Selection Dysfunction

Medical students don't just apply for residency positions at graduation -- they "match" into them. The students rank their favorite programs, the hospitals rank their favorite students, and everyone hopes for the best as an algorithm puts them together. This process, more or less unchanged for decades, across many specialties, went terribly wrong last week during the urology match ...

I still haven't seen any coverage of this in the mainstream media -- I only heard about it on the interview trail, and there are some discussion boards talking about it.

As far as I can tell: the original Urology match on January 24th left several prominent hospitals with unfilled positions. The odds are strongly against places like Hopkins and Baylor having unfilled spots, when so many excellent applicants rank them highly. It became clear that something went wrong.

A re-match was run on the 27th. I don't know if this kind of error, and solution, has ever happened before. A medical student on a discussion board posted part of this email from the AUA:

After hearing concerns about this year's results, it came to our attention that there may have been a problem with our automated match process and that a manual audit was necessary. Upon careful review, we found that one of the criteria in the match was not applied correctly, causing some outcomes to be skewed. With this evidence, it is important for us to perform the match again so that all involved are given the opportunity to be treated fairly...

...We know that this will change some results, but by applying the criteria appropriately, we are confident that the results will align with what you have come to expect from the AUA Residency Match.

We're all told how the match algorithm works, but the actual computer decisions made each year are essentially a black box -- to applicants and institutions. As far as I know, everyone operates on good faith that the computer is working properly. It takes a monumental glitch like this one to trigger a re-evaluation. I can't help but wonder if smaller glitches, maybe affecting fewer programs or individuals, go uncorrected each year.

If you doubt the inconvenience of this, listen to some discussion participants, from the period before the re-match:

...I cancelled 7 general surgery interviews today and told programs to withdraw me from consideration as I matched in Urology. If the AUA match were rerun, damages for those who have made plans based on its results are quite significant (housing commitments, decline of other opportunities, etc.). Damages to those who did not match due to the "errors" are significant as well.


...I had expected to match at my home institution who are completely surprised I did not match with them since they matched someone way down their list. The PD almost called me a liar for telling them that I ranked them 1 when in fact I did. Please help fix this, how can I contact you or give my contact info anonymously. I would like to take action and bring it to the AUA's attention. Something just isn't right here...

This is like those Diebold electronic voting results -- with no paper trail, you just have to hope it's correct. Unless the system is transparent, it will always be vulnerable to head-scratching and conspiracy theories.