Wikipedia Study: Big Online Vote Landslides Aren't Real Consensus
That '99% approved' number may just mean everyone copied the first voters.
When an online community votes on something important, we tend to read the final percentage as a measure of how much real thinking went into it. A 95% win sounds like overwhelming, carefully considered agreement. A new study of Wikipedia's administrator elections — real votes that decide who gets power over the site — suggests that assumption is shaky. The researchers pulled 198,275 written explanations voters left alongside their votes, then compared how strongly someone voted against how much reasoning they actually gave.
What they found was a pattern they call 'vote-text divergence': a firm yes or no attached to a short, deferential explanation like 'I trust the people above me.' This got more common the later someone voted. At first that looks like simple crowd-following, but the researchers dug deeper. Once they accounted for how much evidence had already piled up in the discussion thread, the timing effect disappeared — and the amount of prior evidence took over. In other words, it isn't just that late voters are lazy. It's that the discussion gets saturated, and there's simply nothing new left to weigh.
The effect was strongest among casual voters — people loosely connected to the rest of the voting crowd, rather than the tight-knit regulars. Here's the twist, though: this shortcut-taking did not predict worse outcomes. The administrators who won were, by later measures like activity and staying in the role, just as capable. So the system picked good people — it just dressed the decision up in a bigger mandate than it earned.
Why should you care? Because the same dynamics show up anywhere a group decides something in public: Reddit and Discord polls, workplace votes, shareholder ballots, even online reviews. When you see a huge margin, you're probably not seeing 10,000 independent judgments. You're seeing a few early judgments and a lot of agreement with them. Practical takeaway: if you're running a vote, ask people to write down a reason — it changes what you learn. And if you're reading a poll, read the comments, not the percentage.
- Researchers analyzed 198,275 written reasons from Wikipedia admin elections and found late voters often voted firmly with almost no original reasoning.
- The cause wasn't laziness alone — once a discussion fills up with evidence, there's nothing new left to weigh, so people just echo the crowd.
- The effect was strongest among casual voters, yet the winners still performed fine, meaning big margins exaggerate scrutiny without necessarily producing bad choices.
Why It Matters
Polls, workplace votes, and online decisions with huge margins may signal bandwagoning, not real agreement.