M

Question writing and submission guidelines

Welcome to our question writing guide! Here, you'll learn about our best practices for writing and submitting questions, as well as our content rules and guidelines.

We greatly value the contributions of our diverse community of forecasters, question authors, and forum participants, and we hope that these guidelines will promote, enhance, and safeguard a vibrant community forecasting space for many years to come.

Submitted questions are reviewed by a group of volunteer Community Moderators. The Community Moderators team tries its best to approve all the questions that conform to our guidelines and best practices swiftly, typically within a week or two. Questions that are not immediately ready for publication are provided feedback by Community Moderators.

What types of questions are suitable for Metaculus?

Metaculus hosts questions on almost any topic — science, technology, politics, business, law, economics — you name it! That said, our primary focus areas are Science, Technology, Effective Altruism, Artificial Intelligence, Health, Geopolitics, and Far-Future Forecasting (10 years or more in the future).

Currently, we will consider any question that satisfies our guidelines and rules for publication on the platform.

Our guidelines

Writing incisive questions for forecasting can be challenging and requires a keen eye for detail, careful precision, creativity and imagination, and a host of other skills. Fortunately, many of these skills can be cultivated with a bit of practice.

For experienced question writers, we have a briefer checklist to address common issues here. For a comprehensive guide to writing good questions, see our guidelines below.

Specifying precise resolution criteria

Resolution criteria are the backbone of any forecasting question—they spell out how and when a question will resolve. It's therefore key to spell these out clearly.

  1. Aim for tight resolution criteria. The resolution criteria should leave little room for discretion in deciding the resolution. As best you can, try to limit the scope for ex-post quarrels about what really happened, and who was right.
  2. Define your terms. Questions of the sort “will X occur?” often hinge on how X is defined. It is therefore important to spell out your definitions with extra care. Don't worry about being too pedantic here!
  3. Be concrete. Try to specify precisely and in detail which steps should or shouldn't be followed when resolving the question. Examples are helpful for making these instructions concrete.
  4. Use authoritative sources, when possible. Good options are numerical data regularly published by a reliable publicly available source. Note that you should be sure that the sources will be available at the time of resolution, or otherwise you might want to specify alternative sources of information.
  5. Consider and account for edge-cases. Try to imagine scenarios for which the resolution conditions fail to cleanly apply, or cases that are just on the edge of counting towards resolution. If such scenarios or edge-cases are plausible, you should clarify how the question should resolve when such events rear their head.
  6. Consider fall-back criteria. When you have a resolution that should be easy to check assuming all goes well, try to handle also the case where all doesn't go well. What if the data source you specified stops being published? Is there anything else odd that might happen to make the outcome unclear?
  7. Try to account for unknown unknowns. Think about how the resolution criteria behave when something you don't expect happens anyway.
  8. Avoid requiring significant Admin effort to resolve. Questions should not burden Admins with having to do a lot of work to resolve them in the future. Our current policy is to reject any question that would require more than approximately 15 minutes of Admin effort to resolve, unless we judge the question to be important enough to warrant the amount of effort required.
  9. Leave resolution authority to Metaculus. Only Metaculus Admins can resolve questions and responsibility for determining appropriate resolution should rest solely with them. Questions should not give resolution authority to question authors or any other members of the community.

See the FAQ for more details about how questions resolve.

Formatting and details

Metaculus has developed its own style and norms for writing and submitting questions. Heeding these best-practices will be appreciated by Metaculus forecasters and Moderators alike. Questions which don't follow these best-practices may be rejected.

  1. Write in a neutral and objective tone. Questions in a neutral point of view represent topics fairly, proportionately, and, as far as possible, without editorial bias (think Wikipedia-style).
  2. Put effort into your question submissions. Don't expect Moderators to develop your question from scratch. If you have early ideas for questions but have yet to work these out, you can add a comment in the question-suggestions discussion post. Questions which are not nearly ready to publish and which do not follow the guidelines on this page will be rejected.
  3. Keep an eye out for feedback on your questions. After submitting your question, keep up with the feedback provided by Moderators, including requests for revisions.
  4. Explain why your question is interesting or important. The significance or context of a question is not always common knowledge, so it is sometimes important to explain it in the Background section. Quoting predictions from public figures or institutions is a great way to show the significance of a question. Use the Background to offer information that Metaculus users should know, with links to any high-quality articles or resources that would help them get started on their forecasts. Aim for at least 100 words in the Background.
  5. The title and wording of the question should match the resolution conditions. Make sure that the question wording and title do not give a mistaken impression of when and how the question might resolve.
  6. Delegate the boring stuff to the fine-print. In some cases, the resolution criteria might become quite detailed and involved. To keep questions streamlined, it can be convenient to relegate some of the gory details to the fine-print.
  7. Select ranges that (likely) contain the resolution value. For range questions, make sure the bounds cover most possible outcomes, not just all outcomes that appear most likely. Select open bounds if there is even a small chance that values could fall outside the range.
  8. Avoid excessively wide ranges. Select a range wide enough that the true value is very unlikely to fall outside, but no wider.
  9. Double check. Proof-read your submission, double-check your resolution conditions and ranges, and make sure the formulation of your question is consistent throughout.
  10. Acknowledge the contributions of others. When copying text from an existing question, acknowledge and reference the previous question.
  11. Use appropriate markdown formatting. When posting URLs, remember to embed these. When writing equations and formulas, try using MathJax.
  12. Select appropriate category tags. When submitting a question, make sure that you categorize your question appropriately by selecting relevant tags.
  13. Specify units and times. When the question asks about amounts, be sure to specify the units precisely (e.g. thousands of kilograms, nominal US dollars, temperature in degrees celsius). When specifying dates and times, remember to indicate the timezone or a UTC offset, and write all dates following the format “January 1, 2040”.
  14. Pass Phil Tetlock's Clairvoyance Test. If you handed your question to a genuine clairvoyant, could they see into the future and definitively tell you whether your resolution criteria happened? Some questions like "Will the US decline as a world power?", "Will Country X be at war with Country Y?", and "Will an AI exhibit a goal not supplied by its human creators?" would all be great to know, but as standalone questions they struggle to pass the Clairvoyance Test. For example, what does "war" mean, exactly, in a post-WWII world when countries usually do not declare war and instead engage in "police actions" or "special military operations"? How do you tell one type of AI goal from another, and how do you even define it? If your question fails the Clairvoyance Test, another way to assess the outcome of interest is by asking a series of specific, testable propositions. In the case of whether the US might decline as a world power, you'd want to get at the theme with multiple well-formed questions such as "Will the US lose its #1 position in the IMF's annual GDP rankings before 2050?", "Will the US dollar lose its status as the #1 foreign exchange reserve currency?", and "Will the debt held by the public as a share of US GDP rise above 200%?".

