You didn't get agreement from over 90% of people. That's not at all what it says. You agreed with the majority 90% of the time. If you don't understand the huge difference, let me illustrate an easier to understand example. Suppose 1000 people answered 1000 questions, and all answers individually are in a 50.1% vs 49.9% split. There's essentially no agreement at all in the group, but you got a 90% score yourself, simply by siding with the 50.1% majority on 90% of the questions. Suppose also all 1000 people each choose a unique set of the 100 to dissent about. So, you're in agreement with exactly 0%, the group itself essentially agreed about nothing, yet you still scored a 90%.

Thank you, I understand now!

The problem is the argument over those 10%.

That's not the author's point. The number of people that agrees with all of your answers was likely very small. This is relevant to rule making systems because you have to make a decision on every situation yes/no, you can't leave undefined behavior, and the ruleset is adopted in totality.

If you were part of a huge zoning board for example and you proposed your rules the set of people that have no issue with it less than 1%.