began having odd totals between poll 10-ish minute refreshes. Some could have been legitimate. Some were harder to explain, even for the poll itself. Several times negative counts were recorded - maybe Twitter caught some bots?

EST.....2:43 AM..2:53...3:02...3:11...3:22...3:33 AM...3:39...3:52...4:00...4:12...4:21...4:32 AM...4:53...5:01
BYU.......36........111....658....646......93.......30..........27.....26.....15.......26......51.......69...........32......0
WKU.....36.........-13.......31..1040....-37........30.........27......27.....15......26.......51.......71...........31......43

Having the same number of votes across across multiple refreshes should be extremely unlikely. That's on Twitter. A bot by either team probably wouldn't be able to do that.
WKU's large vote at 3:11 EST had no ramp up or ramp down so is unlikely by real people.
BYU's vote increase from 12:43 MST is probably legit followers of a popular Twitter account because they naturally ramp up and down.
What caught my attention was increments totals in our early yesterday evening of consecutive BYU-WKU 363:365, 529:532, 214:216, 437:442. This seems statistically unlikely to be natural. Either a bot or the twitter poll.
BYU dislikers and sore losers could account for the tripling of WKU's highest previous vote total. Almost but still very suspicious.