angushushu/shu-academic-writing-skill / core/problem-framing
problem-framing
VerifiedFrame the paper as a problem in the reader's existing understanding rather than a report of work done. Use before drafting anything, and re-read the opening against it before delivery. It supplies the operational machinery behind move 1 of `core/reader-layer/SKILL.md`.
Problem framing
A paper that reports work is read by the author's advisor. A paper that changes what the reader believes is read by the field. The difference is not prose and not coverage: it is whether the first two hundred words establish that something the reader currently relies on no longer works.
1. Name the problem, not the topic
Write two sentences before drafting anything:
- What the reader currently believes or does. Not "the literature has studied X" -- the actual operative assumption. "Investigators treat a rising network measure as evidence that viewers track narrative structure, and a dozen published findings rest on that reading."
- Why it is now inadequate. A claim the reader would resist losing, with the reason it fails. "Any quantity that accumulates as a story unfolds rises with the clock, so every one of those findings is equally consistent with viewers tracking nothing but elapsed time."
If the second sentence could be replaced by "but nobody has studied Y", you have a topic, not a problem. Keep working.
2. Write into the reader's existing framework
The reader arrives with a code -- a working model of the field. You are not filling a hole in it; you are showing which part of it must change. So say which part, and make the contradiction explicit:
The standard account holds that A. If A were right, B would follow. It does not: C.
Three sentences, and the reader now has a stake in your result before you have reported one. Without them, the reader has to construct the stake themselves, and readers do not do that.
3. Make the reader's work small
Readers are busy, self-interested, and reading for their own purposes. They will not infer why your finding matters. So:
- The value goes in the first two sentences of the abstract and of every section, not at the end of a paragraph.
- A section heading states the finding: "Much of that association is the clock", not "Results".
- Never write "this is important because" and then supply the reason three sentences later. The reason is the sentence.
4. Value is instability, not coverage
Nobody is rewarded for reading a complete survey. They are rewarded for learning that something they relied on is unstable. So the literature review is not a survey of who studied what: it is the argument that the reader's current position has a crack in it, and here it is.
Every paragraph of the introduction should raise or tighten that instability. A paragraph that only establishes that you read widely is doing no work and should be cut.
5. The self is not the subject
"We believe", "I argue", "our view is" put the author between the reader and the problem. The problem is the subject of every sentence in the opening:
- Weak: "We argue that existing measures are confounded."
- Right: "Existing measures are confounded, and the confound is not a technicality."
6. Significance is a claim about the reader's next move
The last paragraph is not a summary of the paper. It states what a reader who accepts your result can now do, or must now stop doing. "We found X in three films" is a summary. "Anyone who treats an accumulating network measure as evidence of tracking must first show it diverges from the clock" is a claim about the reader's future behaviour, and it is the sentence they carry away.
The two-minute check
Read only the first two sentences of the abstract, then the first sentence of each paragraph in the introduction. If those sentences do not, in order, (a) name what the reader believes, (b) show why it fails, (c) say what you measured, and (d) say what follows for the reader -- rewrite the opening before touching anything else.
What this module is worth
Measured end to end against the same package without it: 6-6 on judged quality (two pairs, two framings, four judges), and compliance 12/12 against 9/12 and 11/12. So it is worth its lines for the mechanical axis and not worth claiming a reader-visible gain it does not have.
The reason is that move 1 of core/reader-layer/SKILL.md -- open on a stake, not a gap -- already
carries the load. Everything here is the operational detail behind that one instruction, and a
competent reviser performs most of it once the opening has been forced onto a stake.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |