“We’re going to talk about eyes today; we all have two,” she says—no, decrees—to 22 students watching her from behind laboratory benches. “Why?” A small pause, short enough to escape conscious notice but long enough to underline the question’s significance. And when Matthews launches into her introduction, she paces and gestures, delivering lab instructions in a short, clear, urgent, almost theatrical fashion. “At the end of the period,” she commands, “this is what you should be able to answer.”

The class breaks into groups, and Matthews, a 29-year teaching veteran who studied opera in college, is darting from group to group. “Does it matter if you have glasses?” she repeats a student’s question. “You tell me. The exercise is about why you need two eyes. Does the fact that you have glasses change the fact that you have two eyes?” The student shakes his head no. “That’s right,” exclaims Matthews, smiling, arms thrown up, and eyes wide open. “See? You answered your own question!” Principal Joseph Wall considers Matthews one of the school’s best teachers.

Upstairs, Cynthia Avezzie’s English class has a decidedly mellow tone, because students are contemplating great literature. They have read canonical works like George Orwell’s Animal Farm and less traditional texts like Amy Tan’s The Joy Luck Club. “So what is good literature?” Avezzie muses, as much to her students as to herself. A girl offers that literature is good when it gives life to a universal feeling, such as loneliness. “That’s very good,” says Avezzie, picking up a pink ball and throwing it to her. “When have you felt lonely?” The student catches and cradles it as if it were made of crystal. “When someone forgets to pick me up,” she says finally, tossing the ball to a classmate to give another example. Principal Wall considers Avezzie also one of the school’s best teachers.

Yet where Matthews is fierce and almost dictatorial, Avezzie is mild and almost maternal. Where Matthews demands, Avezzie coaxes. Where Matthews relies on passion, Avezzie relies on care. But both are good. “They teach from the heart,” says Principal Wall, and that makes the fundamental and national question of what constitutes a good teacher fiendishly frustrating for school administrators around the country.

Consensus in education is hard to find. But it was found in the 1960s, after the Russian launch of Sputnik highlighted the importance of math and science, and it is being found again, today, in teacher quality. Education Week noted recently that “pressure to improve the quality of the teaching force has never been greater,” and everybody—politicians, academics, media, and teachers—has declared that better teachers are the cure to America’s educational woes and the key to continued global leadership. They are the “next big thing in education reform” according to National Journal, an “educational birthright” according to the National Commission on Teaching and America’s Future, and more important than class size or vouchers, according to a 1998 Harris poll and empirical research such as a 1996 University of Tennessee study. “Fine buildings, equipment, and textbooks are important,” argued Eugene Hickok, Pennsylvania’s secretary of education, in 1998. “But it is the skill and dedication of the teacher that creates a place of learning.”

The trouble with this consensus is finding a clear formulation of what constitutes good teaching. “It’s a Holy Grail,” said Eric Hanushek, a professor at the University of Rochester who studies education. “There are many different ways to teach, many different ways to make a good teacher. All we know is: Some people are better at teaching than others.” Last year, a report on teacher quality by the National Center for Education Statistics noted that “[t]eacher quality is a complex phenomenon, and there is little consensus on what it is or how to measure it.” Indeed, good teaching is a complicated issue precisely because it is both elusive and obvious at the same time. It seems to defy definition—how would one, for example, characterize it without falling prey either to the Scylla of the all-too-narrow or the Charybdis of the all-too-vague?—but it also responds to the “I-know-it-when-I-see-it” principle, as spending fourth period with Matthews or Avezzie will make clear. Yet if this is the case, then the key to finding good teachers for American schools may not lie in trying to agree on what good teaching is or attempting to measure it, but in actually watching teachers in action. The key may be to evaluate their performance in real classrooms, in authentic settings, in front of students. We do know good teaching when we see it.

Most efforts to date in evaluating teacher quality have focused on teacher preparation and qualifications for licensing. It’s easy to measure, and easy to certify, based on grade-point averages, a passing score on a standardized teacher exam that tested for basic literacy and math skills, and “seat time”—the number of courses taken in curriculum development or child psychology from a state-approved teacher-training program. All these criteria for licensing, however, have generally only had the limited goal of weeding out bad teachers, never of creaming off good ones—cutoff scores were traditionally kept very low and deliberately so. Test results had “little, if any, power to predict how well people perform as teachers,” a 1987 article in Review of Research in Education noted, mainly because they did not reward creativity, enthusiasm, flexibility or passion—crucial traits of Matthews and Avezzie.

