16th 2021, I started using the server allocator tool which generated logs with slightly different names (e.g. DedicatedServer1.log, versus DedicatedServer01.log). This caused about one day of data to be repeatedly read in and accumulate to stupid amounts. I was able to reverse most if not all of that damage, so if some totals seem like they're less than the last time you looked, that's why.

* Data for July 29th - 30th 2021 is a bit spotty for daily data since I turned off my logging to address some issues. I gather the daily data from my own logs though, which means those two days will only have data if I added it (I only did for a handful that I could verify for sure).