I'd agree with this. I'm in school studying forensic psychology. According to what I've learned so far, the best way to treat offenders and rehabilitate them is to first focus on those who have the greatest chance of re-offending in the future (triage), under the theory that they are the most likely to benefit from treatment (this requires an assessment of future risk for each person). In the treatment itself, it's important to match the appropriate staff to treat certain populations so that they are responsive to the characteristics of the people being treated. It's important to target the criminogenic needs of the prison population being treated; that is, addressing those factors that are known to lead to future recidivism. Why did they commit these crimes, and how can we address the needs that they were trying to get met? Things like antisocial attitudes, substance abuse and dependence and promoting prosocial behaviors are some of the areas that most need to be developed in a treatment team relationship.
It is also known which types of treatment are not effective, types of therapy that can actually increase the chances of recidivism. Traditional psychoanalytic and client-centered therapy don't work well in prison. Neither do sociological strategies that focus on particular subcultures, or retributive programs that focus on punishing the offender (i.e. boot camps). Any program that doesn't address the issues/criminogenic needs that the offender is trying to get met will be unsuccessful in the long run.
Psychopaths are an especially challenging group. As a personality disorder, psychopathy isn't really treatable in modern science as of yet. For a long time it was thought that they couldn't even be treated at all. Treatments that tried to improve empathy in psychopaths seemed to just make them better psychopaths as it taught them how to exploit others. There has been more hope in the last decade that some treatments can help.
Considering that as far as we have been able to determine, psychopathy is present in 25% of all general offenders in prison, 15% of child molesters, and 40-50% of rapists, there is a high chance of recidivism among psychopaths - but these are just estimates as it's very difficult to identify them without a skilled assessment (even then they are easily missed as they're good at imitation/masquerading as normal). It's not their fault they are psychopaths and there is a lot of scientific evidence pointing to biological origins (large areas of the cerebral cortex and the amygdala aren't firing). One of the issues with psychopathy is the failure to learn from their mistakes, again something probably neurological in origin. It's not that they can't see the consequence of a future action they are considering, it's more that they are hyper-attentive to the reward of the decision and ignore the consequence. This allows for greater recidivism in this population.
tl;dr It's best to rehabilitate offenders by treating the problems that got them there in the first place, mainly anti-social behaviors/thinking and substance abuse/dependence. Lack of social support when in the real world is another factor for them. Not all prison populations will be treated successfully, obviously. Cognitive-behavioral therapy is currently the most optimistic treatment method.