“[In most states], you’ve got to take a test of literacy and subject matter,” said Jerome Murphy, dean of the education faculty at Harvard. “Well, a test of literacy is basically, can you read? And a subject test is just about the basics. But even if you go over that hurdle, you can still be a horrible teacher.” But when political pressure turned on teacher quality, many states responded by simply raising the bar on these criteria by requiring more degrees or higher scores. A New York State Regents’ task force recommended in a 1998 report that new teachers should be required to earn a master’s degree before being allowed to teach, and such states as Virginia and North Carolina raised cutoff scores on PRAXIS, an Educational Testing Service exam for aspiring teachers. These reforms quickly came under attack—a New York Post editorial charged the board of regents with sacrificing students on the “altar of credentialism.” It was akin to a struggling basketball franchise raising its minimum height requirement in the hopes of getting better players.

Not only did requiring more degrees and better grades fail to screen out bad teachers, but they discouraged potentially good teachers without traditional education degrees from applying. “If Indianapolis Symphony Orchestra Director Raymond Leppard wanted to teach music at North Central High School,” a 1998 editorial in the Indianapolis News argued, “he would be barred unless he became a licensed educator.” The pool of potential teachers shrunk, and most states soon enacted emergency certification, opening up a back door that still let in unqualified teachers to fill spot shortages. “A growing number of school districts are throwing a warm body into a classroom, closing the door, and hoping for the best,” said Education Secretary Richard Riley. In the end, more credentialist hoops failed to capture what matters most to good teaching: the actual ability to convey knowledge in the classroom. What, states were asking, could they do to move away from teacher preparation—coursework and test scores—and instead focus on teacher practices—in-class teaching? What could they do to look the horses in the mouth and figure out who has what it takes to be a good teacher? And then they came up with a radical solution: You watch them teach.

“Good teaching is one-fourth preparation and three-fourths theatre,” reflected author Gail Godwin. Maybe. The Thomas B. Fordham Foundation cited a study that showed that “it was the verbal ability of teachers—not the degrees they had earned or the experience that they possessed—that had the strongest effect on student learning,” and this is what schools should screen for when selecting teachers. Many states have focused on credentials (preparation) at the cost of in-class performance (theatre), but the trend seems now to be reversing. Consider the following changes, all of which are recent, and all of which focus on in-class performance.

This spring, teachers in California must pass the statewide Teacher Performance Assessment which will oblige them to submit videotapes of their lessons. “That part of it is having a very dramatic impact,” said Dennis Evans, director of credentialing programs at the University of California, Irvine. “Every prospective teacher in California going for a credential will have to convince an observer that they’re competent and can perform well in front of students.”

Wisconsin, after a five-year effort, has instituted licensing requirements which also take actual classroom performance into account. Previously, any graduate from an education program with a 2.75-plus grade-point average and no criminal record could teach. But now, teacher-candidates must first pass a test in reading, writing, mathematics, and more importantly, also demonstrate speaking and listening skills. “We need to see the ability for teachers to act in a verbal way,” said Peter Burke of the Department of Public Instruction. “We are going to videotape teachers.” Teachers will assemble an extensive portfolio that includes documentation of all the main tasks of teaching ranging from rating student work to lesson preparation. Wisconsin is part of the Interstate New Teacher Assessment and Support Consortium (INTASC), an independent coalition of state education agencies, higher education institutions and national education organizations founded in 1987 partly in response to the scathing “A Nation at Risk” report on teacher preparation four years prior. Over the years, INTASC has worked with some 34 states to define and codify good teaching by developing performance-based standards.

In Kansas, the state board of education recently supported dispensing with the current practice under which teachers receive a professional teaching license simply by passing a basic skills and subject test and enduring three years in the public schools. Instead, a new plan would require teachers to complete a teacher-preparation program and only then do they receive a two-year provisional license. During this time, under the guidance of a veteran teacher, rookie teachers must complete an extensive portfolio similar to Wisconsin’s recording their teaching performance.

