---
name: dalio-style-believability-decision
description: Runs one team decision through the decision-making principles Ray Dalio has published on principles.com and in his 2017 TED talk - recognise the case as "another one of those" and reuse or write a principle for it, gather views from the most believable people who disagree, weigh each view by the person's track record on this kind of question and the quality of their cause-and-effect reasoning, compare the weighted view with the plain majority and with the responsible person's own view, record why the responsible person decides as they do, and write the criteria down as a principle for next time. Use when a team is split on a decision, a loud or senior voice is deciding by default, the same kind of decision keeps being argued from scratch, or someone asks for a believability-weighted decision, an idea meritocracy or a decision principle.
---

# Dalio-style believability-weighted decision

These are principles Ray Dalio has described publicly on principles.com and in his 2017 TED talk, "How to build a company where the best ideas win". Not affiliated with or endorsed by Ray Dalio or their companies. The skill takes one decision, weighs the views on it by each person's believability on that question, shows where the weighted view differs from the majority, and writes the criteria down as a principle the team can reuse.

## Ground rules

- Do not attribute quotes or opinions to Ray Dalio beyond the cited sources.
- Do not role-play, imitate, or write in the voice of Ray Dalio; review the user's material against the principles below.
- Believability is about one kind of question, from evidence the user gives (past results on this kind of decision, quality of reasoning). It is not a rating of a person's worth or character. Mark a missing track record "[check: ...]"; never invent one.
- The weights are a tool for the responsible person, not a vote that replaces them (principle 6).
- Transparency here means the reasoning behind this decision is visible to the people it affects. Do not ask for, record or share private, health or personal matters; the sources themselves keep those confidential (principle 8).
- Bridgewater's own tools (the Dot Collector, Baseball Cards) are not available here; this is a written, manual version of the idea, and the output says so.
- Say when the method doesn't fit: a decision with one clearly expert owner and little disagreement doesn't need weighting.

## The principles

1. **Most cases are "another one of those".** Almost every case at hand is a type that has come up before: identify which type it is and apply a well-thought-out principle for that type. Many cases are hybrids, so several principles have to be weighed against each other. Source: principles.com, Life Principle "Use principles."
2. **Write the criteria down while deciding.** Note the criteria you use as you make a decision, write them down as a principle, check them against the outcome, and refine them before the next case of the same type. Source: principles.com, "Use principles." and Work Principle "Create great decision-making machines by thinking through the criteria you are using to make decisions while you are making them."
3. **Who is believable.** Dalio defines believable people as those who have repeatedly and successfully done the thing in question (a strong track record with at least three successes) and who can explain their approach well when probed. Someone with neither is not believable on that question, someone with one is somewhat believable. Source: principles.com, Life Principle "Be clear on whether you are arguing or seeking to understand..." and Work Principle "Remember that believable opinions are most likely to come from people 1) who have successfully accomplished the thing in question at least three times, and 2)...".
4. **Seek believable disagreement and the reasoning behind it.** Find the most believable people who disagree with you and try to understand their reasoning; pay more attention to the reasoning than to the conclusion. A logical theory from someone without the track record can still be stress-tested, and people should say how confident they are. Source: principles.com, Work Principle "Find the most believable people possible who disagree with you and try to understand their reasoning."
5. **Track records build up over time.** Assess believability by capturing people's track records systematically over time rather than starting fresh each day. Source: principles.com, Work Principle "Assess believability by systematically capturing people's track records over time."
6. **A tool for the responsible party, not a replacement.** Believability weighting supplements and challenges the decisions of Responsible Parties rather than overruling them. A responsible person who goes against the believability-weighted view should be able to say exactly what they disagree with, understand the risk of being wrong, and explain their reasoning; if they can't, they should probably go with the weighted view. Once decided, the group gets behind the decision, though it can be revisited later. Source: principles.com, Work Principles "Use believability weighting as a tool rather than a substitute for decision making by Responsible Parties.", "When you're responsible for a decision, compare the believability-weighted decision making of the crowd to what you believe." and "Once a decision is made, everyone should get behind it even though individuals may still disagree."
7. **Neither autocracy nor democracy.** In the TED talk Dalio describes an idea meritocracy, in which neither one leader decides nor every view counts equally, and gives an example of a vote where the majority went one way and the believability-weighted answer was completely different. Source: TED2017 talk, "How to build a company where the best ideas win".
8. **Transparency, as it bears on decisions.** Radical transparency means much more transparency than is typical, not total transparency: it brings problems and how people handle them to the surface. Some things stay confidential, such as private health or personal matters, and exceptions should be rare. Source: principles.com, Work Principles "Be radically transparent." and "Keep exceptions to radical transparency very rare."; TED2017 talk.

## When to use

- A team split on a decision: a hire, a price change, a launch date, a vendor, an architecture choice.
- One senior or loud voice is deciding by default and others stay quiet.
- The same type of decision is argued from scratch every time.
- After a decision, to write down the criteria as a reusable principle.
- Not for decisions about one person's employment or performance (use performance-review-draft or a process with HR), and not where the decision owner is the only one with relevant experience.

## What you need

- The decision as one question, and who is responsible for it.
- The options on the table.
- Who has a view, what the view is, and their reasoning in a sentence or two each.
- For each person, on this type of question only: past decisions of this type and how they turned out, and whether they can explain the cause and effect behind their view. Unknown is fine; it will be marked.
- Any principle the team already wrote for this type of decision.

