Any college basketball fan who watches the annual NCAA Men’s Tournament usually wants (and expects) at least one no-name, small school to climb the ranks and upset much bigger schools with blue chip recruits. If the team is lucky and good enough, they’ll get to at least the Sweet Sixteen (fourth round), or even the Final Four (semifinals). Such teams are not expected to win it all. They’re called Cinderellas for a reason. Over the last week we had our own Cinderella in baseball, and his name is Eric Sogard.

The bespectacled Sogard was the A’s entry into the Face of MLB contest, a series of Twitter popularity polls pitting a player from one team against another player on another team. He was also by far the least known quantity of any of the entrants, which included the likes of Derek Jeter, Felix Hernandez, and David Ortiz, who received a “bye” round. Somehow Sogard worked his way through the first two rounds, besting the likes of young Cubs star Anthony Rizzo and Rockies shortstop (and Fremont High of Sunnyvale product) Troy Tulowitzki.

Well, it wasn’t so much Sogard that did it. It was the ever resourceful and creative A’s fan base that did the bulk of the work. The polls worked by tallying up tweets labeled with the hashtags #FaceofMLB and the name of the player, in this case #EricSogard. MLB put some rules in place to govern the poll: a definite window to vote from 9 AM ET to 8 AM ET the following day and a limit of 25 tweets (or retweets) per Twitter handle. The rules were fair and provided advantages to both East Coast and West Coast voters, as I’ll discuss later.

After clearing the first two rounds, Sogard was matched up against Giants All Star catcher Buster Posey, an apparent mismatch of epic proportions. Yet those scrappy A’s fans came through again, lining up plenty of votes to beat Posey. Next up was Blue Jays right fielder Jose Bautista, yet another seeming mismatch. Oakland fans against all of Canada? Come on, now. Yet A’s fans understood the dynamics and kept plugging away at odd hours, steadily building a big lead as Eastern Canada slept and riding that into the finals.

That cleared the way for the final vote in which Sogard faced off against Mets third baseman David Wright. Wright, nicknamed “Captain America,” also had the look and general popularity to be the presumptive winner going away. The poll started out dead even for the entire morning, with Sogard garnering a 51-49 lead around noon. Wright caught up and again the two were deadlocked until 8 PM ET/5 PM PT, when A’s fans hit Twitter hard with #EricSogard tweets. By this round, the fans had made use of their Photoshop skills, creating some excellent meme-worthy material such as this tweet from @RallyPlantain:

What’s your name? #EricSogard What’s your quest? #FaceofMLB What’s your favorite color? Green & Gold #WayBackMachine pic.twitter.com/zNchaxzo0m

— Rally Plantain (@RallyPlantain) February 28, 2014

While other candidates enticed fans to participate by promising tickets (the Mets) or a follow back in Joey Bautista’s case, all of the momentum for Sogard was fan-generated. It was helped by the team, the radio station, and Sogard’s wife, Kaycee. Even local media jumped on the bandwagon to an extent. Around 9 PM, I saw links to two pages (h/t: @kenarneson) run by a third party company hired by MLB to run the polls, Mass Relevance. The pages are in JSON, a simple text format used to pass data from servers to web apps. They provided the raw vote data I needed to provide real time updates twice an hour throughout the evening. (Wright results/Sogard results)

At 1 AM, I posted my last update for the night, showing that Sogard had an impressive 55-45 lead with nearly 44,000 votes in hand. Based on previous voting days, such a lead seemed almost insurmountable and many fans went to sleep feeling pretty secure about the results. I, too, went to bed.

I woke at 4 with no alarm. I took a peak at the numbers again and was startled. The 44,000-vote gap had been cut by a whopping 40% in only an hour. This was the start of a tidal wave of voting for Wright. During the final 4 AM hour, #DavidWright tweets dominated Twitter. More than 110,000 #DavidWright tweets registered in the final hour. By 4:30 it became clear that Wright votes were going to catch up with Sogard votes. But with the West Coast still asleep, could the early risers there keep up enough of a pace to keep the surge from overtaking them?

At 4:45, what was in question became inevitable. Nearly 2,000 tweets per minute were coming through. Yet MLB’s rules about 25 tweets to a single user remained in place, which meant that people who kept tweeting and tweeting were getting rejected. During that last hour, Wright garnered 54,420 “approved” votes, but also had 50,918 rejected votes. Those 54,420 votes accounted for 20% of Wright’s total for the whole day. Even with the rejections, the sheer volume was enough to surpass Sogard and finish with an 11,000-vote lead. Final percentage posted by MLB and revealed during MLB Network’s Hot Stove morning show: 51% Wright, 49% Sogard.

Vote tallies taken throughout the evening of 2/27 and early morning of 2/28

MLB doesn’t certify results and post hard numbers like a real election board or registrar would do, so the numbers above are technically unofficial. Yet it’s clear how the trends worked out. In the aftermath, many A’s fans screamed conspiracy or that the contest was rigged. MLB can’t rig Twitter, so it’s not a Twitter problem. Everything else is strategic. Teams can entice fans to vote using a number of giveaways or contests, which the Mets did. Fans or teams could create additional accounts to eat up 25 votes. Bots can be set up to do the same. Bots out of South Korea tweeted for Wright, while the sports-unrelated account @LoveQuotes tweeted some Sogard love before deleting those same tweets. Many voters were unclear as to how and when the limit on 25 tweets per user was reset. As I understood it, the reset occurred at the start of voting each day. Others thought it was at midnight, an assertion which wasn’t backed by data. Consider that the final margin of ~11,000 votes represents roughly 440 individual voters or users. It’s such a tiny margin that it looks negligible.

As I looked at samples of tweets during the 4 AM hour, I saw what could be considered bots. However, the vast majority of Wright voters were living, breathing Mets fans. I can’t say how much they were helped by technology, but that pales in comparison to the network effects these types of polls can build. The Mets’ fan base is much larger than the A’s, and the NY Metro is much larger than the Bay Area. When it came time to show in numbers, they did so tremendously.

Hourly voting tallies, including accepted and rejected votes. Click to enlarge

If MLB decides to run the Face of MLB contest next year, they’ll need to make revisions to try to prevent users from gaming the system. There’s only so much they can do. They can’t really weed out the content in each tweet, nor is it easy to ban obvious bots. Besides, what’s the difference between a bot and a person who makes multiple Twitter accounts? I used my own two long-established accounts to vote for Sogard, so am I a cheater? In any case, expect fans from every team who are interested to have a better understanding of the dynamics of a poll like this, and respond accordingly. That means a Cinderella like Sogard is less likely to take hold next year. As we saw just a few hours ago, brute force can overcome any deficit. But don’t be discouraged, A’s fans. There are ways to strategize this. I hope that this effort spills over into future All Star voting efforts, where a small fan base team like the A’s rarely gets position players in. There is hope. And for what it’s worth, the last 48 hours have been one helluva ride.

—–

P.S. – The best tweet of the week came courtesy of former A’s reliever Pat Neshek, who is currently in Spring Training with the Cardinals.

Vote #EricSogard #FaceofMLB – Eric & our rescue dog Lulu, the only man she’s ever gone near. pic.twitter.com/RwIsTWDnSB

— Pat Neshek (@PatNeshek) February 27, 2014

Maybe Eric Sogard is magical, after all.