Connecticut, where Matthews and Avezzie teach, has pioneered performance-based assessment since 1986 and is a good example of the current changing trend. In the mid-1980s, the state started requiring rookie teachers to complete a portfolio to be submitted to the state for evaluation during their second year of teaching. The requirement centers around performance in authentic teaching tasks, not the hypothetical scenarios posed in exams: Teachers must keep daily lesson logs over a period of seven to 10 days, videotape two lesson segments, and include and analyze examples of student work. They must document the planning, the teaching, and the evaluation of student learning. They are also required to reflect on, and analyze, their own teaching. The portfolio’s questions include, “What do you want your students to know and be able to do? Why is this important?” and “What did you learn from this unit about your students as learners and about yourself as a teacher? Overall, what changes would you make in your planning, instruction and/or assessment the next time you teach this unit? Why?”

But perhaps the most representative feature of the new performance-based paradigm is videotaping, currently practiced in Connecticut and New York with many other states such as Indiana, North Carolina, and Arizona to follow. “We want a lens into their classroom,” said Katie Fisk of the Connecticut State Board of Education, who added that trained teachers will watch the tapes and judge, according to ability to organize and implement lessons, understanding of multiple approaches to learning, motivation, and active engagement in learning. “The tapes are a very rich performance sample,” said Edith Hunsberger of the New York State Department of Education, because they highlight performance, delivery, and passion of teachers in the classroom. “The good ones are exciting.” (Incidentally, videotaping has a positive side effect in encouraging self-reflexive evaluation by teachers themselves. “I found that I never smiled, so you think to yourself, I better make sure I smile,” said Caroline Bitterwolf, an elementary school teacher in Moscow, Idaho. “You ask yourself, am I favoring one group over another, do I circulate around the room, can I be heard in the back? You think about things you don’t usually think about.”) Connecticut evaluates the entire portfolio, grading teachers on a scale of one to four. A two or better qualifies for a provisional license; those scoring below par must resubmit the portfolio in their third year of teaching. A second failure bars that teacher from continuing in the public schools. In this sense, the initial licensing acts as an extended audition that gives schools the opportunity to see a teacher in action in an authentic setting before deciding to hire.

This, coupled with assessment based on performance in authentic situations, is the shift to provisional initial licensing. It gives the hiring process of teachers the feel of a trial period, an extended audition, during which beginning teachers are given the chance to succeed or to fail in an authentic classroom setting. The metaphor of the initial license as an extended audition might seem superficial at first—surely academic preparation must always outweigh empty enthusiasm—but educators agree that passion and dedication indicates the presence of academic preparation. “You can have academic mastery with no passion, and that is not effective,” said Patricia Graham, education professor at Harvard University in a roundtable discussion in 1999. “[But] it is extremely rare to have passion for an academic subject and not have a degree of mastery.”

The provisional license allows for a trial period but also reflects the recognition that teaching is not entirely sink-or-swim; teachers do need some time to adjust. Teachers fresh out of school cannot, after all, be expected to perform like veteran teachers, and programs have recognized this as well. Unlike other professions, such as law or medicine that gradually induct their young by giving them easier assignments at first, teachers are often immediately given the full responsibility of a veteran teacher, and often even more difficult classes as veteran teachers opt for honors classes that are easier to manage. “Where does the belief come from that you can pick someone from a teacher-prep program with basic knowledge and say, now you can teach like an expert?” asked Brian Palmer of the Arizona Department of Education. This is why many school districts, such as Rochester, New York, are giving rookie teachers a mentor to help them adjust to the school and also qualify for a license allowing them to continue teaching.

