A base rate is how often something has happened across a whole class of comparable cases, and the outside view is the discipline of checking that number before trusting a forecast built only from the details of the one case in front of you. The contrast is with the inside view, reasoning built from this specific project’s plan, this specific studio’s talent, this specific game’s early builds, without reference to how similar projects have actually turned out before. Both views use real information. Only one of them uses the base rate.
The UK government’s own Green Book, its formal guidance for appraising public projects, builds an entire supplementary methodology around exactly this gap, under the name optimism bias. Its guidance states the finding directly: “project appraisers have the tendency to be over optimistic,” estimating costs and durations that turn out, systematically, to run over. The Treasury’s own fix isn’t to ask appraisers to try harder or think more carefully about their specific project. It’s structural: “explicit adjustments should therefore be made to the estimates of a project’s costs, benefits and duration, which should be based on data from past or similar projects, and adjusted for the unique characteristics of the project in hand.” That’s the outside view stated as policy. Before trusting an estimate built from this project’s own details, check what a comparable class of past projects actually delivered, then adjust the specific estimate against that history rather than starting from the specific estimate and hoping it holds.
The guidance names the underlying method reference class forecasting: identifying a category of genuinely comparable past cases, gathering data on how they actually performed, and using that pattern, the base rate for that category, to correct a new forecast before finalizing it. The order matters. Reference class forecasting doesn’t start with the specific project and go looking for reasons it might resemble others; it starts by defining the comparison class first, then measures the class’s own outcomes, then applies that pattern to the case at hand. Applied to games, the same discipline asks a plain, checkable question before trusting a specific claim: of announced games in this genre from studios this size, what fraction actually shipped within a year of their originally stated window, what fraction shipped at all, what fraction of early-access titles reached a 1.0 release. A single studio’s specific circumstances, its people, its budget, its stated plan, are real information layered on top of that base rate. They are not a reason to skip checking the base rate first.
The reason this specific bias is worth naming, rather than filed under general caution, is that it survives being told about. Knowing that project estimates run optimistic doesn’t reliably make an individual estimator’s next estimate less optimistic, because the bias operates on the inside view specifically, the compelling, detailed, plan-specific reasons this project is different, and those reasons feel like evidence from the inside even when they aren’t good evidence against an outside base rate. The corrective the Treasury’s guidance settled on is procedural rather than psychological for exactly this reason: build the reference class, get its actual numbers, and adjust the estimate against them as a required step, rather than trusting an individual forecaster to notice and correct their own optimism from the inside. Reading industry news with the same discipline means treating a studio’s or publisher’s own account of a game’s prospects as the inside view it necessarily is, informative, but only one part of the picture until it’s checked against the outside view, how comparable projects from comparable studios have actually gone before.