## Steps

1. **Name the type (principle 1).** Say which kind of decision this is ("another one of those": a pricing change, a senior hire, a build-or-buy). If the team has a principle for this type, apply it first and note where this case differs. If it is a hybrid, list each type it contains.
2. **Collect views with reasons and confidence (principle 4).** One row per person: their option, their reasoning, how confident they are. Add at least one believable person who disagrees with the responsible person's current lean; if none was asked, say who could be.
3. **Assess believability on this question (principles 3 and 5).** For each person: track record on this type (successes and failures, from the input), and quality of explanation (a clear cause-and-effect reason, yes or no). Classify as most believable (both), somewhat believable (one), or not yet believable on this question (neither). Mark unknowns "[check: ...]".
4. **Weigh the views (principle 7).** Show the plain count of views per option, then the count with believability weighting (for example most believable = 3, somewhat = 1, not yet = 0.5; state the weights used and that they are our choice, not the source's). Note where the two differ.
5. **Stress-test the reasoning, not the rank (principle 4).** For the top two options, list the strongest reason against each from the most believable person who holds the other view. Name any logical theory from a less experienced person that is worth testing.
6. **Compare with the responsible person (principle 6).** If the responsible person's view differs from the weighted view: write exactly what they disagree with, the risk if they are wrong, and their reasoning. If they can't, recommend going with the weighted view.
7. **Decide and commit (principle 6).** Record the decision, who decided, who disagreed and why, and when it will be reviewed. Everyone gets behind it until then.
8. **Write the principle (principle 2).** Turn the criteria used into a short principle for this type of decision, and set the date and measure to check it against the outcome.
9. **Share the reasoning (principle 8).** Draft a short note for the people affected: the decision, the views, the weighting, the reasons. Leave out anything private or personal.

## Output format

```markdown
# Decision: [question]
These are principles Ray Dalio has described publicly on principles.com and in his 2017 TED talk. Not affiliated with or endorsed by Ray Dalio or their companies. The weights and this written version are our adaptation.

## Type
[Another one of those: type(s)] - existing principle: [text or none]

## Views
| Person | Option | Reasoning | Confidence | Track record on this type | Explains cause and effect | Believability |
|---|---|---|---|---|---|---|

## Majority vs weighted
| Option | Plain count | Weighted score (weights: [x/y/z]) |
|---|---|---|

## Strongest reasons against
[Per top option, from the most believable person on the other side]

## Responsible person
[Agrees with weighted view / differs: what exactly, risk if wrong, reasoning]

## Decision
[Option] - decided by [name] on [date]. Disagreed: [names, reasons]. Review on [date] against [measure].

## Principle for next time
"When [type of case], [criteria], because [cause and effect]." Check on [date].

## Note to the team
[Short, no private matters]
```

## Example

**Input:** "Small online shop, 9 people. Question: should we raise our standard shipping fee from 500 yen to 700 yen? I'm the owner and I lean no. Aiko (ops, has changed shipping rules twice before, both went fine, can show the carrier cost data) says yes. Ken (marketing, joined last month) says no, customers will leave, no data. Mei (customer service, ran the last fee change at her previous job, sales held) says yes but with free shipping above a threshold. Two others say no without reasons."

**Output (excerpt):**

```markdown
These are principles Ray Dalio has described publicly on principles.com and in his 2017 TED talk. Not affiliated with or endorsed by Ray Dalio or their companies. The weights and this written version are our adaptation.

## Type
Another one of those: a price change on a fee customers see. Existing principle: none.

## Views
| Person | Option | Reasoning | Confidence | Track record on this type | Explains cause and effect | Believability |
|---|---|---|---|---|---|---|
| Aiko | Raise to 700 | Carrier costs rose; data shows the loss per order | High | Two fee changes, both held [check: results] | Yes | Somewhat (fewer than three successes) |
| Mei | Raise, free above a threshold | Threshold keeps basket size up | Medium | One, at a previous job | Yes | Somewhat |
| Ken | Keep 500 | Customers will leave | [check] | None | No data yet | Not yet believable on this question; theory worth testing |
| Owner | Keep 500 | [check: reasoning] | [check] | [check] | [check] | [check] |
| Two others | Keep 500 | None given | - | - | No | Not yet |

## Majority vs weighted
| Option | Plain count | Weighted score (weights: 3/1/0.5) |
|---|---|---|
| Keep 500 | 4 | 0.5 + 0.5 + 0.5 + [owner: check] |
| Raise (either form) | 2 | 1 + 1 = 2 |

The plain majority says keep, 4 to 2. Weighted, raise has 2 against keep's 1.5 plus the owner's own weight, so the result turns on how believable the owner is on fee changes [check: the owner's track record]. If the owner's weight is 0.5 or more, the weighted view is at least a tie, not a clear raise; the evidence to settle it is Aiko's cost data. Because the owner is responsible, step 6 applies either way: name exactly what you disagree with in that data and the risk if you are wrong.
```

## Checks

- [ ] The attribution and not-affiliated lines are in the output, and the weights are labelled our choice.
- [ ] The decision type is named, and any existing principle was applied first.
- [ ] Every view has its reasoning; believability is about this type of question and comes from the input or is marked [check].
- [ ] Plain count and weighted count are both shown.
- [ ] The responsible person still decides, and any override has what, risk and reasoning written out.
- [ ] A principle for next time is written, with a date to check it.
- [ ] Nothing private or personal is in the team note.

## Common mistakes

- Treating seniority or title as believability. The sources tie it to track record and explanation on the question at hand.
- Letting the weighted score decide automatically, which the sources say it should not do.
- Ignoring a less experienced person's logical theory instead of testing it.
- Skipping the written principle, so the next case of the same type starts from scratch again.
- Reading "radical transparency" as sharing everything, including private matters.