Because, "opt-in ethics" turns stronger ethical preference into a game theoretic disadvantage.
There are two populations: population A is those who prefer ethical behavior internally and would volunteer for it even if not compelled. Population B is those who don't prefer ethical behavior internally, wouldn't volunteer for it, and must be compelled to act ethically from without if they are to act ethically at all.
In a landscape that impartially disincentivizes unethical behavior, both A and B can coexist.
But in a landscape that DOES NOT impartially disincentivize unethical behavior, everyone acts the same as before—unless there's a benefit to acting unethically. In which case, A, those who prefer to be ethical for its own sake, will inevitably be outcompeted by B, those who engage in any behavior regardless of ethics, so long as that behavior confers advantage.
Enough of that on a long enough time scale and the voluntarily ethical population just disappears.
So, either we enforce ethical behavior (even on those who need no forcing), or we create an unethical free-for-all waiting to happen.
In some cases the "waiting to happen" stage can last a surprisingly long time. Centuries, maybe. But never, as long as B's exist and are free to act, without end.
One solution to this problem that was popular in certain ancient societies was, "Round up and execute all the B's." But A) that's not very ethical, so we can't do that without becoming B's ourselves; and B) you can't tell who is a B just by looking at them, or their skin color or race or eye color or religion or whatever; and C) B's have a maximal incentive to become those who decide who is to be executed for being a B, so the whole thing is prone to MASSIVE corruption.
So the thing Liberal societies have done is create systems where we punish BEHAVIOR rather than trying to classify people. And it works, as long as we do the "impartially disincentivizing unethical behavior" thing I mentioned before.
Typically the behavior is "Very slowly, then suddenly all at once". Which makes issues easy to ignore and by the time all at once happens there is no time to actually deal with it.
There are two populations: population A is those who prefer ethical behavior internally and would volunteer for it even if not compelled. Population B is those who don't prefer ethical behavior internally, wouldn't volunteer for it, and must be compelled to act ethically from without if they are to act ethically at all.
In a landscape that impartially disincentivizes unethical behavior, both A and B can coexist.
But in a landscape that DOES NOT impartially disincentivize unethical behavior, everyone acts the same as before—unless there's a benefit to acting unethically. In which case, A, those who prefer to be ethical for its own sake, will inevitably be outcompeted by B, those who engage in any behavior regardless of ethics, so long as that behavior confers advantage.
Enough of that on a long enough time scale and the voluntarily ethical population just disappears.
So, either we enforce ethical behavior (even on those who need no forcing), or we create an unethical free-for-all waiting to happen.
In some cases the "waiting to happen" stage can last a surprisingly long time. Centuries, maybe. But never, as long as B's exist and are free to act, without end.