Putting police officers on the stand in criminal cases as "experts" is always a touchy issue. They constantly give testimony that goes to an ultimate issue. For example, in gang enhancement cases, the prosecutor always gives the officer a "hypothetical" with the facts of the actual case in front of the jury and asks the officer if, on those facts, the conduct of the defendant was for the benefit of the gang --to which the officer always responds "Yes". The Court of Appeal is pretty much fine with that.
At the same time, we don't allow officers to tell the jury that, in their opinion, the defendant is guilty given the evidence against him. That's too much. Everyone knows that too. (Though, as everyone knows, the line between permissible and "impermissible" testimony in this regard is definitely a fine one.)
The Court of Appeal concludes that the officer's testimony in this case (out of San Diego) is more in the second bucket than the first, so reverses the conviction and remands for a new trial. Which makes sense. Can an officer, as an "expert", testify on "expert" things that a jury wouldn't (allegedly) know; e.g., the nature of a gang? Sure. But what s/he can't do is give "expert" testimony outside of their alleged expertise.
So, here, when the police officer testifies that the defendant -- rather than the other guy in the car -- was the actual shooter, based on the testimony of another witness, that's not expert testimony. The jury could hear that other witness. The officer isn't an expert on who shot who. It invades the province of the jury (or, to put it another way, it's way too influential) for an officer to testify to this ultimate fact based upon his own view when that view is no better (or worse) than the jury's.
Do police officers know a lot? Sure. But they're not "experts" on crime generally.
And when the trial judge let's 'em testify too broadly, as here, we gotta throw out the conviction and do the whole thing all over again.
Unfortunately.