<p>Well we started with the suggestion/premise that MIT’s admissions process was flawed. Then we took it down a notch- some random professor’s claimed that they got it wrong 15% of the time (so 85% of the time the result was fine, OK, outstanding, something other than flawed). But now it seems that it’s not 15%- it’s a couple of kids a year- at most- who want to study string theory, and who take from their MIT rejection that they should become high school gym teachers or something since attending one of MIT’s peer institutions in physics would be too terrible to contemplate. Or that attending Harvard and developing a love for genetics or neuroscience is much too prosaic for such a theoretical 17 year old giant.</p>
<p>So we’re sliding down the slope and now- maybe- it’s one kid that someone once knew who got taken down by John Von Neumann (who family members of mine used to refer to as “Johnny”- they were colleagues). And all of you scientists here (I am not one, just a corporate, boring suit who doesn’t have a cot in the office to sleep on) have used the scientific method and concluded from this one datapoint that MIT admissions are flawed.</p>
<p>With tremendous gusto expended in the process of explaining why the management kids are dumb. The engineers are dumber than the physics majors. Kids who get into Harvard but not MIT have their egos bruised beyond repair and may end up in finance of all things. That people who run MIT admissions are nothing like “real” MIT people and therefore can’t possibly recognize the 17 year old geniuses who they routinely reject for admissions.</p>
<p>Not a scientist here… but any manufacturing company or process I’ve worked with in the past sets an error rate and then develops algorithms and creates fail safes to keep the rate at the acceptable level. Which is why occasionally your cable bill is wrong. Of the tens of millions of bills the cable company sends out, they’ve calculated that if every customer gets a bill with an error about once every two years, that’s an acceptable frequency. Any more accuracy- they’re spending too much money on billing. Any less accuracy, and they will start to lose customers who get aggravated by “my bill is always wrong”.</p>
<p>You with me? So among the boring and stupid MIT management students who are nowhere near as special as the theoretical physics students, they’ve managed to calculate that the heavily manual process of reading undergrad applications has an error rate of what- 3 students a year that they reject but should have accepted, and what— 10 students a year that they should have rejected but in fact accepted. But to create a system with more accuracy would be cost prohibitive for the additional value to the institute. And to create a system with LESS accuracy would erode the mission and values of the institute. So- everyone is on board with their error rate, and nobody loses sleep, within reason, over the fact that the admissions team is not omniscient, and that sometimes, a 17 year old who looks high potential turns out to be a lazy, beer pong playing slug… and sometimes, a 17 year old who looks uninteresting and uninspired turns into a Nobel laureate.</p>
<p>I can’t claim to understand string theory… but I’ve run large and complex recruiting organizations for global companies, and it seems to me that once MIT sets its target on errors and does its six sigma or whatever process improvement to map out the function, the success metrics, and the results, they are smart enough to know that they are never going to be at 100%.</p>
<p>That’s the way a dumb corporate dweeb like me looks at the problem.</p>