• Story 2. On August 15, 2011, the Times reported that 58,000 students had signed up for a free, online course on artificial intelligence taught by Stanford Professors Sebastian Thrun and Peter Norvig. The numbers eventually swelled to more than 100,000. The phenomenal interest in free, interactive online courses, which quickly became known as "Massively Open Online Courses" or MOOCs, spawned ventures such as Coursera—in which Princeton participates—Udacity, and edX.

• Story 3. On March 10, 2011, the Times covered Senator Tom Harkin's hearings that investigated the disturbing saga of Ashford University. That saga begins in 2005. Bridgepoint Education bought a small religious college in Clinton, Iowa, that was founded back in 1893 as Mount St. Claire Academy. By the time that Bridgepoint bought the college, it was called The Franciscan University of the Prairies and had 300 students. Bridgepoint renamed it Ashford University and increased its enrollment from 300 students to 78,000 students in the space of five years. Bridgepoint's programs had drop-out rates ranging from 63% for its bachelor's degree programs to 84% for its associate degree programs; it spent an average of $700 per student on instruction, $2,700 per student on marketing and recruiting, and netted $1,500 per student in profits. 86% of its revenues came directly from the federal government through student loan programs. Despite these deplorable results, the education offered at Ashford University paid off handsomely for at least one person: the CEO of Bridgepoint Education was paid $20.5 million in 2009 for overseeing a university with a drop-out rate in excess of 63%.

• Story 4. In November 2009, the Middle States Commission on Higher Education found that Princeton University had not submitted sufficient evidence of student achievement and demanded a "progress letter" supplying more documentation to prove that Princeton deserved to be reaccredited. In so doing, the commission ignored the judgment of its own peer review team, which had characterized Princeton's assessment efforts as "impressive." The commission wanted evidence of "comprehensive, integrated and sustained processes to assess institutional effectiveness and student learning outcomes." After multiple interactions with Middle States, we replied with a letter describing in more detail our senior thesis requirement, and Middle States reaffirmed your alma mater's accreditation—for the time being.

These four stories illustrate several of the forces reshaping the landscape of higher education. People are concerned about high costs that make education inaccessible or burden students with mountains of debt. They worry about the quality of education and the value of degrees, not just at new, for-profit providers like Ashford University but also at more traditional universities. They hope that online technology will change the cost structure of higher education to make it more affordable. And regulators, prodded partly by debacles like the one at Ashford University, are applying more intrusive rules and requirements not only for new institutions but for established, successful ones like Princeton.

To understand these developments, we should start not with MOOCs but with money, because cost drives much of the change that is now afoot. To state the obvious: if higher education were less expensive, there would be less pressure for change. So how bad is the problem? At most colleges and universities, the average debt is indeed rising, though the problem is not so extreme as the New York Times article that I mentioned earlier makes it look. And now I need to throw some statistics at you. I promise to go slowly, and I hope that the numbers do not prove too dense, but there is no other way to describe the issue accurately. So here we go.

Despite what you read in the papers, not every college graduate leaves school with debt. A substantial fraction of students graduate with no debt at all. More precisely, roughly 1/3 of all college graduates in the United States leave college with no debt; at the country's leading research universities, about 1⁄2 of all graduates leave with no debt. Among those who have debt, the median debt level is in the range of $13,000 and the mean or average debt is in the neighborhood of $26,000. As you might surmise from these numbers, the New York Times story that I mentioned earlier—the one that profiled a student who graduated from Ohio Northern with $120,000 in debt—was profiling some very atypical students at a very atypical university—Ohio Northern happens to have the highest debt levels of any university in the country.

Just for purposes of comparison, let me note that when I graduated from Princeton, I had maxed out on federal student loans: I owed $7,500 in 1983 dollars. Those of who you who graduated around the same time may find it helpful to know that this would be around $17,000 in today's dollars. That was a substantial number—and I'm not going to kid you, it did affect my choices when I left Princeton—but it was by no means a crippling number, and it is in the same ballpark as the median and average debt for students around the United States who graduate with debt now. Things have changed—they have gotten worse—but they have not changed as much or as fast as people often suppose.

Moreover, for students lucky enough to attend Princeton today, the picture is, as you know, much better. Under our no-loan, all-grant financial aid policy, 75% of our students graduate debt-free. Some do take out loans—for example, to help fund a summer internship. For those who do take out loans, the average indebtedness is less than $6,000—which is less in nominal dollars than the $7,500 in debt that I had when I left Princeton 30 years ago. That's an amazing story, and that's very much a result of the generosity of Princeton's alumni, including you who are in this room today.

