Method

How this works

The study exists to place a person on a two-axis map of how they think about machine minds — what they take these systems to be, and what they think should happen to the pace. This page explains what is measured, how a position is reached, and what is done with what you give us.

What is measured

Two axes, scored independently so that a strong view on one never moves your position on the other.

• The nature axis (horizontal). Person at −100, Tool at +100. Whether these systems are beings whose interests count, or instruments built to be used.
• The pace axis (vertical). Acceleration at −100, Safety at +100. How much capability you would hold back, or pace, against the risk.

Alongside those, three further measures are reported without moving your point: how settled your view is, whether it is the kind of position evidence could change, and how you would act when the two aims conflict. Two more statements test whether the two-axis frame fits you at all. Separately, two integrity statements and the pattern of your answers are used to identify responses that were not given care — these can flag a response for review but never change a score, and they are never shown to anyone.

How a position is reached

1. Every statement is answered on the same balanced five-point scale: strongly disagree, disagree, neither, agree, strongly agree.
2. Each answer becomes a value from −2 to +2 and is oriented to its axis, so that a statement phrased one way and a statement phrased the opposite way both point at the same end.
3. Your score on an axis is the total divided by the strongest position you could have taken across the statements you answered, then shown as −100 to +100.
4. Your quadrant follows the signs of the two scores.

Two details worth knowing. Statements are presented in a different order for everyone, so that the sequence itself cannot shape the answers, and the order each person saw is recorded with their response so that this can be checked. And the statements on each axis are mixed in their direction, which means that agreeing with everything lands at the centre rather than at an extreme — the honest result when a set of answers carries no position.

What is stored

• Your answers to the statements.
• The written answer and label, if you chose to give them. The written answer appears publicly on the results page.
• The optional profile answers, if you chose to give them. These are used for analysis only and are never shown next to your answer.
• The order the statements were shown in, and the time the survey took, which are used to check the quality of the response.
• A timestamp.

Not stored: IP addresses, user agents, cookies, accounts or email addresses. There is no analytics script on any page. A rate limiter holds an IP in memory for fifteen minutes to stop automated submissions, and that value is never written to disk.

One thing is kept in your browser, and you should know about it. A random marker is saved in this browser's local storage so that it can tell you that you have already answered rather than silently counting a second response. It identifies a browser profile, not a person: it carries no fingerprint, it is not a cookie, and it is not sent anywhere except with your answer. Clearing your browser storage removes it, and private browsing never has it. If it is absent, the survey simply lets you answer.

Written answers are published immediately and moderated automatically: language that would make the page unreadable is masked, and the original is kept out of public view for review. Anything you would not want quoted, do not write.

What this can and cannot tell you

• It is a self-selected sample. People arrive here because a link reached them. The results describe the people who answered, and not any wider population.
• Positions are relative. A score says where you sit among these statements, not how strongly you feel about the subject in general.
• Statements carry assumptions. Every wording choice nudges. The items are balanced and reversed where they can be, which limits the effect without removing it.
• One person can still answer more than once. The browser marker above stops the ordinary case — the same person returning to the same browser and answering again — and it is deliberately weak. Incognito, a cleared cache, a different browser or a different device all defeat it, and none of those can be detected without the tracking this study refuses to do. Responding twice is possible; it is just no longer effortless.
• Not a validated psychological instrument. It is a structured study designed to be defensible and transparent, not a clinical or academic scale with published norms.

Questions about the method, or a correction to it, are welcome. This page is the claim; if any part of it is wrong, it should be fixed.