Thanks for ganulv and jmm99 for providing those links. Podcasts have the advantage that one can absorb the content whiled doing a workout, detailed papers enable one to dig deeper*.

... confession typically trumps physical evidence
This runs through both sources. There are certainly a couple of reasons for it, some already mentioned.

1) The wide-spread disbelief among agents (police, prosecutors, judges, jurors) that false confessions exist or do so at a considerable level is a big one. The amount of proven innocent in that sample that were actually convicted in a trial is amazingly high, which seems to indicate that jurors tend to grossly overweight confessions compared to physical evidence.

2) If a confession is obtained relatively early it seems that the effort to collect & evaluate solid physical evidence is greatly reduced. Limited ressources tend to get shifted to other cases. A plausible story gets constructed and conflicting evidence, if collected gets pushed away. The unique quality of DNA tests enables it collapse a plausible story built around a false confession.

3) The PEACE method seems to force the investigator to underweight the power of confessions. An interesting question to more knowledgable guys out there: Does the greater qualitiy and quantity of physical evidence (forensic science, information technology like cellphone location etc, etc...) make the PEACE approach more attractive and efficient relative to REID? If so in which cases?

*Many things in the paper were quite disgusting, sadly a considerable amount concerned processes of the justice system.