First, there is the matter of plea bargaining. While plea bargains are usually available in criminal cases, they almost always require a modicum of attorney-client trust. Why? First, if the deal requires a defendant to cooperate with the government, the defendant must trust the lawyer to handle this sometimes dangerous transaction properly. Second, when a deal is available, many defendants will not accept it if they think their attorney has negotiated inadequately or, worse, is in league with the prosecution. The higher the stakes in the case - when a defendant is facing decades in jail, for example - the more a defendant must trust his lawyer.
There is also the matter of trial preparation. Defendants often have a great deal of knowledge that can help secure a better deal, or result in a better trial outcome. They know witnesses. They know the facts of their own crimes. They know their own personal history. But defendants are often reticent about sharing this information with lawyers they don't trust.
Finally, there is the trial itself and, particularly, a defendant's decision whether to testify. Defendants often want to tell their story. Defense lawyers often want them to remain silent. This decision ultimately rests with the defendant. If he doesn't trust his lawyer, he is less likely to listen to her advice.
These subjective notions of trust and solid relationships, Filler argues, are pervasively under-valued by Court precedent. Presumably, the focus of some jurists (e.g., Scalia) on "brightline" rules has something to do with this. It's really difficult to quantify how much "trust" is enough trust, or when a more trusting relationship would have yielded tangible benefits to a given case. However, in my view, where ex post facto review is difficult, that makes it all the more important to utilize pre-event proxies to try and maximize these values.