The stunning numbers behind the MVP's groupthink problem

How MVP voting has changed over time, and why Twitter is to blame

The 2017 NBA MVP race felt historic even while it was happening. In one corner, you had Russell Westbrook, the first player to average a triple-double since Oscar Robertson. His game was all snarls, raw kineticism, excess, and clutch shots. In the other corner stood James Harden, the ultimate min-maxer, constantly pushing the limits of math and rulebook in his craving for ever-greater efficiency.

Outside of a few partisans, onlookers seemed unable to decide who was a more deserving candidate as the season sprinted to its end. It was a “photo finish.” It was “too close to call.” It was a “brutal” decision.

It was also a blowout. Westbrook nabbed 69 first-place votes to 22 for Harden, smashing him in total points and vote share. The hare and the tortoise, if the hare doesn’t stop to take a nap.

And that’s one of the closest races we’ve seen in the last few decades.

The NBA and its media barnacles (*cough*) drum up MVP discussion almost from day one. We have league-sanctioned ladders starting in the season’s second week. Fans and pundits keep things moving, including semi-official straw polls to spur interest and take a pulse check. Bad-faith social media engagement farmers pit candidates against each other with increasingly rancorous rhetoric.

It’s a whole lot of pomp, a whole lot of pageantry. And yet, when the dust settles and the voting is revealed, it’s a one-sided massacre. Always.

Always! The last truly close MVP race was all the way back in 2005, when Steve Nash nabbed 65 first-place votes and an 84% vote share to Shaquille O’Neal’s 58 first-place nods and 81% vote share.¹ Since then, it’s been a bloodbath.

Some context: Since the 1980-81 season, roughly 100 media members have determined the Most Valuable Player (before that, players voted on it). Voters submit a ballot ranking their top five choices. First-place votes are worth 10 points, second-place votes are worth seven points, third-place votes are worth five points, etc.

Any semi-serious awards analysis requires looking at “vote share,” which is the points a candidate earned divided by the maximum number of first-place points available.² Slight changes in vote share actually represent monumental differences in voting totals — a 1% change in vote share roughly represents three to four first-place votes, depending on the number of voters.

Bear with me through this example (or skip ahead to avoid a little math, I don’t want to lose you over technicalities):

Imagine a two-horse race in the year 2050, with players Jikola Eokic and Noel Jembiid each receiving exactly 50 first-place votes and 50 second-place votes. They both would earn a vote share of 85% (the full math is in the second footnote, if you want to see it).

If, in 2051, Eokic earned 70 first-place votes and 30 second-place votes, and Jembiid earned 30 first-place votes and 70 second-place votes, that would seem to be a blowout, yeah? Eokic’s vote share would only be 91%, and Jembiid’s would fall to 79%:

To reiterate, as a very rough rule of thumb, remember that every 1% change in vote share is worth three to four first-place votes. (That’s fuzzy math, but go with it).

Relatedly, here’s a chart showing the MVP victor’s vote share each season (2025 left blank, for now) as well as a five-year rolling average:

Sure, it’s jumpy, but you can see that the rolling average is pretty consistently ~8% higher since the mid-2000s than it was in the decades before. Remember, that’s dozens of first-place votes, and there are only roughly 100 voters in a given season!

There are plenty of reasons for this sea change, but one in particular is striking.

Let’s set the stage. As recently as Michael Jordan’s first three-peat, your favorite team’s broadcaster was casting a secret ballot despite only having access to their team’s games and a scant few nationally televised matches (which were dominated then by flagship franchises like the Bulls and Lakers). Basketball talk was limited to legacy media: Magazines, newspapers, talk radio, and TV. High-level discourse was hard to find and even harder to contribute to.

No part of that paragraph is true today, and several factors have altered voting dynamics internally and externally. League Pass launched before the 1994-95 season, giving analysts, fans, and media members greater access to small-market teams than ever before. The NBA mandated open ballots for the 2013 awards and removed local broadcasters from the voting pool before the 2016-17 season.

But those all pale in comparison to one seismic impact: the opening of Twitter to the public in the summer of 2006.

Let’s look at that chart again:

Look at the difference post-Twitter! It’s almost instantaneous. The lowest winning vote share in the 18 years after the social media app’s launch was 2008’s 87%. 14 of the 26 years before that had a lower winning vote share!

Another way to frame the lack of photo finishes involves comparing the difference in first-place votes received by winners and runner-ups:

As you can see, the MVP victor has had a sizable lead in first-place votes over his closest competitor for the last two decades. The eleven closest margins all occurred in the 26 years before Twitter. In the 18 years since, the victor has received at least 39 more first-place votes than the runner-up.

Perhaps the 2025 MVP race between Shai Gilgeous-Alexander and Nikola Jokic will be different, but current data suggests we’ll be about in line with those numbers again. Max Croes’ award tracker, which looks at what voters have stated publicly in the lead-up to the big reveal, has Gilgeous-Alexander with 32 first-place votes compared to Jokic’s 15.

(In case you’re wondering, that hilarious dip in 1990 is when Magic Johnson won MVP despite receiving 11 fewer first-place votes than second-place Charles Barkley. It’s a wild ballot with pre-championship Michael Jordan in a very close third.)

