However we have not paid sufficient attention to what a small number of philosophers have been saying over the last few years. We will not hesitate to repeat it here, for truths must be told not only until they are adopted by every enlightened person, but until all those who defend the abuses they proscribe are silenced. The prevention of crime is less the effect of the intensity of a sentence than of its certainty; and extreme severity almost always results in impunity. Indeed, a humane man will not denounce a servant who stole from him if the sentence awaiting him is death. The same quality almost always prevents one from denouncing small thefts which, although less severely punished, are still disproportionately so. If, on the contrary, minor crimes were punished only by corrective sentences informed by and essentially punished through public opinion; if for all ordinary offences, and the least wrongdoings, we did not break in an instant all the ties that attached the offender to society (by taking his life, or covering him with permanent disgrace), that is, the last safeguard between him and a life of crime, then all would make it a duty, for the sake of common interest, to denounce criminals. We would be less indulgent, even, were it not for the fact that need reduces people to a dulled state which excuses their crime. Criminal laws, through their severity, and civil laws, because they favor inequality are therefore the cause of impunity for lesser crimes. And they can also be considered the cause of greater crimes since it is the impunity of the former which inspires the confidence needed to commit the latter.
In order for the fear of a sentence to be effective and beneficial the sentence must not outrage. Its justice must be perceptible to average reason, and it must especially awaken the conscience, at the same time as it punishes its silence and slumber. But this will not be so if sentences are too strong and instead of inspiring horror against crime, appear barbarous and unjust themselves; if they do not punish the injustices committed by the rich against the poor; if, when these injustices are not subject to sentences, the laws do not prevent them in other ways; if a judge can arbitrarily harden or soften a sentence; if there are privileges, hereditary, personal or local which offer a legal loophole, direct or indirect. Then the people are tempted to see criminal laws as made against them and in favor of the rich, as the result of an association designed to oppress them. Then they hate more they fear these laws that no longer inform their conscience because they outrage their reason, and this hatred is enough to overcome fear in strong souls and in all those made bitter by the joint feeling of injustice and need. Sophie de Grouchy (1798) Letters on Sympathy, Letter 8, translated by Sandrine Berges.
The "small number of philosophers" De Grouchy mentions in the first sentence above is almost certainly an allusion to Cesare Beccaria (1764) An Essay on Crimes and Punishments and Voltaire's championing it.* Some other time I return to Beccaria and the impact of his thought. But here I call attention to four features of De Grouchy's argument. For she is treating institutional design as a system of incentives on agents that can, if properly aligned, can cultivate virtue and minimize badness.
First, in particular, in thinking about the preventive role of criminal law, she treats would be criminal agents as evaluating (i) the risk of punishment and (ii) length/severity of punishment as distinct factors/variables (in some decision calculus).**
Second, the (i) risk of punishment is not just the chances of being caught in light (iii) of people's willingness to snitch on the would be criminal, but is also (iv) a judge's/jury's willingness to convict. Both (iii-iv) are informed by (v) a sense of proportionality between (ii) the length/severity of punishment and (iv) the severity of the crime. If (ii) and (iv) are felt to be proportionate then the risk of punishment is, thus, taken to be high ceteribus paribus. Because nullification is more likely when (ii) and (iv) are out of proportion, she advocates for less severe punishment for lots of crimes. (At the time one could be hanged or lose whole body parts for relatively minor offenses.) [UPDATED: as Liam Kofi Bright pointed out to me on facebook, this is a nice change from More's argument [recall here] that harsh laws for minor crimes ought be reduced because it incentivises committing worse crimes.]
Third, felt proportionality is a consequence of (vi) a kind of fairness constraint in light of perceptions of the larger institutional status quo ("if there are privileges, hereditary, personal or local which offer a legal loophole, direct or indirect. Then the people are tempted to see criminal laws as made against them and in favor of the rich, as the result of an association designed to oppress them.") So, she thinks that crimes are, in part, understood as consequences of larger social situation (for example, the needy are needy in virtue of the fact that society is badly structured; the feudal rich are protected by "civil law" such as tax and property laws that favor them, etc.) and also interpreted by ordinary people ("average reason") in light of these larger social situation by citizens.
As an aside, elsewhere in the book, De Grouchy also has a rich analysis of how our sense of fairness gets corrupted in unjust societies (about which another time more). But this helps explain how in some political circumstances people are willing to punish petty crimes rather severely.
Finally, criminal reform should, thus, not be decoupled from a larger institutional reform (which make both society and the laws in it fairer). Rather, because people's perceptions about society matter (vi) within the adjudication of criminal cases, would-be-social-reformers can influence criminal outcomes both via social/political reform and criminal reform.
*Perhaps, she is also thinking of Condorcet and early Bentham.
**As it happens, De Grouchy is a utilitarian about this. (Although her utilitarianism is a complicated one, but that's for another time.)