This seems like it's a problem with the people that were used in the study, not inherent within the system that they used. What would you suggest, after having

This seems like it's a problem with the people that were used in the study, not inherent within the system that they used. What would you suggest, after having read this, to not attempt meritocracy at all because it will be biased when those making the decisions are biased? Replace them with algorithms. Feed the algorithms objective metrics that you wish to optimize. This is much easier said that done, but now your meritocracy does what you intended it to do. If you encounter problems, the metrics are not what you want them to be, or the problems are problems at another level, which you cannot solve from changing this part of the system.