Deaths sometimes go backwards?

[Torrenal]

Either I misunderstand how to read the data (I don't think thats the case) or counts of people dead are moving in both up and down over as time passes.

I'm trying to do my own maps of the data, but doing this is highlighting where counts go up and then down.

Listing a few from 77 inconsistences in the us_deaths.csv I pulled early this morning...
Oconee, Georgia, US: 2020-04-14 Count = 1, 2020-04-15 Count = 0
Oconee, Georgia, US: 2020-04-16 Count = 6, 2020-04-17 Count = 0
Caddo, Louisiana, US: 2020-04-15 Count = 57, 2020-04-16 Count = 55
Lancaster, Pennsylvania, US: 2020-04-15 Count = 69, 2020-04-16 Count = 33
Bergen, New Jersey, US: 2020-03-30 Count = 23, 2020-03-31 Count = 4
Essex, New Jersey, US: 2020-03-30 Count = 13, 2020-03-31 Count = 4.0
Union, New Jersey, US: 2020-03-30 Count = 7, 2020-03-31 Count = 0.0
Lancaster, Pennsylvania, US: 2020-04-15 Count = 69, 2020-04-16 Count = 33

I suppose it's possible that the dead are being moved to different buckets on different dates, but you know what moving dead implies....

Zombies.

If I'm not reading the data right, let me know and I'll figure out what I've done wrong.
If the dead are ambulatory, I should be able to cope with it without too much trouble.
If the dead are rising (vampires?), that's a different problem I can't cope with.

-- Side note, good work compiling the data. I'm doing what I can do best, which is plotting data, I'll be hoping there's someone down-stream from me that can turn that into meaningful things for people, or otherwise just help me add final polish onto my work.