For example: I don’t believe in the axiom of choice nor in the continuum hypothesis.
Not stuff like “math is useless” or “people hate math because it’s not well taught”, those are opinions about math.
I’ll start: exponentiation should be left-associative, which means a^b should mean b×b×…×b } a times.
I have this odd, perhaps part troll, feeling that there are two, and only two, roots of the Riemann Zeta function that aren’t on the critical line, and are instead mirrors of each other at either side of it, like some weird pair of complex conjugates. Further, while I really want their real parts to be 1/4 and 3/4, the actual variance from 1/2 will be some inexplicable irrational number.
Multiplication order in current mathematics standards should happen the other way around when it’s in a non-commutative algebra. I think this because transfinite multiplication apparently requires the transfinite part to go before any finite part to prevent collapse of meaning. For example, we can’t write 2ω for the next transfinite ordinal because 2ω is just ω again on account of transfinite and backwards multiplication weirdness, and we have to write ω·2 or ω×2 instead like we’re back at primary school.
The good thing about multiplication being commutative and associative is that you can think about it either way (e.g. 3x2 can be thought of as "add two three times). The “benefit” of carrying this idea to higher-order operations is that they become left-associative (meaning they can be evaluated from left to right), which is slightly more intuitive. For instance in lambda calculus, a sequence of church numerals n1 n2 … nK mean nK ^ nK-1 ^ … ^ n1 in traditional notation.
I’d say the deeper issue with ordinal arithmetic is that Knuth’s up-arrow notation with its recursive definition becomes useless to define ordinals bigger than ε0, because something like ω^(ω^^ω) = ωε0 = ε0. I don’t understand the exact notion deeply yet, but I suspect there’s some guilt in the fact that hyperoperations are fundamentally right-associative.