Princeton, though, is, in this respect and many others, specially blessed. As I have already noted, debt levels elsewhere are higher. And they have been growing since the year 2000, mostly for three reasons. First, families continue to feel the effects of the recession. Second, states have reduced their support for public universities. This point is really important: over the last decade-plus, the fastest increases in tuition have come in the public sector. During the decade from 2002 to 2011, the share of state university expenses paid for by state governments declined from around 40% to about 28%. The public subsidy-per-student at state research universities declined by around 10% in real terms. Tuition revenues rose to make up the difference. Those increases have not made public universities any wealthier; they have more or less substituted for dollars withdrawn by the states. So prices have risen even though the underlying costs of education have remained relatively constant.

Third, the overall numbers on student debt include the numbers from the for-profit sector, and, unless you study education, the for-profit sector is larger than you think. When I talk about "colleges and universities," I suspect the images that leap to your minds are pictures of places like Princeton or, say, the University of Maryland, or maybe even your local community college. But for-profit universities like Ashford University and the University of Phoenix now consume more than 25% of the federal financial aid budget. The Harkin investigation that I mentioned earlier found that for-profit universities on average spent only 17.1% of their funds on instruction, while they spent 22% on marketing and 19.4% on pre-tax profit. And the loan default rates for students from the for-profit sector are high: 96% of students at for-profit universities take out loans, and they account for 47% of all student loan defaults.

Now, I want to emphasize that not all for-profit universities are the same. Some do valuable work. Some genuinely have a public-regarding mission. But the sector includes a significant number of institutions that are reaping large profits at the expense of both taxpayers and students, and the sector is, as I have said, bigger than you think.

Despite the genuine growth in debt burdens, most economists continue to believe that, in general, a college education is a good investment. Catherine Rampell '07 has summarized the evidence nicely in her New York Times blog and related articles. College graduates may not always be getting the jobs they want, but they are getting more jobs than those without college degrees. The jobs they get also pay better than jobs for those without degrees. Rampell concludes one recent column— published on May 3 of this year—by quoting an economic study suggesting that dollars invested in a college education return on average 15.2% per year—and, folks, not even PRINCO can match that record. Despite all those skeptical articles you read in the paper telling kids not to go to college, it pays to get a college degree, especially a residential liberal arts degree.

For those of you who like to read credit agency reports—or, alternatively, for those of you who are having trouble falling asleep at night—you might want to have a look at report about higher education that Moody's issued in January. It articulated a negative outlook for the entire sector of higher education, and, although I disagree with a few claims in it, I think that the overall assessment is remarkably astute. Here is what the analysts at Moody's have to say:

• "Increasing media attention on student loan burdens notwithstanding, the average debt per student graduating with a bachelor's degree is still relatively manageable, despite the dramatic rise in debt burden over time."
• "Overall, postsecondary education remains a valuable long-term investment. ... Families remain willing to pay for college, BUT their capacity to pay higher prices has been largely tapped."

To put this last proviso another way: even if it would be economically rational for families to take on still higher debt levels, they are either unable to do so or not comfortable doing so. And that makes complete sense from a psychological and a prudential perspective: the risks and the short-term constraints generated by educational debt are simply too daunting, even if the projected long-term return on the investment remains positive. For that reason, we need some other solution that can enable colleges and universities to find more cost-effective ways to provide good quality education. In many parts of higher education, traditional funding models are at a breaking point, and unless we can find new models, quality will slip.

Then along came MOOCs. Here, perhaps, was a magic bullet. The hopes and the hype around MOOCs and other technological innovations depend on the possibility that they can help us not only to provide cheaper education but good quality education. Can they? This is the most important question about the changing landscape of higher education. Technology will produce big changes. In fact, it already is producing big changes. Ashford University, after all, is a technology-driven phenomenon—those 78,000 students that Ashford enrolled didn't actually show up in Clinton, Iowa; they took their classes online. The for-profit sector that now consumes 25% of the tax dollars going to federal financial aid, with dubious results, is a technology-driven phenomenon. So we've already seen change, a lot of it disturbing. The question whether technology can change the landscape of higher education in good ways rather than bad ones will come down to whether it can not only reduce cost but enhance quality. And in education, quality is all about creating and sustaining engagement.

Let me explain what I mean by that. I used to tell people that to be an effective teacher you had to know two things: you had to know your subject matter and you had to know your students. The subject-matter part is obvious: if you're going to teach constitutional law or chemistry, you had better know those subjects. If your students know more about constitutional law than you do, you're going to be in trouble. And so here, where we try to produce the best education in the world for some of the sharpest students in the world, we hire world-renowned experts in the fields we teach. But expertise isn't enough. You have to know how to get that material across to students. You have to know what's going on in their heads, so you can anticipate what they will understand immediately, and what they won't, so you can correct their misunderstandings and help them learn.

