<p>I think a computer with a sophisticated enough design would be able to competently produce a satisfactory class (with the help of human admissions officers for the more subjective parts of the application). Numbers may not make up your entire app, but they still count for a good percentage of it. With that part of the app at least, a computer could be very adept at building the class.</p>
<p>Computers can, and at some schools do, perform a first round cut-down to weed out the unqualified who fall below some GPA/SAT threshold. And at mondo state schools that decide the whole issue on stats alone, they can do the whole process quite efficiently. But if you really do want to consider essays and recs you are going to need a human being to read them and decide, at least to a limited degree, what kind of person is behind them. That’s where the subjectivity comes in. Computers don’t do subjectivity.</p>
<p>If a class based on stats alone is “satisfactory,” then computers can indeed produce a satisfactory class.</p>
<p>I’ve only skimmed the thread, but it is entirely possible to separate 1) rating holistic components of the application and 2) deciding whether Applicant X is admitted/waitlisted/rejected. Most admissions officers already “rate” essays, LORs, leadership, etc. (even “personal qualities,” as at Duke). However, these ratings are then considered in the context of a diverse class. A computer is perfectly capable of taking sophisticated models of said diverse class, with objective and subjective “numbers,” and using random number generation to decide tiebreakers.</p>
<p>
</p>
<p>Contrast that with the system we have today. As Robert Frost once said, a jury is a group of 12 people chosen to decide who has the better lawyer. Somebody whose DNA very strongly suggests that he knifed his ex-wife and her male suitor to death, was later found with mysterious cuts on his hands, whose socks and car were found to contain blood of the victims, whose shoe size matched bloody footprints found at the crime scene and which past photographs showed he wore the type of shoes that matched those footprints, and for which witnessed testified seeing the guy wearing a sweatsuit whose fibers matched those found at the crime scene …but can still be found not guilty because of the admittedly ingenious (if racially cynical) defense presented by arguably the greatest (and most expensive) team of criminal defense lawyers ever assembled. {But hey, at least the search for the ‘real killers’ could begin in earnest, right?} In contrast, I suspect that there are plenty of people convicted of crimes they did not commit because they simply didn’t have the money to hire a star legal defense team.</p>
<p>
</p>
<p>Well, I don’t know about that. Let’s face it: most kids deliberately cherry-pick their teacher rec’s by only asking those particular teachers who they know will wrote them strong rec’s. Savvier students will actually identify those teachers early on in their high school years and ‘build social capital with them’ (a.k.a. brown-nosing) so that they can then later ask those teachers for rec’s. Heck, I know one student who had a rather notorious reputation at his high school who then took some community college night ‘enrichment’ courses with the express purpose of hunting for additional instructors from whom he could get recommendations.</p>
<p>
</p>
<p>I think now we’re getting to brass tacks. It seems as if the real reason why schools like to use human adcoms is that it allows them to admit whoever they want, and not only for the purposes of race (or religion) based affirmative action. If the university decides it wants to admit somebody whose family donated a billion dollars, or a prince/princess of a foreign country, or, more mundanely, the child of a current faculty member or administrator (perhaps even the child of an adcom officer), they have the freedom to do so by touting the ‘diversity’ of the class they admitted. They can always claim that somebody’s essays/rec’s/interview were more ‘convincing’ than another’s, and nobody would ever know the real truth.</p>
<p>
</p>
<p>Contrast that with what is happening right now - an entire cottage industry of ‘admissions coaches’ exists to help rich children craft essays that convey the proper emotional weight and otherwise compile the most convincing application packet possible. Some of these coaches were former adcom officers themselves and so they know exactly how to craft an application to garner the most attention. But their services are available only to those who can afford it.</p>
<p>I have a different take on holistic admissions and I’ll use a sports analogy as an example. Some teams in the NFL at draft time place more emphasis on the raw physical skills and combine scores of a player … this is similar to a hard rules/test score based version on admissions. While other teams use a more holistic approach of looking at players … choosing to add their own subjectives evaluations beyond the terst scores of whom will make great/good NFL players. If a computer program picked all the players for the NFL then Drew Brees might not have ever been given a shot at the NFL … he was too short (good but not great GPA), not big and tough enough, and didn’t have the arm strength (good but not great SATs) to be a top flght NFL quarterback … but the computers would be wrong as he has “it” and he has “it” in spades. Even trying to measure his game “smarts” would be incredibly tough … lots of very smart QBs (measured by tests / book smarts) such as Drew Bledsoe did not have superior game smarts. </p>
