According to this paper the most influential criteria engineers and scientists use for selecting information resources are not quality or even familiarity but 1. the time it takes to track down information and 2. the authoritativeness of the resource. Perhaps this explains why researchers are increasingly using Google Scholar rather than PubMed (PubMed is more authoritative, but Google’s ranked results allow you to find publications faster) and why the NCBI still receives more requests for protein-related data than we do (both sites are equally bad at searching, but NCBI may be seen as more authoritative).