The search for Baltimore's 19th school superintendent has been narrowed to a precious few candidates, and the questions swirling about have a familiar ring:

Should the new superintendent be an "insider," knowledgeable about the system and the people in it, or an "outsider," not familiar with Baltimore but with no preconceived notions? Do we need another educator like Richard C. Hunter, the dismissed incumbent, or should we hire a business manager who may or may not have the credentials of an educator? Should the new superintendent be pledged at the outset to get rid of the "deadwood" at North Avenue school headquarters?

Alas, while all are familiar questions and all should be asked, none is the central one. The determining question that ought to be posed to the handful of finalists is this: Which of you is willing, really willing, to turn the system on its head so that you are answerable to the people in the schools, young and old, and not the other way around?

The question to be asked of the city fathers (and mothers) is this: Knowing that changes in education take decades, not years, to bring about, are you willing to wait, and be patient and cooperative, while the new superintendent plots his or her course over a number of years?

Almost a year ago, I was commissioned by the Fund for Educational Excellence, a Baltimore organization dedicated to improving city schools, to write an interpretive history of the system, going back through three decades and the seven superintendents who served during those years. I was struck by two consistent patterns:

* Baltimore has had five "outsider" and two "insider" superintendents, all of whom reorganized the bureaucracy. The central office -- first at 25th Street, then at North Avenue -- has been regionalized and centralized, divided between elementary and secondary and returned to "K-to-12." Enough "deadwood" has been discarded to keep a beach bonfire burning for a month. The central office staff has been reorganized at least 11 times since the superintendency of Roland N. Patterson in the early 1970s.

The superintendents themselves haven't lasted long in office, although Baltimore's average tenure of 4.3 years over the 30 years is a model of stability. (Among other big cities, the average is under three years.) Superintendents take office under great expectations from politicians and community leaders. The presumption is that they will be "reformers." The reality is that, in education, reform takes much longer than most school chiefs ever have.

* All of the superintendents were reformers. All set out, with the support of their school boards, to "turn the system around." Programs initiated by each superintendent were ignored or wiped out by his or her successor, so that the terrain is littered with the corpses of dozens of initiatives (most of them worthy), big and small in concept and ambition.

Veteran teachers can recite the titles: community schools, Right to Read, mastery learning, DISTAR, the Teacher Corps, Parent-Infant Centers for Education, Project STAR, Involving the Very Young, management by objectives, clustering, Model Schools, the City High School Recognition Program, the Model Early Childhood Learning Centers. The list goes on and on.

Most of these programs died not because they lacked merit but because they lost their funding or because they were imposed from above. When that happens, when teachers and students do not feel a sense of ownership in the program, when they aren't adequately trained to carry it out (educators call it "staff development"), they typically revert to old and comfortable ways.

One teacher called it the "barnacle syndrome. . . . When a new wave of reform comes by, people stick to the rock like barnacles. They filter out what they can't use, and they don't change too much. They wait for the next wave of reform to come along, and then they repeat the process."

The bigger the program and the greater the top-down nature of its implementation, the harder it falls.

Two examples from modern history are Right to Read, a federal program in the early 1970s, and mastery learning, a fad that swept the country in the mid-1980s, at one point said to be the teaching technique for 50 million children. Both faded with time, in large part because they required large-scale testing and weren't accepted in the classroom. The trouble with Right to Read was summed up by the faculty of City College in a letter to Superintendent Patterson 18 years ago: "The major causes of reading weaknesses are generally much deeper and more complex than implied by the administration and cannot be erased by presenting the classroom teacher with a few hours of instruction and the opportunity to administer a frightening battery of tests."

These observations don't apply exclusively to Baltimore. In a book published last week, "The Classroom Crucible: What Really Works, What Doesn't, and Why," researcher Edward Pauly noted that what students learn in the classroom has little to do with central office policy and almost nothing to do with state policy (with all of its student testing) and federal policy.

Mr. Pauly says parents shouldn't put their faith in superintendents, governors or the "education president," George Bush, who on Thursday proposed a set of federal policies, including national testing, designed to create a "new generation of American schools." The real control of education, Mr. Pauly says, is wielded by teachers and students in their classrooms, and if parents aren't happy with what is happening there, they should march to the school and insist that their children be reassigned.

The one "reform" Baltimore hasn't tried in modern times is turning the system upside down and having power flow from those classrooms and those schools to the central office. But "school-based management," as it is called, is much easier said than done, even if a superintendent is genuinely willing to cede power to the schools. Schools only "reflect our communities and their conditions," as the Fund for Educational Excellence says in its introduction to my report, and it's hard enough for inner-city schools to maintain functional parent-teacher organizations, let alone involve teachers and parents in real local governance.

There are a number of other questions. What happens when a student goes from School A, which has chosen one curriculum, to School B, which has chosen another? How does the idea jibe with the state's reform effort, "Schools for Success," which requires extensive testing? (The first statewide "report card" last November showed Baltimore failing in all eight categories, four of which require testing.)

Will North Avenue really give up its power? In Chicago, where school-based management was ordered (from the top, of course) on a system-wide basis in 1988, the central bureaucracy not only dragged its feet; in some cases it sabotaged the process. One evaluator late last year found the bureaucracy in a state of confusion, with many "demoralized people in cubicles."

Baltimore schools took tentative steps toward school-based management last fall with an invitation to schools to participate in a "pilot" program. This effort was put on hold, pending employment of a new school chief. But despite the problems in Chicago and elsewhere, the concept seems well worth pursuing. (Sometimes it's good to begin experimentally, not to try something so complicated in one fell swoop, as Chicago did.)

An earlier Baltimore superintendent endorsed the idea. "One of the greatest needs of public education, especially in our larger cities, is to recapture, in a practical way, the close and intimate relationship between the school and the parents, between the school and the total community," said William Lemmel. "This can best be done, it seems to me, at the principal-teacher-building level. If this relationship is to be effective, it must not be superficial; the community must be a part of the school and the school a part of the community." He said it in 1948.

Mike Bowler is the editor of The Evening Sun's Other Voices page.