Not very. Our first option, then,
...distributes punishments proportionately, so that the worst crimes are matched with the worst penalties, and so on down the line. This method dictates only relative levels of desert, rather than requiring any particular objective measure of what criminal acts deserve what treatment. We might call this version of retributivism the "proportionate penalty" theory.Note that the ceiling comes first: choose the most serious offense, define the most serious possible punishment, and work downward. The problem, then, is what criterion we use to justify the most serious punishment. Is it death? Torture? Isolation? Attending a Hannah Montana concert?
There's another option.
The second, and more promising strategy is to attempt to establish a moral equivalence between crimes and permissible punishments. This strategy asserts that the perpetrator should suffer an amount equivalent to the harm or moral evil inflicted on the victim, but the kind of harm or moral evil involved need not match. That is, instead of either assigning the same harm or evil as punishment that the offender inflicted on his victim, or fixing penalties proportionately by making sure that the right intervals obtain between levels of punishments, we can match crimes with punishments on an absolute scale, but establish only a rough moral equivalence between the two. We would seek to inflict on the perpetrator by way of punishment the nearest morally permissible form of punishment to the act the perpetrator committed. Let us call this version of retributivism the "moral equivalence" theory of justified punishment.Finkelstein traces this idea back to Kant. It suffers from a similar problem; we need to determine exactly how to correlate, say, a fine with a felony. We have a reason for doing so, but not a method.
Any debater running up against a retributivist case should ask some tough questions in CX about justifying particular punishments. These options provide a way out, but not a destination. That's up to you.