<p>Are students so different? In my town in each graduating class over the last 5 years or so if you were to ask me who would be the most successful in life … the list would be far from an ordered list of the kids with the best stats … some of the kids who look good on first draft on paper don’t have “it”; they willl do fine but not as well others … these other kids despite their lower “stats” just give the vibes of future success … it might be their confidence or the way they interact with others or their passion … whatever it is people have the same perception of their future. Why should colleges not want these kids? and shouldn’t they want to try to find them?</p>
<p>
</p>
<p>I’m afraid that I can’t agree with that logic. Gaming of the system is already pervasive under the current human-based admissions process. Plenty of high school kids engage in EC activities not from true interest but just because they think it will make them look good to college adcoms, as no easy measure exists to ascertain how dedicated somebody is to a particular EC. {I remember during my high school extracurricular activities, probably no more than 25% of the participants truly cared, the rest were there simply to check a box on their college applications.} As I mentioned in a previous post, rich families hire admissions coaches to shepherd their children’s application process, and that may include advice on which extracurricular activities they should choose in order to ‘stand out’ by building a ‘hook’, and tips on how to write impressive and emotionally invested essay answers. </p>
<p>I therefore don’t see that a computer-based system would be any more ‘gameable’ than a human-based system. The computer-based system would have to incorporate factors that are not easily gamed, such as the 2400/4.0 that you mentioned previously. The system would also not have to be published publicly, which then obviates the concern that somebody would be able to easily search for the formula that determines admission.</p>
<p>Since you invoked the specter of high school applicants unearthing a quantitative admissions formula through a ‘simple Google search’, I would proffer the counterexample of the Google Search Engine itself, as well as the intertwined Google Adwords advertising platform. The order of Google search results and the accompanying advertising is determined not by human editors but by Google’s proprietary algorithms. Every website in the world would like to be the first result returned for a corresponding text string input to the Google search engine, and similarly with every advertiser and Adwords. But nobody has been able to consistently ‘game’ the Google search algorithms. Nor can you find the details of those algorithms with ‘a simple Google search’. {Granted, certain basic functionality of the algorithm is widely known, being within the public domain, and intelligent guesses can be made regarding certain features of the algorithm, but nobody outside Google knows exactly how the algorithm works, for Google treats it as a closely guarded secret.} </p>
<p>Nor is Google the only company that has been able to successfully retain corporate secrets away from prying eyes. Coca Cola, most famously, has been able to conceal the exact ingredients of the formula of its soda for more than a century. KFC’s 'finger-licking-good" recipe of 11 herbs and spices that season its fried chicken has remained secret for decades, locked in a computerized vault, access of which is available to only two KFC executives, neither of whose identities the company will reveal. McDonald’s refuses to disclose the exact ingredients of its Big Mac’s ‘special sauce’. {Granted, many guesses have been made as to the exact contents of each of these secret recipes, but nobody really knows for sure.} Or, heck, how about the complete source code of Microsoft Windows/Office, or to Apple Mac OS X? Computer hackers would love to have that to find and exploit all of the security weaknesses they would surely find. But you can’t just perform a ‘simple Google search’ for the source code of Windows. </p>
<p>The point is, companies successfully maintain trade secrets as a matter of routine. Hence, I see no reason to believe that universities couldn’t maintain the secrets to their ‘admissions algorithm’. The universities could tweak those algorithms as necessary just as Google regularly does to its search algorithms to ensure that the public never knows the latest version.</p>
<p>
</p>
<p>The question is not whether a computer-based algorithmic system would be perfect. Everybody agrees that it would not be. Everybody agrees that errors would be made due to the limitations of the algorithm.</p>
<p>The real question is whether an algorithmic system would be better than the human-based system, and let’s face it, the human-based system makes mistakes as well. To extend your sports analogy, history is replete with examples with players drafted highly for holistic reasons rather than purely physical traits…yet who nevertheless turn out to be busts. For example, I distinctly remember the Golden State Warriors drafting Todd Fuller in the first round despite his lack of athleticism under the rationale of his ‘great character’ and ‘disciplined play’ - this during the ugly aftermath of the Latrell Sprewell choking incident. Fuller’s ‘great character’ was good for 3.7 points and 3 rebounds a game during his NBA career. </p>
<p>I’m also not sure that your Brees vs. Bledsoe example quite demonstrates your point, at least not yet. Whatever else you might want to say about Bledsoe, at least he did quarterback the Patriots to one Superbowl appearance (in 1997) and helped them to a Superbowl victory in 2002 by filling in for an injured Brady during the AFC Championship Game, where he led the Pats on the go-ahead touchdown drive. Hence, Bledsoe has a ring. Brees, so far, has never led the Saints to any Superbowl appearances. We’ll see what transpires in the next few weeks.</p>
<p>
</p>