I just threw a lot at you. Let’s simplify things. If we define “close” races as years where second place was within 20 or fewer first-place votes of the MVP winner (which is not particularly close), here’s what we’ve got:

Hopefully, you’re convinced of the change by now. So why was Twitter such a pivot point?

Twitter (now X) and its social-media ilk are a double-edged sword. On the one hand, there has never been easier access to quality basketball analysis. Social media has enabled incredible two-way knowledge sharing. TNT’s Jared Greenberg, a voter, told me for Fansided a few years ago that social media "gives you more to chew on... for me, it allows me to put more ideas on the plate and digest all this information. In every aspect of life, more information, more data is better."

The last few decades have been to sports analytics what the Renaissance was to art. Has it helped us zero in on the best candidates more accurately than before? Maybe advanced statistics and elevated discussion have simply made identifying the deserving winner easier. Wisdom of the educated masses, so to speak.

On the other hand, there is a real fear of being “wrong” and drowning under a deluge of criticism. As longtime voter Ben Golliver noted to HoopsHype several years ago, “It’s media-on-media backlash too, in some cases. The Twitter backlash and media backlash is worse than any backlash from players or agents who see your ballot.”

Basketball opinion-sharers learn to have a thick skin, but nobody enjoys being savaged by stans online. In 2023, a reporter named Andy Larsen, who covered the Utah Jazz, received thousands of vile messages for daring to pick Jazz center Walker Kessler for Rookie of the Year over consensus favorite Paolo Banchero. Larsen wasn’t even the only one who voted for Kessler! (I’ll save you from yet more graphs, but yes, we’re seeing similar changes in the other awards, too.)

Voter fatigue may be an issue, too, although we won’t crack open that particular can of worms since it’s already received thorough coverage elsewhere. The phenomenon’s very existence is contentious (“Voter fatigue is bullsh*t,” Greenberg told me), but a conscious or subconscious sense that players shouldn’t receive an award more than historical precedent allows can effectively remove a leading contender from the pool, concentrating votes in the remaining strongest candidate(s).

This vote consolidation may not be entirely subconscious, either.

Owen Phillips

at

The F5

ran an excellent piece a few years ago that speculated the NBA itself may be punishing highly divergent voters by removing their vote.

Groupthink has negative connotations for good reason, even if it doesn’t often create an obviously terrible result. Are we really so sure as a collective that any given award winner is that much better than his peers? Why are we certain year after year after year? Isn’t it funny that, over and over again, 100 voters will agree a race is really close, and then 80 will pick the same guy?

But there is some good news.

While outlier opinions and close races are on life support, we’ve staunched the bleeding. If we isolate the five-year rolling average for MVP to only the post-Twitter timeline, you can see that things have stabilized:

And while stabilizing at a vote share in the low 90s isn’t ideal, that’s still leaving 20-25 first-place votes up for grabs for everyone else. It could be worse.

In other words, groupthink is real, but the hivemind is not. There’s inherent value for media members (whose livelihoods depend upon audience interest) to have some diversity of thought; analysts may feel internal or external pressure to find a way to stand out from their peers to garner more attention. It’s not clickbait, exactly, but if a voter feels two candidates are equal, there may be more upside in vouching for the path less taken.

And some slight, defensible homerism will always be involved, even with the removal of team broadcasters. Many voters primarily cover one team, such as if they work for the city paper. They have so much more exposure to their local star’s greatness (and that’s leaving aside the fact that they have to deal with said star and his representation daily!). Given the incentives at play, it's easy for a media member to convince themselves that they have inside knowledge of how incredible their team’s guy is compared to everyone else.

In other words, imperfect but compelling systemic safeguards are in place to ensure we’ll never become one voting consciousness. We likely won’t see a race like 1990 again, and even 2005 feels like a long shot. We also can’t simply “force” close races; history has shown us that seasons with equally compelling candidates were uncommon even before social media. But I also don’t anticipate a string of unanimous MVPs. The media machine (*coughs again, reaches for a lozenge*) always needs more fuel for the fire.

Like the Red Hot Chili Peppers, the perpetual-motion awards discourse every season can’t stop. Whether there’s any drama involved is beside the point.

Thanks to Mark and Peter for their help on this project.

1. All numbers in this piece are from Basketball-Reference’s extensive databases.

2. An example: In 2024, there were 99 voters for MVP. Shai Gilgeous-Alexander received 15 first-place votes (15*10 = 150 points), 40 second-place votes (40*7 = 280 points), 40 third-place votes (40*5 = 200 points), three fourth-place votes (3*3 = 9 points), and one fifth-place vote (1*1 = 1 point). Add it up, and you get 640 points.

Divide that by the maximum possible points for a candidate, which would be 99 first-place votes (99*10 = 990 points), for the vote share of 640/990 = 64.6%.

Because you get points for second, third, fourth, and fifth places, the total vote share for all candidates adds up to way more than 100% (although no one can earn more than 100% individually). That same year, Nikola Jokic earned 926 points, so he won MVP with a 93.5% share. That can confuse people, so I just wanted to call it out here.