Today’s agreement on teacher evaluation appeals wasn’t a complete loss for the union – just 87 percent of one.

When talks over an evaluation system broke down last year, the conflict centered on who should have the final say on teachers rated ‘ineffective’ under the new evaluation system. The city wanted all appeals to be decided by the chancellor, while the union wanted an independent third party to make the final call.

The subsequent deal that was struck as part of today’s statewide teacher evaluations on paper appears to favor the city. Eighty-seven percent of first-year ineffective rating appeals will still be heard by the chancellor. Second-year ineffective ratings will go straight to a 3020-a termination process that takes into account, but does not depend on, a third-party reviewer’s assessment of a teacher’s quality.

The fact that the union managed to salvage a sliver of its demand – getting the city to agree to refer 13 percent of ratings to a third party – is a small win. Bloomberg and the Department of Education initially walked away from the negotiating table in late December and refused to return until the union gave in to all of their demands.

In an interview today, Mulgrew said he was content with winning the 13 percent figure, which he said was based on the proportion of “unsatisfactory” ratings that were overturned before Bloomberg took office. In a statement, he called the deal “the kind of independent, third party component that the UFT has been seeking.”

This afternoon, city officials offered more details on the agreement, which won’t go into effect until the union and city officially settle on a complete evaluation system. Today, Mulgrew repeatedly indicated that he would not cooperate with the city further on negotiations if they continue to move forward on plans to close and reopen 33 schools.

“I will put every legal remedy on the table and we will do everything in our power,” he told GothamSchools today.

Under the agreement, the union has the option to challenge and refer 13 percent of first-year ineffective ratings to a panel of third-party reviewers. City lawyers said the union would be limited to teachers whose low rating might have stemmed from “harassment” by their principal.

Another set of third-party reviewers, called “validators,” will be assigned to all teachers whose first ineffective ratings are upheld. A “teacher improvement plan” will be created for the teachers and the validators will monitor them over the course of the second school year.

Whether the validators’ assessment of each teacher’s performance matches the principal’s will be crucial if the teacher receives a second low rating. Currently, to the city’s chagrin, the burden of proof in 3020-a termination proceedings is on the city, meaning that lawyers must convince a third party that a teacher is incompetent and should be fired. Under the new agreement, the city will still bear the burden of proof if the validator doesn’t agree with the city’s rating. But if a validator has supported the principal’s low rating, the teacher will have to prove she is not incompetent in order to keep her job — stripping her of a protection the city says has made it nearly impossible to fire weak teachers.

The validator role is modeled after a similar position in New Haven, Conn., where the teacher evaluation system has been cited as a model. UFT Secretary Michael Mendel said tonight that the emergence of the role in negotiations was key to bringing together the agreement.

“The independent validator we believe was a huge win for our members,” Mendel said.

City officials envision that the validator positions would be filled by “master teachers” and experienced evaluators who would be hired as vendors working with the Department of Education. Although city officials said they would like to work with the union to pick the vendors jointly, they added that the UFT would not have the final say. That decision would be made by the State Education Department.

“It’s the only thing we can do to ensure fairness,” Mulgrew said of the need for the independent evaluators.