As I said, I used to say that you needed to know your subject matter and your students. That's not quite right, either, because knowing your students is not enough; you have to know how to engage them. You need to get your students energetically involved with you in pursuit of learning.

Back when I was teaching first-year contract law at the New York University School of Law, I used to engage the students through what is famously or infamously known as the Socratic method—meaning I would call on students, whether they had volunteered or not, to analyze assigned cases in class. Those of you in the audience who are lawyers probably have bad memories of this technique, because it puts students at risk of embarrassing themselves in public in front of their classmates. But practiced gently enough, it is a great tool for engagement—the professor gets constant feedback on what the students are understanding and what they are not, and the class is involved in a collective process of discovery, where students learn that by careful analysis they can unearth insights that they had previously missed.

So in about my third year of teaching, I started off the academic year asking the same questions that I had asked in the preceding two years, and I was amazed. I thought that I had my smartest students ever. They easily and immediately answered questions that had stumped their predecessors. So I was able to ask some follow-up questions that we never had time for in the preceding years—and you can probably guess what happened. The class quickly fell apart. Unbeknownst to me, students from the two preceding years had compiled all of the questions-and-answers from class discussion into an exhaustive set of notes, and students in the third year were simply parroting the answers aloud. They had no idea how to derive those answers. They were resisting engagement, and until I got to a new set of questions, they weren't learning a thing.

At Reunions last year, Don Randel '62 *67, a musicologist and the former president of the University of Chicago, gave a sermon in the University Chapel where he touched on this topic. He said that musicians have a saying that a great music teacher is not someone who can tell you where to put your fingers but rather someone who can inspire you to practice. Same idea. Engagement is crucial, and it is hard to sustain.

Williams College recently did a study to determine what practices on its campus most contributed to the goals of a liberal arts education—that is, what practices at Williams best inculcated the kinds of knowledge and habits that students wanted in order to lead meaningful lives. Some of you may have read about this study, because Adam Falk, the president of Williams, wrote about it in The Wall Street Journal. What Williams found was, basically, that the practices that mattered most were those that maximized engaged learning—faculty supervised research projects, for example, and intensive reading and writing projects. Interestingly, Williams found that any direct student/faculty contact had a positive correlation with student learning. So, in other words, if students were having lunch with faculty members, or running into them at campus events, those interactions correlated positively with learning, regardless of whether or not the students and faculty were discussing academic subjects. If you have kids in college now, tell them to take their professors to lunch. At a minimum, tell them to go to office hours. Your children will get more out of college, even if all they talk about with their professors is baseball or popular music. This is a statistically proven fact! (I have a theory that explains why this fact is true, though I will save that for the question-and-answer period if any of you are interested).

I said that these conclusions were not surprising to me, and I suspect they are not very surprising to any of you. What Princeton offers, and a lot of the reason why we all come back and wear these glorious orange jackets, is intense and unrelenting engagement. The senior thesis—that's about student engagement in learning, discovery, and a personal relationship with a faculty member. Precepts and freshman seminars—they're about independent thinking and student/faculty contact. Residential life, from the chapel to the athletic fields to the eating clubs, about engaging students with one another and with mentors in a myriad of ways that interlace with what they are learning in classrooms. Some of the most memorable discussions with students about academic topics that I have had since returning to Princeton as a faculty member have in fact taken place at eating clubs, which nowadays occasionally invite professors to visit for dinner discussions. This may reflect a more serious turn amongst our students—I don't remember discussions with faculty members at my own club, though I do remember a lot of very engaged conversations with classmates.

Engagement is essential to learning, and, in general, engagement is expensive. It is expensive because it turns on putting students in contact with faculty members in as close to a 1-on-1 relationship as you can get. So engagement drives labor costs, and labor costs are what make education expensive. You see newspaper articles every once in a while attributing the rise in higher education costs to the creation of fancy climbing walls and other recreational amenities. I can tell you this, after serving as Princeton's chief budget officer for nine years, climbing walls do not drive educational cost—not here, not anywhere. What drives cost on a college campus is your human capital, and especially the teaching faculty. If you cut cost there, you will almost certainly cut quality, because you are diminishing knowledge or engagement or both, and those are things that matter to educational quality.

A lot of online learning is what people call "distance learning." Residential learning like you received here, engaged learning, is the opposite of distance learning. It is "full contact" learning. The question about technology is whether it can increase or sustain the kind of contact and engagement that are essential for learning, rather than simply putting more distance between students and teachers.

