Writing at Astral Codex Ten (post), Scott Alexander re-examined the 2020 Biological Anchors report, Open Philanthropy’s compute-anchored estimate of when transformative AI would arrive, and concluded that its single largest error was wildly underestimating algorithmic progress; hold that one variable to what actually happened and its predictions would have been remarkably accurate. The newsletter runs it alongside METR’s 10x-a-year horizon doubling and Bostrom’s “swift to harbor, slow to berth” as three readings of the same acceleration. In the recursive-self-improvement trajectory it is the forecast postmortem that pairs with December’s AI 2027 accuracy check and the 2034 date two days earlier: the variable the models are now improving for themselves is the one the field’s canonical forecast got most wrong.
Development
The 2020 forecast's biggest error was algorithmic progress
A retrospective on the field's canonical compute-anchored timeline locates the miss in the one variable recursive self-improvement acts on: the compute assumptions held, and the algorithms outran them.
Claim
Scott Alexander's postmortem on the 2020 Biological Anchors report found its largest error was wildly underestimating algorithmic progress, without which its predictions would have been accurate.
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to February 13, 2026 (theinnermostloop.substack.com)
- What Happened With Bio Anchors? (www.astralcodexten.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.