Setting question dates

When submitting a question, you are asked to specify the closing date (when the question is no longer available for predicting) and resolution date (when the resolution is expected to occur). The date the question is set live for others to forecast on is known as the open date.

  • The open date is the date/time when the question is open for predictions. Prior to this time, if the question is active, it will have "upcoming" status, and is potentially subject to change based on feedback. After the open date, changing questions is highly discouraged (as it could change details which are relevant to forecasts that have already been submitted) and such changes are typically noted in the question body and in the comments on the question.
  • The close date is the date/time after which predictions can no longer be updated.
  • The resolution date is the date when the event being predicted is expected to have definitively occurred (or not). This date lets Metaculus Admins know when the question might be ready for resolution. However, this is often just a guess, and is not binding in any way.

In some cases, questions must resolve at the resolution date according to the best available information. In such cases, it becomes important to choose the resolution date carefully. Try to set resolution dates that make for interesting and insightful questions! The date or time period the question is asking about must always be explicitly mentioned in the text (for example, "this question resolves as the value of X on January 1, 2040, according to source Y" or “this question resolves as Yes if X happens before January 1, 2040)".

The close date must be at least one hour prior to the resolution date, but can be much earlier, depending upon the context. Here are some guidelines for specifying the close date:

  • If the outcome of the question will very likely or assuredly be determined at a fixed known time, then the closing time should be immediately before this time, and the resolution time just after that. (Example: a scheduled contest between competitors or the release of scheduled data)
  • If the outcome of a question will be determined by some process that will occur at an unknown time, but the outcome is likely to be independent of this time, then it should be specified that the question retroactively closes some appropriate time before the process begins. (Example: success of a rocket launch occurring at an unknown time)
  • If the outcome of a question depends on a discrete event that may or may not happen, the close time should be specified as shortly before the resolve time. The resolve time is chosen based on author discretion of the period of interest.

Note: Previous guidance suggested that a question should close between 1/2 to 2/3 of the way between the open time and resolution time. This was necessary due to the scoring system at the time, but has been replaced by the above guidelines due to an update to the scoring system.

Question content

In addition to specifying precise resolution criteria and following accepted formatting and practices, we have some rules that we adhere to regarding question content.

  • Questions should not contain inappropriate or offensive material. Please refrain from posting content that is predictably disruptive.
  • Questions should be on topics that are notable. Questions should reasonably be expected to be of interest to at least some other users. If your question idea is only interesting to you or a small group of friends, you can make a private question, and share it with them.
  • Public questions should not concern the personal lives of non-public figures. Questions about non-public figures are generally only appropriate as private questions.
  • Questions should not concern the mortality of individuals. Questions should never aim to predict the mortality of individual people or even small groups. In cases of public interest (such as court appointees and political figures), the question should be phrased in other more directly relevant terms such as "when will X no longer serve on the court". When the topic is death (or longevity) itself, questions should treat people in aggregate or hypothetically.
  • Questions should typically not concern harmful acts. Questions should avoid being written in a way that incentivizes harmful or illicit acts — that is, if one were to imagine that the stakes of getting a question correct were high enough to motivate someone to interfere in real-world events to change a question's resolution, those actions should not be by their nature illegal or harmful. Exceptions may be made in cases where such predictions are clearly of wide public interest, which is to be decided by Community Moderators.

Question content rules (with the exception of the notability requirement, and the no non-public figures rule) apply equally to private questions as they do to public ones.

About the Community Moderators

Metaculus Community Moderators are committed members of the community who volunteer to help the platform run smoothly. We greatly appreciate their time and talents, and we hope you'll do the same!

The main responsibilities of Community Moderators include:

  • Providing constructive feedback on user-submitted questions, including asking for revisions when necessary.
  • Accepting community-suggested questions that are ready to go live for forecasting.
  • Mediating discussions about question resolutions.
  • Answering users' questions about how the platform works, when they arise.

Additionally, Community Moderators can help draw extra attention to bugs, feature requests, or other issues that affect the platform. Community Moderators are appointed democratically, by way of election. If you'd like to register interest in becoming a Community Moderator, please do so by commenting to this effect in the comment section of this discussion thread, and we'll reach out to you before the next election takes place.