<p>I hardly see the proposal as a ‘worsening’ of the elitism within the process that exists right now. Again, the question is not whether the SAT or any other numbers are perfectly reliable, but whether they are reliable relative to the human-based system used currently. As a counterpoint, there are threads on CC regarding students with low SAT scores - but who were likely admitted by ‘holistic’ criteria - who have also flunked out. </p>
<p>
</p>
<p>Yet it also boggles the mind that a human-based system could reliably make judgments regarding those issues. Regarding the oddly appealing essay, that begs the question: oddly appealing to who? An essay that one adcom officer finds oddly appealing may be downright repulsive to a different adcom officer, and so the fate of your application now rests on which particular adcom officer it happens to reach. One may not be able to properly program every single keyword and variable into an algorithm, yet those keywords/variables may not be properly weighed within a human-based system. Who’s to say that a human adcom officer will make the right choice regarding the worthiness of polo-horse breeder vs. an oboe player? </p>
<p>The takehome point is that while a algorithm-based system would not be perfect, neither is a human-based system. Let’s face it: people make mistakes all the time. Let’s dispense with the false trope that human-based systems are always able to make perfect decisions.</p>
<p>I have no idea how a computer could analyze an essay for “fit” or analyze which extra curricular activities are more outstanding than others, or read letters of recommendation to get a feel for the student.</p>
<p>Could public schools use an automated system? Yes.</p>
<p>
</p>
<p>Why only public schools? Because of their size? If that’s the rationale, then I should point out that certain private schools such as NYU are actually larger than many public schools. </p>
<p>
</p>
<p>One stopgap maneuver would then be for humans to assess only those components, and leave the rest of the application to be analyzed algorithmically. That would then limit the power of an adcom officer to indulge his personal preferences, for example, in admitting his own child or the child of his friend. </p>
<p>In the future, I don’t see why an algorithm one day would not be able to assess the quality of essays, rec’s, or extracurricular activities through some combination of natural language processing and textual analysis. After all, right now, computer algorithms have been designed that can compose their own highly respectable works of music, which means that software has been developed that can ‘learn’ how to assess the quality of music to the point of being able to produce original works. If software can analyze an inherently subjective and humanistic activity such music, then I don’t see why software couldn’t be used to analyze creative texts someday. </p>
<p>And again, I would reiterate the point that humans don’t exactly analyze subjective answers perfectly either. Do adcom officers always know which EC activities are more impressive than others? Are they always able to properly judge essays for fit? Are they always able to properly determine a feel for the students through rec’s? Or do they just take those questions simply as opportunities to indulge their own personal preferences, such as admitting the children of their friends by claiming that their essays and rec’s exhibited perfect ‘fit’, an inherently unfalsifiable statement?</p>
<p>Software algorithms surely aren’t perfect. But neither are human beings. But algorithms carry the advantage of determinism - you always know exactly why somebody was admitted or not. Algorithms also never suffer from problems of agency: they have no hidden agendas they’re trying to fulfill. In contrast, if a human being chooses to admit or reject somebody, you never really know exactly why. If a corrupt adcom officer simply wants to admit all his friends and reject all his enemies, nobody would ever really know.</p>
<p>wow
Magister dixit</p>
<p>With the recs thing, I think it’s very hard to assess them. You could definitely point out the recs that don’t recommend the student, but you can’t pick out which recs point to a really good student. </p>
<p>Most teachers don’t know how to write good recs. </p>
<p>The top1%/top5%… probably says a lot more about the student.</p>
<p>Someone could probably learn the code/algorithm and then make their essay gibberish, but get accepted anyway.</p>
<p>sakky–don’t take “oddly appealing” to literally mean <em>odd</em> to the point of seeming repulsive to another reader. I just meant that some things can’t be quantified. Unlike math, where there’s a right or a wrong answer, English essays rely on countless word choices and sentence structures. An essay can be more or less persuasive or interesting depending on the skill of the writer. And I don’t see how a computer algorithm could test those things. Sure, a student’s essay could be “buzzword compliant” and score high according to the computer, but a human would see if the thing actually made sense.</p>
<p>An analogy–a computer program has been invented to create music in the style of Mozart. What it produces is very good, BUT…it’s not Mozart. It lacks that [italics]je ne sais quois<a href=“creativity?”>/italics</a> that musicians will recognize as essentially Mozart.</p>
<p>What if computers do the preliminary essay read and eliminate the first round, and then an actual human reads the second round? </p>
<p>So computers could eliminate applications down to 2-3 times the number of people that would be accepted, and humans read the finalists?</p>
<p>Another advantage of having a computer making selections would be the huge saving in 30+ adcom salaries+expenses. I guess this would be like a few million/year. Maybe 100 more students could get financial assistance with this.</p>
<p>^We all know that admissions people are overworked and underpaid.</p>