Joseph Conrad's The Secret Agent is a difficult novel to assess because its plot centers on a botched bombing that the narrator treats with cold irony instead of suspense. Students who write about it often summarize the attempt on the Greenwich Observatory rather than explain how Conrad shapes the reader's judgment of Verloc, Winnie, and the London underworld around them. A rubric helps separate those two kinds of writing before the first essay is marked. Without one, graders tend to reward fluent plot recaps and quietly penalize students who take riskier interpretive positions.

The best rubrics for this novel are built around what the assignment actually asks students to do, not around generic essay categories. If the prompt asks how irony affects the reader's sympathy for Winnie, then the top row of the rubric should describe an essay that traces irony through specific scenes and explains its effect on sympathy. A middle row might describe an essay that names irony correctly but only gestures at its effect. The lowest row should describe an essay that retells events without making any claim about technique.

Four criteria usually cover what teachers care about in a Conrad essay: a debatable thesis, accurate and well-chosen textual evidence, analysis of narrative technique, and control of sentence-level writing. Weighting these differently signals priorities to students. Many English teachers give analysis the heaviest weight, since The Secret Agent punishes surface readings, and give mechanics a smaller share so that a strong idea buried in imperfect prose still earns credit. Whatever weighting you choose, publish it before students begin drafting.

Write Level Descriptors That Describe Work, Not Quality

Descriptors such as "excellent" or "needs improvement" tell students nothing about what to change. A more useful descriptor says what the writing does, for example "explains how the narrator's detached tone in the Verloc kitchen scene makes the violence feel routine." That kind of language lets a grader point to a specific sentence and say which level it matches. It also lets students self-assess a draft before submitting it, which tends to raise the baseline quality of what arrives.

• A thesis that makes a claim about Conrad's technique, not just the plot
• Quotations that are short, accurate, and followed by explanation
• Analysis of narrator tone, irony, or time shifts rather than events alone
• Paragraphs that build one argument instead of retelling chapters in order
• Prose that is clear enough that the argument is never in doubt

Test the Rubric on Sample Essays First

Before grading a full class set, score three or four sample essays against the rubric and note where your own judgment hesitates. Hesitation usually points to a descriptor that is vague or two criteria that overlap, such as evidence and analysis blending together. Fixing those spots takes a few minutes and prevents hours of second-guessing later. If more than one teacher is grading the same unit, scoring the same samples separately and comparing results reveals disagreements early.

Students also benefit from seeing annotated samples at each level. Showing a strong paragraph about Winnie's marriage of convenience next to a weaker one that only summarizes it makes the abstract language of the rubric concrete. Teachers who do this often report fewer regrade requests, because the standard was visible from the start. The samples do not need to be perfect; they just need to illustrate the difference between description and argument.

Where AI Grading Fits Into a Rubric-Based Workflow

An AI grading tool such as GraideMind works best when it is given the same rubric a teacher would use by hand, including the specific descriptors tied to Conrad's irony and narrative distance. It can apply those criteria to every essay in the same way, which removes the drift that sets in around the twentieth paper on a long grading night. The teacher still decides what the criteria are and reviews the output. The tool simply handles the repetitive first pass so that human attention goes to the essays that need it most.

This division of labor matters for a novel as ambiguous as this one. A student who argues that the narrator is more sympathetic to Stevie than he lets on is making a claim that deserves thoughtful engagement, not a checkbox. Reviewing flagged or borderline essays yourself keeps interpretive risk-taking rewarded, while consistent first-pass scoring keeps the baseline fair. The result is a grading process that is both faster and more defensible when a parent or student asks how a score was reached.

Revise the Rubric After Each Unit

After grading, look at where students clustered. If most essays landed in the middle row for analysis, the descriptor for that row may be too forgiving, or the class may need more instruction on how to move from noticing irony to explaining its effect. Keeping a short note on these patterns makes next year's rubric stronger than this year's. Over several cycles, the rubric becomes a record of what your students find hard about this particular novel.

Treat the rubric as a teaching document and not just a scoring sheet. Walk students through it before they write, return it with the graded essay, and refer back to it during conferences. When the same language appears at every stage, students start using it in their own revisions. That shared vocabulary is often the clearest sign that a rubric is doing its job.