What is this?
Everyone here gets one vote. We count those votes twice: once counting everybody, and once counting only the people sentimentmaxxing - the ones who put money behind their vote. The gap between the two numbers is the entire point.
Why not just run a normal poll?
A free poll costs nothing to answer, which is what makes it easy to answer carelessly. Economists call that cheap talk: a signal that costs the sender nothing tells you very little.
Attach a price and only people who care past that threshold bother. That is a costly signal, and the useful part isn't the money - it's the filtering.
Why are the questions so trivial?
The pettiness is the method. Ask about anything consequential and the paid answer stops meaning what it claims to: anyone with money riding on the outcome has reason to move the number, and a group willing to coordinate could buy a result and quote it as evidence.
Nobody has a strategic interest in the direction of a toilet paper roll - cats excepted, and they have so far declined to pay. So if somebody pays to weigh in, the only plausible explanation is that they genuinely care, which is the one thing we are measuring.
It is also why paying can't make anyone right. Nothing here resolves. In a prediction market money pushes toward accuracy because being wrong is expensive; here it can only reveal how much someone cares - which ordinary voting cannot see at all, since a shrug and a conviction both count as one.
How does the money part work?
Your one vote counts once in the first tally, however much you spend. In the second it carries weight, and each conviction vote costs double the one before: $1, $3, $7, $15. Four is the ceiling, so no amount of money buys a result outright.
A money tally stays hidden until at least 5 people hold conviction votes on that question, and whenever one of them holds too much of the weight. One person's opinion wearing a percentage sign is worse than no number at all.
Where does the money go?
We donate 25% of purchases to Bay Area nonprofits focused on education and economics.
How much should I trust these numbers?
Less than a representative survey. Nobody here is a random sample: the free tally is whoever found the site, and the paid tally is whoever found it, had a spare dollar, and felt strongly enough - a selection bias stacked on a selection bias. Paying also correlates with having money to spare.
None of which makes the comparison worthless. It just means the interesting question is the shape of the gap, not the precision of either number.
The terms cover how funds work. Got a question worth asking? Suggest one.