How it works

Seven layers. One at a time. Nothing skipped.

A Houses of Thought run walks the same seven layers every time, based on John Trapasso’s classroom model, derived from the Paul–Elder framework for critical thinking. Six of the seven are checked by a panel of nine independent reviewers, one per standard, each blind to the others, before the run is allowed to move on. Select a layer below for a quick summary, or jump straight to the full write-up further down the page.

01Layer 1 of 7 · Nine-standard review panel

Frame

Pins down exactly what is being decided, defines the terms that matter, and sets the scope, without arguing for any side yet.

Read the full write-up ↓

Why a panel, and why nine standards?

The nine standards are Paul and Elder’s Universal Intellectual Standards, the same nine a critical-thinking classroom would apply by hand. Each one is graded by its own independent reviewer, seeing only its own question, so a strong score on one standard can never paper over a weak one on another. What each standard actually means changes by layer, since a layer can only fairly be graded against what it is actually trying to do. The full glossary, including that per-layer nuance, is below.

A layer that fails a standard doesn’t quietly pass anyway. It loops and redoes the work. Frame, Global Assumptions, Global Evidence, Conclusions, and Implications hold the whole run until they pass, since nothing else backs them up. Perspectives instead drops just the one stance that failed, since the other independent perspectives already provide redundancy. Breadth Scoping is the one layer with no panel at all: a single judgment call about how many perspectives the question deserves, made once.

The seven layers, compared

Layer	Review panel	If it fails a standard
1. Frame	Nine independent standards	Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.
2. Breadth Scoping	None	No panel. A single judgment call, made once.
3. Perspectives	Nine independent standards	Drops the failing stance. The other independent perspectives already provide redundancy, so only the one that failed is dropped.
4. Global Assumptions	Nine independent standards	Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.
5. Global Evidence	Nine independent standards	Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.
6. Conclusions	Nine independent standards	Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.
7. Implications	Nine independent standards	Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

The review panel

Meet the nine standards.

Paul and Elder’s Universal Intellectual Standards. Six of the seven layers get graded against all nine. Open a standard below to see what it specifically means at each of those six layers.

Clarity: Is it understandable, free of vagueness and ambiguity?

Frame

Is the question stated plainly enough that you would instantly know what is being decided?

Perspectives

Does the stance commit to one clear position, instead of hedging about what it even argues?

Global Assumptions

Is each assumption stated as one clear claim, not a vague generality?

Global Evidence

Is it plain what each piece of evidence would actually show?

Conclusions

Is the verdict direct and committed, not hedge-everything mush?

Implications

Is each implication a clear, concrete statement of what follows, not hand-waving?

Accuracy: Is it true, and free of distortion or error?

Frame

Does the framing faithfully capture what was actually asked, without quietly narrowing it?

Perspectives

Are its claims and evidence real and defensible, not overstated or misrepresented?

Global Assumptions

Are these genuinely implicit in the question, rather than misattributed to it?

Global Evidence

Is anything not yet verified honestly flagged as such, rather than stated with false confidence?

Conclusions

Does it represent what the evidence and assumptions actually established, without overstating them?

Implications

Does it actually follow from the stated conclusion, rather than inventing an unrelated consequence?

Precision: Is it specific enough — exact details, not loose approximation?

Frame

Are the key terms defined concretely, instead of in vague generalities?

Perspectives

Is the argument specific to this exact question, not generic pro/con boilerplate?

Global Assumptions

Is each assumption named specifically, not a fuzzy gesture at a theme?

Global Evidence

Is the kind of source needed specific, not a vague "some research shows"?

Conclusions

Is it specific about what it claims, and under what conditions?

Implications

Are who is affected and the timeline named specifically, not generic?

Relevance: Does it actually bear on the question at hand?

Frame

Do the definitions and scope actually bear on answering the question?

Perspectives

Does everything in the stance actually serve its own case, rather than drifting off-topic?

Global Assumptions

Does each one actually matter to answering the question?

Global Evidence

Does each item bear on the question itself, rather than just one stance’s argument?

Conclusions

Does it actually answer the question that was asked?

Implications

Does it actually follow from adopting the conclusion, rather than being a tangent?

Depth: Does it engage the real complexities, not just the surface?

Frame

Does the framing account for how many considerations are genuinely at stake?

Perspectives

Does it argue its strongest real form, and does its self-attack target real weaknesses?

Global Assumptions

Do they surface non-obvious assumptions the conclusion would quietly depend on?

Global Evidence

Does it engage the genuinely contested parts of the question, not just easy surface facts?

Conclusions

Does it engage the real tension in the evidence, rather than a surface "the evidence supports X"?

Implications

Does it explore real downstream effects, not just the most obvious first-order one?

Breadth: Does it consider other relevant viewpoints, not just one?

Frame

Does the scope leave room for genuinely different angles, rather than pre-narrowing toward one?

Perspectives

Does it draw on a genuine range of angles, rather than one point repeated?

Global Assumptions

Do they genuinely cut across perspectives, rather than repeating one stance’s own list?

Global Evidence

Does the evidence span a spread of angles, rather than all pointing one direction?

Conclusions

Does it account for the range of perspectives and evidence gathered, including the ones that cut against it?

Implications

Does it span positive, negative, and uncertain, and different people affected?

Logic: Does the reasoning actually follow — no contradictions, no non-sequiturs?

