In actuality, when lawyers refer to "Brady" they are referring to two related requirements. The first, which finds its roots in Napue and Giglio, is the requirement that the prosecution turn over any evidence the government has which would impeach a prosecution witness. The second, coming out of Brady, is a requirement that a prosecutor turn over any evidence the government has which is exculpatory. The case law is clear that in either case the prosecution's requirement to disclose is not just limited to what he knows. If it is known by any government agent the prosecutor is required to disclose it, even if the prosecutor never knows of the evidence's existence. There are cases in the lower appellate courts which state that if exculpatory evidence is not turned over prior a guilty plea is not valid.
At the heart of Brady is an abiding belief that the government will play fast and loose, or downright cheat, to get its man. As unpalatable as that thought is, anyone who reads the decisions applying Brady can see that in some cases prosecutors have. In such cases the prosecution deserves to get dinged. There's even a valid reason for the refusal to allow the prosecution a “good faith” exception. After all, if someone was willing to deny a defendant evidence that he might be innocent it's not much of a step to claim no knowledge of the evidence at all or to engage in willful blindness. Thus, the rationale for the Brady rests on solid foundations.
Even when everyone is acting in good faith, there are problems when Brady bumps into reality. In reality, there are two major problems with Brady. The first is that the prosecutor and the defense attorney are going to have different views of what evidence is exculpatory. Any defense attorney worth her salt is going to see just about everything as exculpatory; she'll want to know if 3 decades prior the complaining witness' grandmother once uttered a slur about Colombians in front of the witness because that could effect his bias toward people of Colombian descent (the immigrant defendant's place of birth). On the other hand, prosecutors are going to take a rather narrow view of what constitutes exculpatory evidence; if nobody else has confessed to the crime and the DNA hasn't come back as belonging to someone else the prosecutor probably isn't going to see it as exculpatory. How can this be resolved? Well, there is always the option of having the judge review the evidence in camera (reading over it in private in his chambers) in order to see if he thinks there is evidence which the prosecution must turn over. Of course, this is not practical. No judge is going to sift thru the evidence in every single case. Thus, short of something unusual happening, Brady evidence will depend entirely on the prosecutor's interpretation of whether it is exculpatory or not.
The second problem is in almost every case the prosecution is constantly in technical violation of the Brady line of reasoning. If the lab which did the drug analysis fired a guy 5 years ago for sloppy work that's impeachment material. If a single witness has been interviewed more than once he will have said something which isn't exactly the same as he said the first time. Those variations could technically be used to impeach the witness. If there are two witnesses there is no way their stories will be exactly the same; the fact that the witnesses vary in describing the defendant as wearing blue or green shoes is technically exculpatory because it could bear on the identification.
Now imagine a serious, complicated case – you know, the kind wherein the prosecution has dedicated an entire file cabinet just to this case. 2 people were killed and initially the police didn't know who did it so there is a drawer full of forensic reports from two labs (ie: one analyzed things like carpet fiber and DNA – the other checked the deceased's computers). 20 police canvased a ten square block area and got three "confessions." One was from Crazy Joe, the town drunk who confesses to everything. The other were by two smartaleck college fraternity pledges who were required to walk up to an officer and confess as a fraternity prank. The patrol cops don't even bother to tell the investigating detective. A year later a guy in jail contacts the police and tells them that his cell mate has admitted to the murder. It is too late for him to get a sentence reduction; however, 2 years later, a month after the trial finally comes to its conclusion the inmate has a parole hearing (which the prosecutor didn't know about until he got a letter 3 days before it asking for "a good word." Et cetera. &cetera. Etc. There will definitely be dozens, probably hundreds, maybe even thousands, of things in this kind of case which will cause even a prosecutor who has an open file policy to fall short of Brady requirements.
The federal supreme court has tried to further clarify its position in a case called Kyles. This case held that a Brady violation was neither subject to a "harmless error" nor a "sufficiency of the evidence" review. Instead, it established the "verdict worthy of confidence" standard; the evidence not turned over must be "material to either guilt or punishment" and there must be a "reasonable probability" that all the cumulative undisclosed evidence would "undermine confidence in the outcome." While I don't think this is as low as probable cause, the court does portray it as a lower standard than a preponderance that the undisclosed evidence would have resulted in an acquittal. In the court's own words,
"The question is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence.The court goes on to lay the responsibility for determining when the cumulative evidence meets the level that its disclosure is required in the laps of the prosecution and makes it clear that it is no excuse if the prosecutor did not know of exculpatory evidence in the hands of someone in the government.
. . .
The possibility of an acquittal on a criminal charge does not imply an insufficient evidentiary basis to convict. One does not show a Brady violation by demonstrating that some of the inculpatory evidence should have been excluded, but by showing that the favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict."
Brady remains an unsatisfactory situation for all parties involved. Like much of the justice system, it's an imperfection lacking a more perfect solution. Defense attorneys are not going to be satisfied with government disclosures. Prosecutors are left with an amazingly high sounding standard which provides little in the way of actual guidance. Appellate courts are in an even worse position. What puts a case in a "different light" in rural Pitcairn County could be quite a bit different from what does that in the City of Shire. Different jurors, different judges, different attitudes about violence or drugs or the police, &cetera, could all bring a different meaning to a "different light."
It's a great principle. It's just not pretty to watch in action.