• This is not what Felix Cohen would say. It's a perfectly good thing to say--although a little dogmatic unless the model doesn't actually take any account of whether arrests ripen into convictions--but it's not about Felix Cohen's point, and you'd have done better to leave him out of it.

This nonsensical method of law enforcement, detached from social actualities and ignorant of the consequences of its actions, is reinforced by a judicial system that operates in much the same way. In Tulia, the arrested individuals took turns appearing in court, denying every allegation, and were convicted on the basis of Coleman’s testimony alone

• Not quite true: there was cocaine admitted in the cases, just with no fingerprints that could trace it to the custody of the accused persons.

and sentenced to between 10 and 360 years in prison, until the others had no choice but to take plea bargains for less time. On a case by case basis, this hardly seemed objectionable: Yes, the officer was white and the defendant black, but surely this doesn’t prove racism. No, the officer had no corroborating evidence, but surely his word is credible until proven otherwise. The only role left for the judge, then, was to mechanically provide the appropriate rule and leave the rest to members of the all-white jury, the individuals best able to determine the truth.

In Jerome Frank’s view, to trust that such a system is capable of discovering truth is to ignore the subjectivity and uncertainty that is necessarily inherent in trial court decisions.

• Certainly not. Frank would say that any real system of appellate review of criminal convictions (which Texas doesn't have) would have detected the insufficiency of the evidence sustaining those convictions and reversed them outright. Frank's concern isn't with the impossibility of determining the sufficiency of evidence to convict. You're stopping short of precision in the analysis.

To believe that a conviction in such a setting signifies anything more than “that the trial judge or jury believed the testimony of some witness rather than that of others” [2] is to succumb to legal magic.

• This isn't a clear sentence. If you mean that there wasn't enough to sustain the convictions, you are right. But the problem isn't therefore with the system of law enforcement that attempts to arrest drug dealers, it's with the inadequate functioning of other parts of the legal system: the willingness of prosecutors to bring inadequately prepared cases, the failure of the trial court to require sufficient evidence, and the absence in the Texas judicial system of real appellate supervision of criminal adjudication. Obviously Coleman, like the Vermont rogue narc, Paul Lawrence, who created a similar avalanche of wrongful convictions in St. Albans, Vermont in the 1970s, can reasonably be blamed for the whole affray, but it would be more useful to ask about all the other institutions that failed, and it would be overtly incorrect to blame the one institution you do hold responsible, namely the institution of the narcotics sweep task forces, which have not been shown in any way to have failed solely because there was an itinerant rogue cop running a con game against small police forces.

• I don't think you've shown that at all. What you've shown is a picture in which racism enabled the convictions by biasing local officials and jurors, and in which the savagery and incompetence of Texas justice, already a story and a byword through the civilized world, performs a star turn, as expected. For comparison, the cases in which Paul Lawrence framed people in Vermont did not result in equivalently savage sentences, and although the numbers are much higher (Lawrence arrested 600 people over 6 years, and most--but not all--of his arrests were frames), defense lawyers had the state defender general pay for private detectives to investigate Lawrence early in the avalanche, and only the personal negligence of the State Attorney General (who shelved the resulting report during the heat of a political campaign for reelection) prevented his earlier discovery. When he moved from St. Albans to the larger Burlington force he was almost immediately apprehended.

• The pathologically lying policeman is a problem to all justice systems, but not--as the comparatively infrequent occurrence of these episodes shows--a particularly intractable one in the US context. The problem of Texas, on the other hand, is apparently insoluble.

• I think the Cohen and Frank elements of this essay were misplaced, and did much more harm than good. I think, for reasons outlined in comments above, that there are some significant lapses in legal analysis. I think, once corrected for these problems, the essay would read very differently.