Please, that's such a copout. To "hire good people" you need to 1. attract and 2. identify them. If you don't create attractive conditions, good people will go elsewhere; if you do offer, say, financial incentives and personal prestige, bad people will flock too, so you need to identify good people. Further, the whole show is dynamic, good people can turn bad if wrong incentives exist, or you can lose them if more attractive alternatives present themselves.
So it all becomes a process problem, how to select, retain and reward good people. Which is exactly the problem impact factors, grants, publication metrics etc. try to solve, in an objective and reproducible manner. You might have a better system and results, but it's relevant only if it's not based on you personal nose and intuition to "hire good people".
This is particularly a problem for academic labs looking for bioinformatics specialists. Many of the bioinformatians I know with strong stats backgrounds have taken jobs at banks etc where the pay and benefits are significantly better. It's hard for academics to offer their staff something competitive to retain them. Moreover, people are realizing that so much of medical research is less about the patients and more about using their samples to get big papers with little impact on their prognosis
The best strategy to hire and train good bioinformaticians is to let them live where they want and involve them in world-spanning virtual groups. Pay them as well as can be allowed, which is not so great in terms of their fair market value in other fields, but with life flexibility that gives intangible benefits that are awesome. It's been said that having time flexibility with one's work is the surest way to increase our happiness and lifespan.
We need to move past the hierarchical model of staff/leadership to a fluid one in which everyone is respected and does the roles that they can and want to fill. This can happen in a distributed manner. Labs and groups which take a flexible attitude to these work relationships are and will be successful. Eventually it will catch on.
I agree with you. I think more PIs are realizing that. I know of some who have machine learning scientists that have second jobs in industry. They tolerate it because they still do everything asked of them well and in a reasonable time frame. I am not sure how they handle conflicts of interest though.
So it all becomes a process problem, how to select, retain and reward good people. Which is exactly the problem impact factors, grants, publication metrics etc. try to solve, in an objective and reproducible manner. You might have a better system and results, but it's relevant only if it's not based on you personal nose and intuition to "hire good people".