As most of you know, Princeton is currently offering a small number of MOOCs through the Stanford-based firm called Coursera. We began this experiment for two reasons. The most obvious one was to make some of our on-campus content accessible to a wider audience. But another reason was that we thought the experiment could actually increase levels of engagement on the campus. The idea was this: many lectures have a rather passive feel to them. Mark Twain described college as a "place where the professor's lecture notes go straight to the students' lecture notes, without passing through the brain of either." That's distance learning, or non-learning, even if everybody's together in the same room. It has nothing to do with engagement. Engagement takes place, if at all, outside of the traditional lecture, such as during office hours. So the question that we asked, and that many people are asking, is whether it was possible to "flip the classroom": could we put the transmission of knowledge online, in recorded lectures, and then use the liberated class time for more interactive forms of teaching? And could we make the online lectures themselves more interactive, by embedding questions into the online format, so that students would participate more actively online than they would in the traditional classroom?

I would say that the jury is still out on these questions. We do not yet know whether online lectures can produce more engagement—or even as much engagement—as in-person lectures. But this much is certain: the faculty members participating in the Coursera experiment have been energized by this opportunity to reimagine how they teach their courses. And that creativity benefits engagement, regardless of whether or not what you've created is legitimately called a "flipped classroom." It goes back to that story I told you about my contract law class at NYU. There is always a temptation, for both students and faculty members, to do what Mark Twain lampooned, to let learning lapse into routine or rote. Rote learning is less effective and less interesting than engaged learning, but also requires less energy and less work. In general, anything that injects energy into the classroom is good. And right now, online learning is energizing classrooms here and elsewhere.

All this leads to two questions, for Princeton and for higher education policy more generally. One of the questions is how do we maximize the engagement of every student, undergraduate and graduate, on this campus? For purposes of sustaining and improving Princeton education, that is the most important of the two questions. If we can get more engagement, our education will be better, and it may not be particularly important to say how much better or better in exactly which ways.

The second question, though, is about how we measure success and demonstrate value. This question is important from the standpoint of public policy and the landscape of higher education more generally. We have plenty of evidence that engagement increases value. But we need to be able to talk about which forms of engagement are most cost-effective, and that in turn will require us to be more explicit about what we are trying to achieve with various forms of education in this country: cost-effective means cost-effective at accomplishing specific goals, and those goals are going to vary from campus to campus, from one institution of higher education to the next.

It is here that I want to fulfill the promise that I made at the beginning of these remarks, the promise to say something not just about higher education but about law. Regulators are struggling to adapt traditional measures of educational quality to new forms of instruction. How can we tell whether online forms of education—whether part of online universities, or part of the curriculum at traditional residential institutions, including Princeton—are providing the value promised to students? Regulators do not want to stand in the way of innovations that might make education more affordable or more effective. But they also do not want to license the kinds of abuses that occurred at Ashford University.

Federal law attempts to address this by requiring institutions undergoing accreditation to provide evidence of "success with respect to student achievement in relation to the institution's mission." (For all my friends at the Program in Law and Public Affairs, I want to emphasize that I have just quoted from Chapter 20 United States Code Section 1099b(a)(5)(A)—so I think we're all kosher now with this being a law lecture!) I think that's pretty good language, actually—but of course the key question is what kinds of evidence counts. Accreditors have started doing something that I regard as a huge mistake—they have said that "evidence of ... student achievement" means quantitative evidence of what they call "student learning outcomes." Some of them would like to see us do something like what the public schools have to do these days: namely, administer standardized tests to our incoming freshmen, administer them again to our seniors, and report on the differences.

That is what led to the fourth story that I told you at the beginning of these remarks, the story about the progress letter that the Middle States Commission demanded from Princeton. Middle States wanted more evidence of so-called "student learning outcomes" at Princeton. When we asked them what, other than standardized tests, might satisfy them, one of their commissioners told a story about another university that had filled an entire room, he said, with fat black binders stuffed with pages and pages of data. He did not say what the data was about. He just said there was lots of it, and it was in big binders that filled a room. He admired the big-binders approach, and he recommended that Princeton adopt it.

I don't need to tell you this, but I'll say it anyway. This fetish for "student learning outcomes" is no friend to education and no friend to engagement. Black binders of data, standardized testing at the college level, new bureaucratic red-tape: these are not engines of learning. They do not promote engagement. They stifle it.

Princeton has accordingly been working actively to promote better interpretations of 20 USC 1099b—better interpretations of "evidence of student achievement." Last November, I joined a small meeting, organized by the American Association of Universities, that brought accreditors and provosts together to discuss this topic. That meeting produced a statement that has now been endorsed by all the major associations of colleges and universities in the United States and by most of the accreditors. It identifies three relevant kinds of evidence of student achievement: