In the Colombian referendum last year, people got the options of “yes” or “no” to a thick pile of paper outlining the exact conditions for peace with the Farc guerilla. The result was even, but the no side won.
Referendums in general must be the least democratic process for decision making that you can possibly participate in (while still actually participating in a decision).
The problem is that you are not involved in the processes that should come before a decision, like defining the problems to solve and coming up with constructive solutions.
That leaves you with a decision that is not about solving peoples actual problems, but some other thing that the politicians can’t make up their mind about.
When you ask a confusing question and it’s unclear what actual problem it pretends to solve, you will get a random result. Like in the referendums in the UK or Colombia last year.
If the suggested solution was unclear to Colombian voters, the issues addressed must have been extremely confusing to UK voters.
Which problem did the people who voted for Brexit try to solve? A lot of the debate was about immigration. But “leave the EU” and “less immigration” are both suggested solutions, not actual problems. So the issues must be something else, but what were they?
Looking at the statistics, a qualified guess is that two immediate concerns for many of the no voters are a secured income and affordable housing.
… most of those not working voted to Leave. More than half of those retired on a private pension voted to leave, as did two thirds of those retired on a state pension.
Around two thirds of council and housing association tenants voted to Leave.
When acute problems don’t get addressed in other ways, they spill over into referendums and general elections, where you at least get a chance to piss off the academic elite.
If our democratic processes were more about addressing peoples needs, and less about token participation, we would use methods for real participation and problem solving instead of confusing referendums.