Foucault: What are we to understand by “security”?

What are we to understand by “security”? I would like to devote today and maybe next week to this question, depending on how quickly or slowly I go.
I will take an example, or rather a series of examples, or rather one example modulated in three stages. It is a very simple, very childish example, but we will start from there and I think it will enable me to say certain things.
Take a completely simple penal law in the form of a prohibition like, say, “you must not kill, you must not steal,” along with its punishment, hanging, or banishment, or a fine.
In the second modulation it is still the same penal law, “you must not steal,” and it is still accompanied by certain punishments if one breaks this law, but now everything is framed by, on the one hand, a series of supervisions, checks, inspections, and varied controls that, even before the thief has stolen, make it possible to identify whether or not he is going to steal, and so on. And then, on the other hand, at the other end, punishment will not just be the spectacular, definitive moment of the hanging, fine, or banishment, but a practice like incarceration with a series of exercises and a work of transformation on the guilty person in the form of what we call penitentiary techniques: obligatory work, moralization, correction, and so forth.
The third modulation is based on the same matrix, with the same penal law, the same punishments, and the same type of framework of surveillance on one side and correction on the other, but now, the application of this penal law, the development of preventive measures, and the organization of corrective punishment will be governed by the following kind of questions.
For example: What is the average rate of criminality for this [type]? How can we can predict statistically the number of thefts at a given moment, in a given society, in a given town, in the town or in the country, in a given social stratum, and so on? Second, are there times, regions, and penal systems that will increase or reduce this average rate? Will crises, famines, or wars, severe or mild punishment, modify something in these proportions? There are other questions: Be it theft or a particular type of theft, how much does this criminality cost society, what damage does it cause, or loss of earnings, and so on?
Further questions: What is the cost of repressing these thefts? Does severe and strict repression cost more than one that is more permissive; does exemplary and discontinuous repression cost more than continuous repression? What, therefore, is the comparative cost of the theft and of its repression, and what is more worthwhile: to tolerate a bit more theft or to tolerate a bit more repression? There are further questions: When one has caught the culprit, is it worth punishing him? What will it cost to punish him? What should be done in order to punish him and, by punishing him, reeducate him? Can he really be reeducated? Independently of the act he has committed, is he a permanent danger such that he will do it again whether or not he has been reeducated?
The general question basically will be how to keep a type of criminality, theft for instance, within socially and economically acceptable limits and around an average that will be considered as optimal for a given social functioning.
These three modalities seem to me to be typical of different things that we have studied, [and of] those that I would now like to study. You are familiar with the first form, which consists in laying down a law and fixing a punishment for the person who breaks it, which is the system of the legal code with a binary division between the permitted and the prohibited, and a coupling, comprising the code, between a type of prohibited action and a type of punishment. This, then, is the legal or juridical mechanism.
I will not return to the second mechanism, the law framed by mechanisms of surveillance and correction, which is, of course, the disciplinary mechanism.
The disciplinary mechanism is characterized by the fact that a third personage, the culprit, appears within the binary system of the code, and at the same time, outside the code, and outside the legislative act that establishes the law and the judicial act that punishes the culprit, a series of adjacent, detective, medical, and psychological techniques appear which fall within the domain of surveillance, diagnosis, and the possible transformation of individuals. We have looked at all this.
The third form is not typical of the legal code or the disciplinary mechanism, but of the apparatus (dispositif ) of security, that is to say, of the set of those phenomena that I now want to study. Putting it in a still absolutely general way, the apparatus of security inserts the phenomenon in question, namely theft, within a series of probable events. Second, the reactions of power to this phenomenon are inserted in a calculation of cost.
Finally, third, instead of a binary division between the permitted and the prohibited, one establishes an average considered as optimal on the one hand, and, on the other, a bandwidth of the acceptable that must not be exceeded. In this way a completely different distribution of things and mechanisms takes shape.

Some text from "Security, Territory and Population" [Read more]
บางส่วนของเนื้อหาในบทแรกจากหนังสือ Security, Territory and Population (2007)