We wanted a quiz. The world already had surveys. Plain versions ask whether you love research, whether you handle failure, whether you tolerate poverty with grace. Then they score your agreement and wish you luck.

Self-report has one known flaw. People answer the question they wish were asked. Ask “Do you love your field?” and the yes arrives before the doubt can speak. The yes is sincere. It is also a costume.

So we changed the target. We did not want your answer. We wanted the behavior around your answer. The result is a questionnaire of 18 questions, four minutes, no login. This is the story of how it works and why it always ends the same way.

We started with a clock

The first change was silent timing. The quiz records when each question appears and when your answer starts. The opening prompt is soft: name three things you love about your life. Some people answer in two seconds. Some people stare. How fast you reach a feeling is a fact about the feeling.

The next prompt is harder: “Why this PhD?” The box counts revisions. You type, you delete, you retype. Churn on your own stated reason is the tell. A stable reason survives contact with a text box.

Then comes the page with the statistics. Median time to a degree runs near six years in many fields. Roughly 40 to 50 percent of students never finish. Stipends sit around 30 to 38 thousand dollars while classmates compound salaries. The quiz measures how long you stay on that page before you continue. The raw number is not the point. The number against your own pace is the point.

You against you

That comparison needed a rule. Absolute timing is a typo contest. Fast typists win. Slow typists lose. Neither fact touches the question we ask. So the quiz never compares you with other people. Every comparison stays inside your own run. How long the money question took against your own average. Never the person beside you. Your numbers, your baseline.

The floor

Even that rule had a hole. Some people start any free text slowly. Their whole run drags. The drag is style, not doubt. So the quiz sets a floor before it measures anything. Two questions open the run: “Where are you sitting right now?” and “What is one thing you did this morning?” Anyone can start those instantly. They look like the other prompts. Nothing marks them. Your floor is the time they take.

Every timing claim on the results page is a multiple of your own floor. The debrief names the two floor-setters at the end. The reveal is part of the design. You deserve to know what measured you.

The verdict

Every completed run ends the same way. The screen says: Not ready.

This is not cynicism. It is arithmetic about scope. A quiz can certify non-readiness. A quiz cannot certify readiness. A person who needs a ten-minute webpage to say yes to a six-year commitment is answering a different question.

So the real exit is behavioral. The verdict page offers three buttons. One of them says: Something’s wrong with this result. Click it and a box opens. Write a specific rebuttal. Name what the quiz missed. That act shows the one signal no self-report can fake. You considered your own case and you decline to let a webpage overrule it.

The final card then tells the truth. The verdict was never computed. Readiness is self-issued.

We also measure the pause between the verdict and the click. Some pauses last minutes. We like those people.

The Committee Room

The quiz has a sibling: The Committee Room, a defense simulation. The first version let people pitch their own ideas. We replaced it. Now you study one of five short research proposals. School start times. Walking groups. Green roofs. Small vendor loans. School gardens. All realistic. All secretly doomed.

The shared proposal is the point. When everyone defends the same text, the attacks target the proposal’s actual flaws. Not your nerves. Not your accent. The flaws.

Three examiners wait: Dr. Suhaj, Prof. Kang, Dr. Varga. Each has a name, a manner, and a distinct typing rhythm. The cruelty is deliberate and calibrated. One examiner plays ignorance: “What is ‘engagement inventory’? Again.” One repeats your weakest line with rising contempt. One tracks your drift and calls the collapse: “First answer was better. You’re getting worse at your own proposal.”

Leaving mid-defense is allowed. The exit logs as data. A quick exit is also an answer. A debrief then decodes every trick and ends with decompression.

The plumbing

An appendix for anyone who builds pages inside WordPress. Two filters fought us the whole way.

The first is convert_chars. It turns every ampersand into an HTML entity. Inline JavaScript with a logical AND now holds two entities. The script dies at parse and tells you nothing. The fix: write JavaScript with zero ampersands. A tiny AND(a, b) helper keeps the meaning.

The second is wpautop. It injects paragraph tags and real newlines around anything that looks like a block tag inside your strings. A raw newline inside a string literal is fatal too. The fix: write every tag-opening angle bracket inside strings as \x3c.

One rule outranks both. The stored content and the served page are not the same bytes. Run every check against the served page. Trust nothing else.

One more lesson. Some themes pin a fixed header over your content. Do not edit a theme you cannot touch. A scoped position: static !important override, written in the page content, fixes that.

The standing offer

Three pieces, one question. The quiz takes four minutes and tells you what you already suspect, in your own units. The Committee Room takes longer and leaves marks. The framework article explains the reasoning behind all of it.

The standing offer is simple. Take the quiz. Read the verdict. If the verdict is wrong, tell it why. That reply is the only credential we ever needed from you.