Frame

Does the framing avoid quietly assuming the answer before the reasoning even starts?

Perspectives

Does the reasoning hold together, and does the counterargument actually land on the real claims?

Global Assumptions

Is the case for why these are cross-cutting actually coherent, not just asserted?

Global Evidence

Does each claim follow reasonably from its source, without an unjustified leap?

Conclusions

Does it genuinely follow from what came before, without overreaching beyond what was established?

Implications

Does it logically follow if the conclusion is adopted, without a non-sequitur?

Significance: Does it focus on what actually matters most, not a minor point?

Frame

Is this question substantial enough to justify the full process?

Perspectives

Does it focus on what would matter most to someone who actually holds this view?

Global Assumptions

Would the answer meaningfully change if one of these turned out false?

Global Evidence

Is this evidence that would actually move the answer, not tangential trivia?

Conclusions

Does it address what is actually at stake, not a minor corollary of it?

Implications

Would this implication actually matter to someone deciding, not a trivial side-effect?

Fairness: Is it even-handed, not self-serving or one-sided?

Frame

Is the language neutral, rather than tilted toward one side?

Perspectives

Is the counterargument a real, honest attack rather than a strawman, and are opposing views represented fairly?

Global Assumptions

Do they scrutinize every side evenly, rather than singling one out?

Global Evidence

Does it represent multiple sides, rather than being cherry-picked toward one?

Conclusions

Does it weigh multiple sides, rather than reading as pre-decided?

Implications

Are the upsides and downsides both explored honestly, rather than stacking the deck one way?

In depth

01Layer 1 of 7

Frame

Pins down exactly what is being decided, defines the terms that matter, and sets the scope, without arguing for any side yet.

Frame is checked twice: once before the rest of the run to confirm the question is actually answerable, and once after framing to confirm nothing essential was missed. Everything downstream, including every perspective and every piece of evidence, inherits whatever gets decided here, so a vague or quietly slanted frame turns the rest of the run into work built on an unstable foundation.

Checked by the nine-standard panel (see Meet the nine standards). Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

02Layer 2 of 7

Breadth Scoping

A single judgment call: how many genuinely distinct, worthwhile perspectives does this question deserve? Made once, with no review panel and no redundancy to fall back on for this step.

Some questions genuinely split two ways; others have four or five defensible angles worth building out. This layer sets that count before a single perspective is drafted, so the run builds exactly as many independent stances as the question supports, with no padding to hit a round number and no flattening a genuinely multi-sided question down to a simple for/against.

No review panel. A single judgment call, made once.

03Layer 3 of 7

Perspectives

Builds several independent, honestly-argued stances, each with its own claims, sub-questions, assumptions, evidence, and a real counterargument against itself.

Building a perspective happens in two rounds: every stance stakes out its position independently first, then each one is fleshed out with sub-questions, assumptions, evidence, and a counterargument against itself. Because each stance is reviewed on its own, one weak perspective doesn’t stall the run. It’s dropped, and the other independent stances carry the weight instead.

Checked by the nine-standard panel (see Meet the nine standards). Drops the failing stance. The other independent perspectives already provide redundancy, so only the one that failed is dropped.

04Layer 4 of 7

Global Assumptions

Surfaces the assumptions sitting underneath the whole question, not just one stance’s.

These are different from the assumptions inside any one perspective. A perspective’s assumptions only need to hold for that stance’s own argument; a global assumption is the kind of premise that would change the shape of every stance at once if it turned out false, the kind that is easy to miss precisely because no single perspective is responsible for naming it.

Checked by the nine-standard panel (see Meet the nine standards). Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

05Layer 5 of 7

Global Evidence

Gathers evidence that bears on the question itself, not just one stance’s case.

Where a perspective gathers evidence to support its own case, this layer gathers whatever bears on the question directly, including evidence that cuts against wherever the run eventually lands. It is the layer most likely to complicate a tidy story.

Checked by the nine-standard panel (see Meet the nine standards). Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

06Layer 6 of 7

Conclusions

Draws the verdict(s) that actually follow from everything that passed review.

A conclusion here isn’t a vote among the perspectives, and it isn’t required to land in the middle. It’s whatever the vetted global assumptions and global evidence actually support, which can mean siding closer to one perspective than another, as long as the reasoning that gets there is shown.

Checked by the nine-standard panel (see Meet the nine standards). Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

07Layer 7 of 7

Implications

Maps what follows if the conclusion is adopted: positive, negative, and uncertain effects, who is affected, and on what timeline.

A conclusion that helps one group and costs another isn’t neutral just because it’s correct. This layer is what makes that visible: it sorts what follows by who bears it and on what timeline, rather than stopping at whether the conclusion itself holds up.

Checked by the nine-standard panel (see Meet the nine standards). Hard-blocks. Loops until it passes; the run cannot continue on a failed version of this layer.

↑ Back to top

Where this comes from

This isn’t a house style invented for an app. It’s a real classroom model, taught by a real teacher, built into software because it worked on paper first.

The seven layers are John Trapasso’s classroom framework; the nine standards each layer is checked against are Richard Paul and Linda Elder’s Universal Intellectual Standards for critical thinking. Houses of Thought is one particular implementation of both, not the other way around.

Read the full story →

See it for yourself

Pick a question you can't crack.

Try it instantly Browse examples

No sign-up needed to try it.