A friend asked my me my opinion of referendums. I'm generally opposed to direct referendum voting.
Now, all of this is provisional. It matters the health of the legislative and executive systems we're comparing it to. It matters the threshold.
(I'm using the plural referendums because sometimes even I need to give the smug erudition a freaking rest.)
. . . . . . . . . . . . . .
Problem one: disconnect between the values/intentions of voters and the outcome of referendums
Sometimes the issue is clear and the politics make sense, eg Massachusetts recently voting to legalize pot (which I support), or various states voting to block gay marriage (the blocking of which I oppose). But much of the time the tradeoffs at play are obscure, the wording doesn't fairly reflect the core questions, and people don't really understand what they're voting on, or how it got here. if one group is doing an end run around a legislative compromise, for example, or trying to get their shoddy hospital funded before a more carefully planned one can be approved.
Part of this problem: not much incentive for deliberation. the typical voter doesn't have much of a reason to study up on the various referendums. At least a legislator is much more likely to cast a crucial vote, whether from the perspective of the odds of the bill passing, or the focus which their vote will receive.
The classic example of problem one is the observation that referendums that propose new government spending generally pass, but referendums that cut taxes also generally pass. (As far as I know, states must all balance their budgets each year, so these are even more contradictory then they are on the federal level.) The same referendums, worded to explicitly reflect the full impact of these proposals, would likely not pass.
Compare this to the federal budgeting system, where (IIRC!) even though Republicans have enjoyed a two House Majority and the presidency for two different periods in the last 20 years, they have passed relatively few unfunded tax cuts. It's just a bit less willy nilly with legislators.
Problem two: the values/intentions of voters being wrong
The first aspect of this is status quo bias. Our federal and state systems have a heavy status quo bias, in that there are lots of ways for legislation to fail and only one or two ways for it to pass and become law. Thus our legislative systems have the value built in that "I'm not sure" becomes "let's not change it". This bias is not without its problems, but I think it's generally a good thing, and the country's founders certainly agreed (though they were wrong on plenty of stuff).
The yes or no format of referendum questions demolishes this. The two options don't seem significantly different in prudence, and if anything, I think psychologically many people have a preference for saying yes if we can't think of a reason to say no. With the current system, why not just keep throwing a referendum like Brexit or against the wall until it sticks? It only needs to pass once!
Voters should be asking themselves, "Am I SURE that it is a good idea to..." before reading referendum questions, but they generally don't. They're wrong not to.
The counterargument is that often populations seem to be ready for a social change before their legislators are. The public turned against the Vietnam War long before Congress did, for example, and the Massachusetts legislature is only begrudgingly cooperating with the instructions of the marijuana legalization referendum. But I think most of the time, a legislature will slowly come around, and it's very much worthwhile having that slowness for other purposes.
Then there is the fact that most people just generally aren't that smart or knowledgeable about most things, and aren't good at even assessing or acknowledging their own ignorance. Of course, neither are legislators. But at least legislators have had to jump through some hoops of competence and organization to get where they are, and spend part of their days discussing the meaning and impact of legislation. And at least legislators must face budget and procedural constraints that require them to justify their proposals to some degree in a broader context, rather than skipping this context entirely, as referendums do.
. . . . . . . . . . . . . .
I think referendums would be workable with a higher threshold for passing, like 60% of eligible voters ... I'm sympathetic to the need for an end run around pervasively corrupt governments, I just think, for the reasons above, that 50% of those who happen to vote that day is a ridiculous threshold for a permanent change to the law.