“People always wonder if Rochester is a country club or a city,” said Carl O’Connell, coordinator for the mentoring program there. “Trust me, it’s a city. It’s poor.” Thirty-six percent of the population is officially poor, but the school system is a different story. While Recruiting New Teachers, a national non-profit educational research organization in Massachusetts, has found that, in urban districts, new teachers leave at a rate of 50 percent in the first three years, Rochester has retained close to 90 percent of its teachers, which is remarkable considering that, between 1987, when the program was started, and 1996 the city became poor: The percentage of children receiving subsidized lunches increased from 44 percent to 80 percent, yet the dropout rate declined by half. And although test scores have declined slightly, Rochester students perform almost as well and often better than the rest of the state: In 1998, Rochester students outperformed the state on every subject category of the Regents Competency Test. The key to Rochester’s program is its mentoring program that helps teachers before they are evaluated for a professional license. “Let me put it like this,” said Ann Chazan, a ninth-grade history teacher in Rochester. “If it weren’t for this program, those first-year teachers would walk out.” Chazan currently is a mentor and helps the beginning teachers, offering advice and support ranging from classroom management to personal advice, helping them to secure materials such as audio-visual equipment, and guiding them through the paperwork. The mentoring process—really a support mechanism to help teachers adjust before the system decides whether or not they should stay on—has positive side effects as well by changing teaching from a very isolated profession into a more collegial endeavor, and has the power to retain teachers who, according to research, reach their maximum potential after three to five years.

The paradigm shift to watching teachers in the class and to giving them a chance to succeed or fail in authentic teaching situations before granting them a professional license is a way to have standards without overly standardizing. “We used to say, we want you to build a race car with this much rubber, this much steel wiring and aluminum,” said Tom Hansen of the Indiana Professional Standards Board. “Now we say we want the race car to go this fast and it has to be competitive. We’re not going to tell you what to put in it, we’re not going to hamstring you. You’ll get a better car under the second method.”

To be sure, this performance-based paradigm with its focus on authenticity is not without its flaws. First, it is very labor-intensive to apply, making mentoring programs costly. When veteran teachers spend time outside their own classes advising other teachers, they cannot teach in the classroom. Annual costs for mentoring can run as high as $13 billion, according to Richard Rothstein, an education writer for The American Prospect. States have also been generally reluctant to fund such programs: Although 28 states mandate or encourage such programs, only 10 provide full or partial funding. Performance assessment is similarly expensive. “It takes 10 times as long to score a videotape as it takes to score an essay,” said Hunsberger.

To save labor, schools might simply decentralize the hiring of teachers, putting principals in charge assessing and recommending them for a professional license after a year-long trial period of performance. “We think deregulating the labor market and increasing accountability for administrators is just a more effective approach,” said Dale Ballou, professor at the University of Massachusetts at Amherst, to Education Week. Raising the bar by requiring acceptable in-class performance may reduce the supply of teachers a bit—not everybody, even if they have successfully completed a state-approved education curriculum and passed the entrance exams, will be able to trade their provisional license in for a professional license. This might be a concern in the face of a potential teacher shortage: This year, the U.S. Department of Education predicted a need of 2.2 million new teachers over the next decade, and even though interest in teaching has gone up in the last several years (10 percent of college freshmen are now considering teaching), the shortage is still dire. One-fourth of teachers are over age 50 and will retire just as student enrollment is projected to climb.

But focusing on authentic teacher tasks and in-class performance rather than credentials also has the potential to increase the pool of applicants who come from non-traditional fields. Recently, Education Secretary Richard Riley called for a plan that would provide for alternative certification that relies on exactly performance-based assessment and provisional licensing. Recent college graduates or retired military workers would receive a one-year initial license after passing a basic skills test and having their performance assessed. During this time, the school has the authority to terminate these teachers if they are not deemed adequate. In essence, school administrators can watch if the new teacher has what it takes to succeed in the classroom, and base hiring on such an assessment. After three years, these teachers can apply for a renewable professional license that relies heavily on classroom performance. The program is extremely promising, not only because precedent (the 1994 Troops-to-Teachers program that lets military personnel teach classes) showed participants were much more diverse and willing to teach in high-poverty areas, but also because it promises to increase the potential supply of teachers.

Ultimately, you need to be able to perform if you’re going to teach, and teachers should be evaluated by what actually happens in the classroom. It is this mindset that has the greatest potential of finding and letting in the Jaime Escalantes, the Mr. Chipps, and the Mr. Hollands into America’s schools.

Alexander Nguyen is a writing fellow at The American Prospect