I think IRV at least has fairly unpredictable failure cases, which means they can't be exploited.
I remember seeing a 2D chart representing voter preferences on the X and Y axis, coloured by who wins under those preferences (and some symmetry-breaking assumptions). Some systems showed clear biases towards certain outcomes. IRV had a pretty symmetric shape but with a splotch in the middle that was a random jumble of different colours - an unpredictable, chaotic outcome.
I remember seeing a 2D chart representing voter preferences on the X and Y axis, coloured by who wins under those preferences (and some symmetry-breaking assumptions). Some systems showed clear biases towards certain outcomes. IRV had a pretty symmetric shape but with a splotch in the middle that was a random jumble of different colours - an unpredictable, chaotic outcome.