[dragonwriter]

> A referendum is structurally incapable of enacting policies with minority support, for good or for bad.

With a single-subject rule, that’s often (but not always) true; without it, it is less true, because policies can be packaged to achieve a combined majority, so long as there isn’t a majority that thinks it is important enough to defeat any part to overcome any support within that majority for other parts. (Even with a single-subject rule, this can sometimes be done, so long as the policies packaged relate to sufficiently closely related subject matter as to fit within the way the rule is applied.)

This is common in legislative bodies, and it works with citizen-legislators, too.

[ThrustVectoring]

Even without a single-subject rule, there's zero enforcement mechanism to prevent other interest groups from reneging on the packaging agreement and defeating one or more parts in detail. There's no stable point you can get via referenda if (policy A + policy B), (policy not-A), and (policy not-B) all would garner majority votes.

[dragonwriter]

> Even without a single-subject rule, there’s zero enforcement mechanism to prevent other interest groups from reneging on the packaging agreement and defeating one or more parts in detail.

Sure, after the package passes, groups can try to form separate coalitions to pick things out of it; there are methods to protect them, generally (such as putting together a similar coalition to put up a trigger bill that deletes the parts the other groups want to protect conditioned on the other repeal passing, undermining its support.)

> There’s no stable point

Yeah, real world politics is generally not about finding stable equilibria, as much are the things most easily amenable to theoretical analysis.