Okay, working with some more reliable sources now: Biden had 49,000 votes posted on the City of Milwaukee website as of 4 AM Wednesday ( https://web.archive.org/web/20201104040300/https://county.milwaukee.gov/EN/County-Clerk/Off-Nav/Election-Results/Election-Results-Fall-2020 ) and 194,000 votes as of 3 PM Wednesday. 102 wards added more than 500 votes, 16 added more than 1000. State margin 20k

And here are the Benford's law numbers for the leading digits of the early counts and the late counts. (shrug?)

Lmao, is this how city elections normally look? This is the percent of each ward's votes Biden won in the final counts

Close relationship between the percentages by ward posted early and late; little relationship between the totals posted early and the percentages posted late; many wards had counted very few ballots as of 4 AM but had very high %s as of 3 PM, which I guess we know.

And here's the Benford's chi-squared tests from the firstdigit.ado stata package, for the early and late totals.

The Jorgenson totals don't violate Benford's law, fwiw

So, here are four wards where early counts (as of 4 AM Wednesday morning) had fewer than 350 Biden votes but gained over 1300 Biden votes by afternoon. Ward 138, 177, 183, and 185. Maybe there's a simple story about why, apart from in person being lower than mail throughout.

To explain the (not great) variable names above: in Ward 138, Biden had 311 votes at 4 AM, 1773 votes at 3 PM, for a gain of 1462. in the same ward, Trump had 63 votes at 4 AM, 129 votes at 3 PM, for an increase of 66. That is, initial votes were 80% Biden, later votes 96% Biden.

This is Biden's % in early counted ballots versus Biden's % in the counts that were added after 3 AM yesterday.