So at this point, your Chesterton fence alarm should be blazing red, because it is unlikely that you, sitting in pure contemplation, can correctly decide that something so widespread and stable across centuries is "preposterously stupid".
So let's try to understand why this is necessary.
To start, we do not live in a democracy, we live in a system where large majorities are needed for something to pass. Democracy is three wolves and two sheep voting on who to have for dinner. It scares the hell out of people, and rightly so.
To defend against this, we institute a number of undemocratic consensus requirements that effectively boil down to large supermajorities being needed in order to get a law passed. This, by the way, is why people are so confused that Biden can't get his infrastructure bill passed even though he has that oh-so-sexy 2 seat margin in the house and a 50-50 senate.
Now imagine you are a team leader with 10 reports, and you are evaluating some course of action, and 6 people feel strongly that option A is the way to go, but 4 people feel strongly that A is a menace and the opposite of A is the correct approach.
Do you go ahead and do A, because 6 is a bigger number than 4? If you do, you are a fool, because you tore your team apart, four people may leave, and you are screwed. You need to govern by consensus, and that means:
1. Abandon option A and find another option, A' that a large majority (8 out of 10) can support. This, of course is the ideal solution. A majority so large that the remaining holdouts feel like assholes for pressing the issue.
2. Come up with a mix, where you do A but also B, where the group that hates A loves B, and is willing to trade doing A for getting B. And the group that love A doesn't like B so much, but is willing to accept B in order to get A. So everyone is a little happy.
#1 we can call "find the one pure course of action we all agree on" approach
#2 we can call "generate a collective basket of actions that, together, we can all agree on".
And very little introspection should tell you that option 1 is almost never available, but option 2 is virtually always available.
So one senator gets more grain subsidies, and in exchange he's willing to tolerate removal of a tax deduction and the subsidy of some solar panels, and another senator hates the solar panel subsidy, but loves getting rid of the tax deduction more, and doesn't much care about the grain subsidy, and the third senator hates getting rid of the tax deduction but loves the grain and solar subsidies more.
Thus there is a policy mix where all of them can agree, even though on each specific policy there is no consensus. This magic by which the sum of policies is able to achieve consensus while each individual policy is not really confuses people who don't understand convex preferences. They think preferences should be linear, not convex. Some even have bizarre moral beliefs about this, which is truly a WTF situation, but that's where we are, people get outraged about literally everything, including convex preferences.
You may have noticed that this basket approach is very reminiscent of the gains from trade you learned in your econ 101 class, and you'd be correct -- this consensus formation is even called "horse trading".
Thus welfare maximization via convex preferences is the process by which a super-majority agreement is reached when there is no one pure action that we all agree on, allowing government to continue while protecting us from the tyranny of majority rules.
The only thing "preposterously stupid" is the shallow dismissal of this consensus formation, seeing as how it is the bedrock that allows consensus to be reached in any group situation, from being an engineering team lead to running a local Kiwanas club to running congress.
Next time you face a divided membership in whatever team you are leading, start some creative horse trading -- there is a good chance you will be able to reach consensus.