After a student errs this time, can learning continue? Chapter 4 examined feedback within the classroom, and Chapter 10 traced how evaluation reshapes opportunity. This chapter asks a further question: when a single error enters a file, a tracking decision, or a qualification decision, how can evidence obtained later gain a place.
At the same school, Class 3 piloted a growth portfolio in which every error and every correction was preserved, yet subsequent decisions drew only on the accumulated errors; Class 7 abolished public scores while also requiring teachers to offer encouragement in place of pointing out errors. These are two specific arrangements and do not represent the general form of process-oriented evaluation or score-free evaluation. To compare them, we need to separate what is recorded, how judgment is made, and how feedback is given.
What a Single Error Can Provide
An error can reveal a step not yet understood, but it can also arise from misreading, carelessness, or the design of the task itself. Whether it advances learning depends on factors such as whether the feedback is intelligible, whether the student has the conditions to work with it, and whether the task is appropriate. Failure does not automatically generate competence, and pointing out an error does not guarantee that the student corrects it at once.
Here, accommodating failure means preserving meaningful opportunities for subsequent learning; it does not mean asserting that every failure is beneficial or costless. An error in safety training may require immediately halting the operation, while a deviation in ordinary practice may suit staged treatment. How to correct depends on the risk profile of the task and the situation of the learner, and cannot be settled uniformly by the single injunction to get it right at once.
A single failing result should likewise not automatically become a judgment about the whole person. If a test can only show that a particular skill has not yet met the standard for the current period, it lacks sufficient grounds to directly prove that the student is lazy, lacking in potential, or forever unsuited to a field. Limiting the scope of interpretation protects both the credibility of the evaluation and the student's place to continue learning.
The Two Arrangements Omit Different Things
Class 3 kept records of corrections, yet at tracking time it consulted only the accumulated errors. Even if students later mastered the material in question, their new performance could hardly influence the decision. Within this setup, the problem is not that the records are too detailed but that the uses of evidence are asymmetrical: failure affects qualification, while correction has no corresponding point of entry.
Suppose some students respond by submitting only the steps they are confident about, while others ask parents to tidy up the corrections on their behalf; the reduction of errors in the portfolio then cannot directly indicate an increase in understanding. This is a behavioral hypothesis that needs to be tested: to make such a judgment in a real school, one would have to compare assignments, feedback, and students' actual performance, not infer motives from the name of the arrangement.
The problem in Class 7, by contrast, is that withholding public scores was bundled with the cessation of error correction. If students do not know where a calculation went wrong, and no other diagnosis and feedback are available, they may carry the difficulty into the next task. What is missing here is usable feedback, not the number itself. Even with the hundred-point scale retained, a total score without explanation could produce the same problem.
Conversely, we can imagine a non-numerical arrangement: teachers review the work against publicly stated task requirements, indicate what has been accomplished and what still needs improvement, retain the materials from before and after revision, and allow another teacher to review the judgment. It issues no total score, yet it still has standards, evidence, and a path of correction. This counterexample suffices to refute the claim that without scores only impressions remain; it need not first prove itself suitable for every course.
Scores, Standards, and Evidence Cannot Substitute for One Another
A score is one way of expressing an outcome, standards state the basis on which judgment is made, evidence is the material left by actual performance, feedback helps the learner see the next step, and review checks whether the original judgment holds. The five are related, yet no one of them can replace all the rest.
A test with precise numbers may cover a limited range or be affected by scoring errors; a narrative comment, in turn, may correspond accurately to specific steps in the work. Language can be checked: which performances a comment cites, whether it conforms to the public requirements, and why different reviewers diverge can all become objects of examination. We should not declare text inherently unauditable, nor treat numbers as inherently objective.
Nor do standards need to be unified into a single scale across all tasks. Basic computation can carry fairly explicit correctness requirements, while creative tasks may call for multidimensional reasons and professional judgment. The existence of room for judgment does not mean arbitrary verdicts are permitted; what should be stated is the range within which reasons must be given, and how learners can submit differing material.
