A masters student had an idea. Six months later, it was a published preprint.
Aravind Balaji, Northeastern University, 2026. No paper. No structure. A curiosity about quantum memory and graph neural networks.
Used CRITIQ to find the research question hidden inside the idea.
// RESULT
QEMA-G
Quantum-Enhanced Memory Architecture
for Graph-Based AI Systems
// SUBMITTED TO
ACM Transactions on Quantum Computing
TechRxiv preprint — February 2026
What you'll be able to do by the end of this.
| By the end of this, you will be able to… |
Bloom's Level |
| Distinguish a research topic from a testable research question |
Analyze |
| Use /brainstorm to move from a raw observation to a working hypothesis |
Apply |
| Identify the six questions /brainstorm asks — and why each one matters |
Understand |
| Recognize what a complete /brainstorm output looks like, using a real paper as anchor |
Evaluate |
| Initiate your own /brainstorm session with a genuine curiosity you already have |
Create |
CRITIQ is a peer review tool for serious researchers. Today it's for you.
For experienced researchers
- Full eight-dimension peer review
- Methods, stats, structure, ethics, writing
- Reviewer response drafting
- Journal selection + pre-submission checklist
- Used before high-stakes journal submission
For your first paper
- Start with /brainstorm — a curiosity, not a paper
- Six questions that surface your research question
- Teaches as it goes — explains every step
- Same rigor. Better explanation.
- From zero to a question worth pursuing
// Same tool. Same standard. Different starting point.
The problem isn't writing. It's not knowing what to write about.
"I'm interested in AI."
That's a topic. Not a paper.
- Most first papers fail before a word is typed
- The gap: a vague interest where a testable question should be
- "I'm interested in AI" + six hard questions = a hypothesis worth testing
- /brainstorm is the tool that runs those six questions
The obstacle is not intelligence. It is the missing question that turns an interest into a research problem.
/brainstorm asks six questions. Each one does a specific job.
01
What caught your attention?
Surfaces the raw observation — the thing you noticed before you had a framework for it.
02
What's your gut explanation?
This becomes the working hypothesis — the mechanism you think is operating.
03
What have you read about this?
Locates the gap — the space between what's known and what your question addresses.
04
Who would care about the answer?
Identifies the audience — and therefore what counts as a contribution.
05
What study would you run?
Shapes the method — observational, experimental, theoretical, archival.
06
What result would prove you wrong?
Tests falsifiability — the requirement that makes this science, not advocacy.
Before the example. Your turn first.
PAUSE.
What's a topic you're vaguely interested in? Write one sentence. Don't worry if it's not a research question yet. That's exactly the point.
// CHECK FOR UNDERSTANDING
Aravind noticed something. That's where every paper starts.
"Graph Neural Networks face critical memory bottlenecks when scaling. I wondered if quantum memory could change that."
— the observation that became QEMA-G
- GNNs choke on large graphs — memory and compute both break at scale
- Gut read: quantum memory primitives might change the scaling relationship
- Not a hypothesis yet — a hunch about a mechanism
- /brainstorm question 1 surfaced it: "What caught your attention?"
- Most students are never asked that question in academic terms
By question 2, a hunch becomes a testable claim.
Before /brainstorm
"I'm interested in quantum memory for GNNs."
A topic. Cannot be tested. Cannot fail.
After question 2
"QRAM-backed graph store could reduce retrieval depth from O(N) to O(log N) — testable against a classical baseline."
A mechanism. Can be tested. Can fail.
// Same student. Same idea. Two minutes later, it's a hypothesis. That's the tool working.
Who cares, and what don't they know yet? That's the paper.
The Gap — Question 3
GNN scaling is a known problem — well-studied
Quantum integration with GNNs is theoretical — not yet combined
No existing framework provides hardware targets
The space between these bodies of work is the paper
The Audience — Question 4
Quantum hardware researchers who need actionable engineering specs
GNN researchers exploring scaling solutions
ACM Transactions on Quantum Computing readership
Knowing the audience determined what counted as a contribution
// "Actionable engineering targets for the quantum hardware community" — that language came from question 4.
What result would prove you wrong? That's not a trap. That's science.
"We acknowledge that fault-tolerant QRAM remains experimentally immature."
This is not a weakness. This is the falsifiability that makes the paper credible.
- Question 6: "What result would convince you your explanation is wrong?"
- Aravind's answer shaped the realistic regime analysis — NISQ-era noise might eliminate any quantum advantage
- QEMA-G explicitly acknowledges the disadvantage regime — that's the falsifiability built in
- A paper that can't fail isn't a paper. It's advocacy.
CRITIQ will not accept a hypothesis that can only be confirmed. The acknowledged disadvantage regime is what got this paper submitted to ACM.
At the end of /brainstorm, you have five things you didn't have before.
01
A research question
Specific, answerable, scoped. Not a topic — a question with a knowable answer.
02
A working hypothesis
Testable and falsifiable. A mechanism you believe is operating — that could be wrong.
03
A gap statement
What's known, what's not, and why your question sits in the space between them.
04
A rough study design
The simplest viable approach to test the hypothesis — not the ideal, the feasible.
05
Concepts to learn before drafting
Type /learn [term] for any concept you don't recognize yet. CRITIQ will explain it before you continue.
Back to your topic. Which output would be hardest?
PAUSE.
Look at the topic sentence you wrote in the first Pause. Which of the five /brainstorm outputs would be hardest for you to produce right now — the research question, the hypothesis, the gap, the study design, or the concepts to learn?
// CHECK FOR UNDERSTANDING
Open CRITIQ. Type /brainstorm. Start with what you already noticed.
/brainstorm
Answer six questions. Get a research question, a hypothesis, a gap, a study design, and a reading list.
- You don't need a polished idea — you need an observation or a curiosity
- Answer the six questions honestly, in plain language — no academic framing needed
- CRITIQ teaches as it goes — each answer gets a one-sentence explanation of why it matters
- If your first hypothesis is unmeasurable, CRITIQ will tell you — and show you how to fix it
- Aravind started here. The preprint is on TechRxiv. The journal submission is in review.
The paper was always there. You just needed the right questions.
The paper was always there.
// TESTABLE
// FALSIFIABLE
// YOURS
Start with /brainstorm. When your question is locked, /outline builds the IMRaD structure. When you have a draft, /review applies the same standard used for journal submission.
// ARAVIND'S PAPER
doi.org/10.36227/techrxiv.177162482.24655380
CRITIQ — from curiosity to submission // Nik Bear Brown