Guidelines/heuristics concerning "no action better than (possibly) wrong action"

During testing of a website, we found an unexpected behavior when the user misses a target and, due to a very slight mouse movement while clicking (i.e. accidentally dragging the mouse 1 pixel to the side), the opposite action is triggered, which is very disturbing.

I would like to reference some UX heuristics/guidelines to convince the developers this is a bad pattern, in the hopes that this will better explain them why such design is not good, so that future iterations of the website will converge faster.

However, the closest heuristic I found related to the subject would be this one:

Error prevention: Even better than good error messages is a careful design which prevents a problem from occurring in the first place. Either eliminate error-prone conditions or check for them and present users with a confirmation option before they commit to the action.

But I am not sure this communicates the situation clearly. In particular, the unwanted triggered action is not necessarily an error, and the "confirmation option" certainly does not apply.

The overall idea is that no action is better than possibly wrong action.

If the user really wants to trigger the behavior, they only need to repeat it with a larger intensity. As in smartphones, when a slight drag does nothing, and you have to drag it a decent amount before it actually does something.

Is there a more specific heuristic/guideline that would apply here?