Independent Estimate Round¶
Elicitation protocol — instantiates Reference-Class Planning Calibration
Collects each expert's estimate privately and simultaneously, before any sponsor target or group discussion can anchor the room, then reveals the spread.
The moment a number is spoken aloud in a planning meeting — a sponsor's hoped-for date, a senior engineer's confident guess — every subsequent estimate bends toward it, and the group's apparent consensus is really an echo. Independent Estimate Round is the facilitation ritual that prevents this by changing the order of operations: each participant produces an estimate privately and at the same time, with no target announced and no discussion, and only after all estimates are in does anyone see the others. Its defining move is the quarantine of the anchor — the sponsor's target, the loudest voice, the first number — is kept out of the room until the independent estimates are locked, so the spread that emerges reflects genuinely separate reads of the problem rather than convergence on whoever spoke first. It protects the independence of estimates; it does not, on its own, tell you which estimate is right or reconcile them against historical outcomes.
Example¶
A video-game studio is scoping a new multiplayer mode and needs an honest read on how many sprints it will take before the publisher is told anything. The producer runs a Wideband-Delphi-style round.[n1] The publisher's preferred ship window is deliberately withheld. Each of seven contributors — two gameplay engineers, a netcode specialist, a designer, a QA lead, an artist, a producer — writes a private sprint estimate for the mode against a shared feature brief, at the same time, with no talking. The estimates are revealed together: they range from 6 sprints to 15. That spread is the payoff — not because it converges, but because it does not. The netcode specialist's 15 and the designer's 6 reveal a real disagreement about how much online-stability work the mode needs, one that a normal meeting would have buried under the first confident number. The round surfaces the divergence so it can be discussed on its merits; the outliers explain their reasoning, and a second private round is taken. The publisher's target never touched the estimates that matter.
How it works¶
- Withhold the anchor. Any sponsor target, prior estimate, or authority opinion is kept out of the room until the round closes. This is the quarantine, and it is the whole reason the protocol exists.
- Estimate privately and simultaneously. Each participant commits a number without seeing or hearing others', so no estimate is contaminated by another. Same brief, same moment, separate answers.
- Reveal together and read the spread. All estimates surface at once. The dispersion is the primary signal — a tight cluster means shared understanding, a wide one means a hidden disagreement worth naming.
- Iterate on divergence, not toward the anchor. Outliers explain; a further private round may follow. Convergence, if it comes, follows from exchanged reasoning — never from an announced target.
The protocol is agnostic about what the estimates are based on; its guarantee is only that they were formed without cross-contamination.
Tuning parameters¶
- Panel composition — who estimates. A diverse panel surfaces more independent perspectives and a truer spread; a homogeneous one converges fast but may share the same blind spot.
- Rounds and stopping rule — one shot or iterate until dispersion narrows. More rounds refine understanding but risk grinding genuine disagreement into false consensus.
- Anonymity depth — whether revealed estimates are attributed. Anonymity frees juniors to contradict seniors but weakens accountability for a wild guess.
- Aggregation of the spread — whether the output is the median, the full range, or the flagged outliers with reasons. Reporting only a central number discards the disagreement the round was run to expose.
When it helps, and when it misleads¶
Its strength is protecting local knowledge from premature convergence: it is the direct countermeasure to anchoring and to sponsor pressure, and it treats a wide spread not as a failure to be smoothed but as information about where understanding diverges. Sequencing estimation before discussion — the logic of Wideband Delphi and of planning poker — is what keeps the group's answer from collapsing onto the first or the most senior number.[n1]
It misleads when independence is only cosmetic: if participants can guess the sponsor's target, or if a dominant personality's likely view is common knowledge, the "independent" estimates quietly re-anchor on it and the round launders a foregone conclusion. It also misleads when the group treats the estimates as calibrated in themselves — seven independent inside-view guesses can be independently, uniformly optimistic, and averaging them corrects the anchoring bias without touching the optimism bias. The guarding discipline is to genuinely enforce the quarantine (no target, no discussion, real simultaneity) and to feed the resulting estimates onward into a reference-class comparison rather than trusting the panel's average as a finished forecast.
How it implements the components¶
independent_estimate_channel— its core mechanism: each expert's estimate is captured through a private, simultaneous channel so no estimate contaminates another.sponsor_target_quarantine— it holds the sponsor's target (and any early anchor) outside the room until the independent estimates are locked, so the target cannot shape them.
It does not compare the estimates against a base-rate distribution or apply the outside-view correction (outside_view_adjustment_rule) — that reconciliation is the Reference-Class Forecasting Workbook's job; this round only guarantees the estimates were formed cleanly, then hands them on. It also does not stress-test the story for missed failure modes (class_membership_challenge); that is the Premortem as Auxiliary Probe.
Related¶
- Instantiates: Reference-Class Planning Calibration — supplies the un-anchored independent estimates the calibration then adjusts.
- Sibling mechanisms: Reference-Class Forecasting Workbook · Premortem as Auxiliary Probe · Three-Point Estimate with Base Rates · Historical Project Outcome Database · Schedule and Cost Risk Register · Contingency Reserve Formula · Launch or Commitment Readiness Gate · Forecast Backtesting Review
Editorial Notes¶
Form Classification¶
Form family: Communication, Facilitation & Learning
Rationale: The mechanism is a structured elicitation event that privately collects simultaneous expert estimates before discussion and then reveals the spread.
Nearest alternative: Protocol, Workflow & Routine — Collection and reveal are ordered, but participatory elicitation of independent judgment is the operative form.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Computer Science & Software Engineering
Origin pattern: Convergent development
Present-day reach: Multi-domain
Rationale: The private-commit-then-reveal estimation round is most characteristically formulated in software engineering through Wideband Delphi and planning poker.
Related originating lineages:
- Futurism & Strategic Foresight — RAND Delphi practice established anonymous independent elicitation before group revision.
- Operations Research — Formal project estimation and planning supplied a parallel quantitative decision tradition.
Review resolution: Both independent reviews place the primary lineage in computer_science. The queued differences (reported_ambiguity, alternate_origin_disagreement) concern secondary metadata rather than primary provenance. The final retains futurism_foresight, operations_research only where a reviewer supplied a formative-lineage rationale; this does not convert downstream applicability into origin. origin_mode=convergent because the reviewers document independently established or materially co-developing traditions. domain_reach=multi_domain records application breadth separately from provenance.
Attribution caveat: The mechanism combines the Delphi lineage with a specifically software-project estimation form.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] Wideband Delphi, a structured estimation method popularized by Barry Boehm, and its lightweight descendant planning poker, both have experts commit estimates privately before any discussion, then reveal and iterate — deliberately sequencing independent judgment ahead of the conversation that would otherwise anchor it. ↩a ↩b