Since Harris apparently failed to make his own point on the law implications of his viewpoint on free will, I'll have to quote that from one of his main critics, Dennett:
Harris, like the other scientists who have recently mounted a campaign to convince the world that free will is an illusion, has a laudable motive: to launder the ancient stain of Sin and Guilt out of our culture, and abolish the cruel and all too usual punishments that we zestfully mete out to the Guilty. As they point out, our zealous search for “justice” is often little more than our instinctual yearning for retaliation dressed up to look respectable. The result, especially in the United States, is a barbaric system of imprisonment – to say nothing of capital punishment – that should make all citizens ashamed. By all means, let’s join hands and reform the legal system, reduce its excesses and restore a measure of dignity – and freedom! – to those whom the state must punish. But the idea that all punishment is, in the end, unjustifiable and should be abolished because nobody is ever really responsible, because nobody has “real” free will is not only not supported by science or philosophical argument; it is blind to the chilling lessons of the not so distant past. Do we want to medicalize all violators of the laws, giving them indefinitely large amounts of involuntary “therapy” in “asylums” (the poor dears, they aren’t responsible, but for the good of the society we have to institutionalize them)? I hope not. But then we need to recognize the powerful (consequentialist) arguments for maintaining a system of punishment (and reward). Punishment can be fair, punishment can be justified, and in fact, our societies could not manage without it.
This discussion of punishment versus medicalization may seem irrelevant to Harris’s book, and an unfair criticism, since he himself barely alludes to it, and offers no analysis of its possible justification, but that is a problem for him. He blandly concedes we will – and should – go on holding some people responsible but then neglects to say what that involves. Punishment and reward? If not, what does he mean? If so, how does he propose to regulate and justify it? I submit that if he had attempted to address these questions he would have ended up with something like this: Those eligible for punishment and reward are those with the general abilities to respond to reasons (warnings, threats, promises) rationally. Real differences in these abilities are empirically discernible, explicable, and morally relevant. Such abilities can arise and persist in a deterministic world, and they are the basis for a justifiable policy of reward and punishment, which brings society many benefits – indeed makes society possible. (Those who lack one or an-other of the abilities that constitute this moral competence are often said, by everyday folk, to lack free will, and this fact is the heart of compatibilism.)
If you think that the fact that incompatibilist free will is an illusion demonstrates that no punishment can ever be truly deserved, think again. It may help to consider all these issues in the context of a simpler phenomenon: sports. In basketball there is the distinction between ordinary fouls and flagrant fouls, and in soccer there is the distinction between yellow cards and red cards, to list just two examples. Are these distinctions fair? Justified? Should Harris be encouraged to argue that there is no real difference between the dirty player and the rest (and besides, the dirty player isn’t responsible for being a dirty player; just look at his upbringing!)? Everybody who plays games must recognize that games without strictly enforced rules are not worth playing, and the rules that work best do not make allowances for differences in heritage, training, or innate skill. So it is in society generally: we are all considered equal under the law, presumed to be responsible until and unless we prove to have some definite defect or infirmity that robs us of our free will, as ordinarily understood.
And what Dennett talks about with indefinite treatement is more real than you might know; various US states have such statues of indefinte civil confinement on the books; typically for sex offenders. And the detention of these hardly differs from [life] [im]prison[ment].
So basically if were to posit that law would dispense with any notion of responsibility (and thus perhaps even with criminal law altogether), a generalized "civil" containment model as hinted by Dennett is not too unreasonable of an inference.
And Harris (or Dennett) is hardly the first guy to talk of this:
Derk Pereboom has argued for a position called
hard incompatibilism, which asserts that we are simply not the sort of creatures who can
be morally responsible for our actions. [...] But if criminals cannot be held morally responsible for their actions, then there is
a serious problem with [the current justice] practices. Since criminals do not deserve to be punished, it
would appear as though punishing them would be unjustified. I explain that in order to
have properly justified crime control practices, we need to either form a different
conception of moral responsibility, or we need to change our current crime control
practices. In other words, we need to be revisionists about either moral responsibility or
about punishment. Manuel Vargas has argued that we ought to do the former, and
Pereboom has argued that we ought to do the latter.
Vargas suggests that we reconceptualize moral responsibility, but maintain our
responsibility-characteristic practices for pragmatic reasons. Pereboom, on the other
hand, suggests that we stop holding people morally responsible
altogether and instead
revise our system of criminal law to account for this change. [...]
own suggestion for a system of crime control [is] the incapacitation approach. This model
suggests that we justify incapacitating criminals by invoking the right to protect ourselves
as an analogy to quarantine.
A slightly more elaborate/concrete proposal, by a biologist (Cashmore, 2010):
Progress in understanding the chemical basis of behavior will make it increasingly untenable to retain a belief in the concept of free will. To retain any degree of reality, the criminal justice system will need to adjust accordingly. However, to retain a degree of orderliness in society it will still be necessary to incarcerate individuals found guilty of certain criminal acts. This is rationalized in various ways including the following: To a), protect society; b), protect the offending individuals from society; c), provide such individuals with appropriate psychiatric help; d), act as a deterrent (the act of incarceration and the presence of a criminal code forming part of the environment); and e), alleviate the pain of the victim. The proposal is a pragmatic one, based on the belief that the welfare of society at large is more important than the welfare of the individual offender.
One might ask: How does this proposal differ from the present system? Whereas in some ways, not significantly; in other ways it differs fundamentally. The primary difference would be the elimination of the illogical concept that individuals are in control of their behavior in a manner that is something other than a reflection of their genetic makeup and their environmental history. Furthermore, psychiatrists and other experts on human behavior should be eliminated from the initial judicial proceedings—the role of the jury would be to simply determine whether or not the defendant was guilty of committing the crime; the mental state of the defendant would play no part in this decision. However, if a defendant were found guilty, then a court-appointed panel of experts would play a role in advising on matters of punishment and treatment. This is a system that would hopefully minimize the retributive aspect of criminal law; concerns about this aspect of law, which have probably been around since laws were first introduced, include those expressed by Wootton (33), Menninger (36), and, more recently, Greene and Cohen (37). Also I note that I am not the first to propose that psychiatrists should be excluded from the initial court proceedings; Glueck (38) and Menninger (36), for example, who both had substantially more expertise than I have in this field, long ago made similar suggestions.
According to this view, juries should only establish external facts, leaving the inner working of the defendant entirely to the punishment/treatment phase, which is to be handled by experts.
It's interesting to note in connection to these proposals that a 2012 controlled experiment published in the high-impact Science journal, found that when US judges are presented with a "biomechanism" for psychopathy (in this case a combined genetic and neurological explanation: low MAOA activity and atypical amygdala function) they tend to give a somewhat shorter prison sentence (from 13.93 years to 12.83 years). Note that his was not a forgone conclusion (that the sentence would be shorter), as the likelihood of recidivism is presumably higher under a biological determinant, and this has to be taken into account by US judges upon sentencing (this is elaborated in the introduction of the paper). Nevertheless, the judges saw on average the biomechanism as more mitigating than aggravating.