Bottom line for me is that Columbia or any other school which as a policy rejects a class of medical students for any reasons other than their individual abilities says more about the institution then the potential applicants.
A problem we have here is the only real and accurate way to gauge one's quality is to see them work in the field for an extended period of time. No one can do that when there's thousands if not more applicants for a few spots.
I've taken courses in Industrial (AKA organizational/personnel) psychology. There is a lot of published data about the process by which institutions take applicants. Unfortunately no method that is used is very good. Letters of recommendation barely have any validity. The best indicators are standardized test scores, but as we know, that is not a very good method. We've all known, I'm sure, several people who score poorly on such tests but are great clinicians.
From the institution's viewpoint, given their limited resources in taking applicants, many programs do the stereotypical "we'll make several piles, anyone below this score, a D.O., an FMG, (etc) will not even be consisdered."
While this method is arguably a "fast food" method of filtering through applicants, unfortunately there really isn't much that can be done. I know plenty of D.O.s and FMGs that IMHO are better than many AMGs. The only two residents I know of that were kicked out of my program were AMGs who both scored very well in medical school and the USMLE. That written, no matter how you slice and dice it, if you start doing the "pile" method, you will in general tend to get better applicants though you will also throw out several great applicants that you never would've even had the chance to consider.
It's all statistics. If I had 10,000 applications and I use the pile method, however flawed, when I widdle down the applicants to 1,000, I will on average have better applicants than not vs the 10,000 I started with. I would not have the time to give every single applicant the time and consideration that person truly deserved. A rough analogy is a google search. That search engine uses an algorithm to pick the best sites, but any algorithm is not perfect.
The real shame of this type of method is I've seen people that IMHO would've been great doctors but they could not pass the USMLE, or they were blown out of the water while in pre-med by an undergraduate college that had a weed-out policy where several courses were designed to fail a large portion of the students--no matter what. Several of these people are victims of the way the system is designed, but unfortunately, it's the best we got considering the circumstances.
By the time people are eligible for fellowship, things change. There are far fewer applicants for fellowship positions and as a result, the programs can give more individualized time in weighing a candidate, though the reverse is also true, the programs will also accept poor candidates due to lack of applicants that otherwise would've been widdled out using the "pile" method.