It would require very little data to precisely pinpoint who an applicant is from a chance thread and crossmatch his or her screen name to an actual person in an admissions database.
If a school wanted to push its yield up, I think it would be easy to create software that could search an applicant database and pick out applicants based on scores and other info from chance posts. A full-time AO mole to compile the data would be useful in such a venture.
It might be the next step to propelling yield another 5-10% for schools looking to move up the ranks.
Not that I am advocating this, mind you, but just saying it might be a pragmatic way to increase yield.
Whether this is/has been done on a small scale already is anyone’s guess, but with yield being an important factor in USNWR rankings, it would not surprise me.