Removing an unsuitable indicator can improve evaluation; removing all intelligible requirements may make discretion harder to examine. For families without outside help, vague in-school feedback may be especially hard to compensate for. But this is a conditional risk, and from it we cannot conclude that a certain kind of reform always benefits middle-resource families and harms low-resource families. Group-level conclusions require specific institutions, participants, and outcome data.
Records Should Serve Explicit Uses
Instructional diagnosis, certification of current-period qualification, and selection across stages are different uses. Whether early records remain meaningful requires stating their connection to the task at hand. Historical results can help in understanding change, yet they should not permanently determine every opportunity merely because preservation is convenient; later evidence, likewise, should not automatically overturn a still-relevant record of risk merely because it appeared later.
Class 3 could retain the original coursework while changing the rules of consultation: show recent performance and corrections in judgments about current skills; annotate the status of records that are missing or disputed; and allow students to state the specific conditions that affected performance. This does not erase the past; it lets users see the range and the variation of the records.
Records also involve who may see them, for what use, and for how long they are kept. The closer the material comes to an individual's difficulties and family information, the more it needs restrictions on use. Making class-level instructional problems public does not require publishing every student's complete materials. Aggregates and individual cases each have their uses, and each has its omissions.
Comparison across periods can show progress or persistent difficulty, and should not be prohibited wholesale. What needs to be avoided is comparing non-comparable tasks, ignoring changes in conditions, or inflating a single discrepancy into a label on the person. A report sent home that states the task, the time, and the direction of improvement is better positioned to support concrete discussion than a permanent number without context; it still cannot guarantee how parents will use the information.
What Support Correction Opportunities Require
"You may try again" requires time, guidance, and bearable cost. If remedial slots are insufficient, or the scheduled times conflict with caregiving, the institutional opportunity may not be usable in practice. Evaluation reform therefore needs to check teacher workload and student resources at the same time, and cannot hand all the added feedback over to individual overtime.
Opportunity also has boundaries. Certain qualifications require demonstrating current competence first, and continued learning cannot by itself immediately authorize high-risk operations; selection with limited slots cannot guarantee that everyone keeps their original place. The conditions of learning support, of reapplication, and of current qualification should each be stated separately, so that a single failure to meet the standard is not written as a permanent absence of exits, nor a second chance as a guarantee of the same outcome.
Teachers are responsible for providing feedback proportionate to the task, and students also need conditions under which to participate in correction. This does not constitute a symmetrical exchange of immediacy: students may need additional support, and teachers may need to adjust their explanations, tasks, or feedback time. Failure to get it right immediately cannot by itself prove that a student refuses to exert effort; persistent encouragement that leaves identifiable difficulties unaddressed cannot either count as fulfilling the responsibility to teach.
Review can handle disputes over scoring or records, but the upholding of the original judgment after review does not mean the channel has failed. What should be examined is whether the reasons correspond to the materials, whether errors have been corrected, and whether the necessary downstream decisions have been updated accordingly. Where conflicts of interest are serious, separation of roles is needed; the concrete configuration must further weigh professional capacity and available resources.
Letting Change Enter Judgment
Returning to Class 3 and Class 7, the directions of revision are not the same. Class 3 needs to let new evidence actually participate in subsequent decisions; Class 7 needs to restore intelligible diagnosis and feedback. Either could use scores or forgo them, so long as the chosen form of expression suits the task and does not replace reasons, materials, and opportunities for correction.
Evaluation cannot remove all difficulties on the student's behalf, nor can any single procedure guarantee that learning continues. But it can reduce the blockages of its own making: not inferring total ability from a single failure, not automatically extending old records to every use, and not treating encouragement as a reason to stop teaching.
When the student returns to the classroom with this assignment still unfinished, what is needed is not to be declared permanently qualified, but to know where the problem lies, how to continue, and how new effort can be seen. A community's stance toward failure is tested in exactly these concrete conditions.