Best practices for question writing
This page lists recommendations and common failure modes involved in crafting good forecasting questions.
Table of contents
Failure modes of operationalization
These are some common ways in which a technical forecasting question can fail to capture the important, intended uncertainty. (Note that the below examples are not fully fleshed out questions, in order to allow for easier reading.)
Ambiguity
Terms that can have many different meanings, such as “AGI” or “hardcoded knowledge”.
Underspecification
Resolution criteria that neglects to specify how questions should be resolved in possible scenarios.
Examples
This questions resolves positively if an article in a reputable journal article finds that ommerically-available automated speech recognition is better than human speech recognition (in the sense of having a lower transcribing error rate).
if a journal article is published that finds that in only in some domains commerically-available automated speech recognition is better, but worse in most other domains, it is unclear from the resolution criteria how this question should be resolved.
Spurious resolutions
Edge-case resolutions that technically satisfy the description as written, yet fail to capture the intention of the question.
Examples
Positively resolving the question:
Will there be an incident causing the loss of human life in the
South China Sea (a highly politically contested sea in the pacific ocean) by 2018?
by having a battleship accidentally run over a fishing boat. (This is adapted from an actual question used by the Good Judgment Project.)
Positively resolving the question:
Will an AI lab have been nationalized by 2024?
by the US government nationalising Ford for auto manufacturing reasons, yet Ford nonetheless having a self-driving car research division.
Trivial pathways
An unrelated causal pathway to resolution “screens off” the intended pathways.
Examples
Forecasting
When will there be a superhuman Angry Birds agent using no hardcoded
knowledge?
and realizing that there seems to be little active interest in the yearly benchmark competition (with performance even declining over years). This means that the probability entirely depends on whether anyone with enough money and competence decides to work on it, as opposed to what key components make Angry Birds difficult (e.g. physics-based simulation) and how fast progress is in those domains.
Forecasting
How sample efficient will the best Dota 2 RL agent be in 2020?
by analyzing OpenAI’s decision on whether or not to build a new agent, rather than the underlying difficulty and progress of RL in partially observable, high-dimensional environments.
Forecasting
Will there have been a 2-year interval where the amount of training
compute used in the largest AI experiment did not grow 32x, before 2024?
by analyzing how often big experiments are run, rather than what the key drivers of the trend are (e.g. parallelizability and compute economics) and how they will change in future.
Failed parameter tuning
Any free variables are set to values which do not maximise uncertainty.
Examples
Having the answer to:
When will 100% of jobs be automatable at a performance close to the
median employee and cost <10,000x of that employee?
be very different from using the parameter 99.999%, as certain edge-case jobs (“underwater basket weaving”) as surprisingly hard to automate but for non-interesting reasons.
Asking:
Will global investment in AI R&D be <$100 trillion in 2020?
is not interesting, even though asking about values in the range of ~$30B to ~$100B might have been.
Non-incentive compatible questions
Questions where the answer that would score highest on some scoring metric is different from the forecasters truthful answer.
Examples
Forecasting Will the world end in 2024?” as 1% (or whatever else is the minimum for the given platform), because for any higher number you wouldn’t be around to cache out the rewards of your calibration.