One idea can carry gains, costs and questions that have no answer yet. PMI splits them into Plus, Minus and Interesting. This generator is for assessing one clear change, not for drawing a conclusion about your life from a plus-minus total. The output keeps the evidence, the check and the owner behind each reason, and produces a three-column diagram for discussion plus a file you can keep editing.

The optional 1–5 weights are only there to sort out how much you personally care. The tool adds the Plus weights and the Minus weights and shows P − M; Interesting never enters the net figure. A bottom-line condition is entered separately and no total can offset it.

1. 01Write the change and its boundary, for example “trial one day of remote collaboration per week for six weeks, without affecting on-site service”.
2. 02Write the plus factors first, then the minus factors seriously; anything whose direction cannot yet be judged goes into Interesting rather than being forced into a neutral score.
3. 03Enter a subjective weight and the evidence for each P and M. Merge similar reasons so that one benefit is not split into three and counted repeatedly.
4. 04Give each open question a next check or small experiment and name an owner. Where conditions allow, prefer a reversible trial over adding more arguments indefinitely.
5. 05Review the three-column diagram and the weight summary, and export an image for communication, Markdown for review and JSON for restoration. CSV only outputs the one table you selected.

Edward de Bono's thinking training emphasises deliberately changing the direction of attention, and PMI belongs to that structured way of looking. The point is not to prove a first impression but to set aside time for the favourable, the unfavourable, and the interesting or still-to-be-explored. His books can serve as method background, but this tool does not claim to implement a validated psychometric scale. Author's book

An earlier example of weighing pros and cons is Franklin's 1772 letter to Priestley: reasons thought of at different times are laid out together, their weight compared, while admitting that reasons cannot be measured as precisely as algebraic quantities. That was a two-column weighing, not the later PMI third column, and it is not the source of this tool's 1–5 scores. Letter text

Compared with a decision matrix, PMI does not require several options to share one set of criteria, so the PMI net figures of different options cannot be compared directly. Compared with a risk register, a Minus is not necessarily a risk that has not happened yet but can be a certain cost, while an Interesting can be an opportunity, a problem or a condition worth experimenting with. A table makes missing information visible, but it cannot guarantee that the reasons are complete, true or independent of one another.

In the “one remote day per week pilot” example, less commuting is weighted 4, easier focus on tasks that do not need on-site equipment is weighted 2, and weaker immediate support for new colleagues is weighted 4, so the current net weight is 6 − 4 = 2. The open question “which collaboration has to be synchronous?” takes no part in the addition; the next step is to sample meetings and documents to confirm whether the remote day really improves collaboration.

In the “evening course enrolment” example, systematic feedback is weighted 5, the cost of two evenings is weighted 4, and the squeeze on the budget is weighted 3, for a current net weight of −2. That figure cannot by itself decide against enrolling; confirm the full cost and the refund conditions first, and consider whether one trial class is enough to test the interest. Necessary rest and family arrangements are separate bottom lines.

Plus: systematic feedback and practice, weight 5, source: course syllabus
Minus: two evenings a week, weight 4, source: course timetable
Minus: squeeze on the budget, weight 3, next step: confirm the full cost
Interesting: could I try it first? next step: attend a short workshop

• The weights are ordinal subjective judgements, not currency, probability or benefit units.

• The tool adds them up to help you organise, and that does not prove the numbers support strict utility arithmetic.

• A blank weight is not silently filled with a neutral value; it is flagged and excluded.

• Repeated wording raises a prompt, but near-synonym repetition and hidden bias still need a human check.

• Up to 80 reasons.

• The image is easy to read but cannot restore the editable structure; JSON keeps every field.

• Weights from different people cannot be merged directly into a group consensus, and discussing the disagreement is usually more valuable than averaging the scores.

Do not decide the conclusion first and then adjust the weights. Nor should an unacceptable condition be ignored because the net score is positive. Rewrite “a bit worried” as an observable question, such as “does the average wait for a new colleague's question get longer during the pilot?”, and state the observation period, how it will be recorded and who will check it.

Interesting is not a bin: every key unknown should have a checking action, or an explicit statement that it cannot be verified for now. Something that cannot be verified can still matter.

• Suited to course choices, trials of a way of working, habit changes, event plans and reversible personal plans.
• Use a decision matrix when several mutually exclusive options need consistent criteria; use a decision tree when complex uncertain branches need step-by-step decisions.
• Safety, medical and legal matters cannot be decided on pros-and-cons weights alone.

No account is needed; input, totals and export happen in the browser. Autosave may be affected by private mode or browser cleanup, so keep a JSON backup. The fragment of a share URL contains the whole document unencrypted and can be read by whoever holds it; above 8 KB, switch to a file. Do not enter sensitive personal history, customer lists or unpublished HR information.

• de Bono, Teach Your Child How to Think: the author's work on thinking training; the publisher page is used to check the edition and does not mean this implementation is certified. （访问日期：2026-09-22）
• Franklin to Priestley, 1772-09-19: primary historical material on weighing pros and cons, and on the limits of quantifying their weight. （访问日期：2026-09-22）