The most credible so far, IMHO, is SRI's independent testing/replication of results of
Brillouin's NiH reactor system. For whatever reason (probably due to the ongoing dispute with Rossi), Dewey tends to minimize IH's involvement with Brillouin, but apparently they are an investor, and are involved.
I hope you are right about SRI's testing of Brillouin's device. I fear probably it is a false positive. I would also remark that it is not a replication, but simply involvement in the calorimeter design and running the device at SRI's facility. They saw similar results when the device was operated at SRI as was reported by Brillouin.
After many personal attempts to partially replicate Brillouin's electrolysis system, I came to see that the most likely error would be on the input power measurement. I have suspected that BE has made mistakes with this all along, because it is quite difficult to measure the input power precisely with the Q-pulse system. BE used an ohmic control in previous electrolysis experiments, which I do not trust based on a lot of experience with this producing false results. You absolutely must perform an electrolytic control using material theoretically inactive (e.g., the only thing changed is the cathode material).
Tanzella of SRI noted that he just basically relies on the expertise of BE's electrical engineering for the input power measuring system. While understandable to an extent given that he may have felt it was beyond his expertise, it is the weak point of the entire system. They need to calibrate using a theoretically LENR inert metal in the reactor using the full range of pulse width and input power settings. They need to run secondary and tertiary methods of measuring input power. Rest assured, it is not overunity measured at the wall socket.
I don't blame IH for not investing further. They (BE and SRI) have not disproven potentially valid alternative explanations at this time.