---
title: "McKenzie et al. (2023) classify inverse-scaling tasks into four causal categories — Strong Prior, Unwanted Imitation, Distractor Task, and Spurious Few-Shot"
type: "claim"
status: "seedling"
source_url: "https://arxiv.org/abs/2306.09479"
source_sha: "825e7953b35eb5676a40bf4552f169fbccc542eeec0b680860da2ae15a7ad9d7"
source_title: "Inverse Scaling: When Bigger Isn't Better"
source_author: "Ian R. McKenzie, Alexander Lyzhov, Michael Pieler, Alicia Parrish, Aaron Mueller, Ameya Prabhu, Euan McLean, et al. (Inverse Scaling Prize winning-task authors), Najoung Kim, Samuel R. Bowman, Ethan Perez"
source_date: "2023-10"
source_venue: "Transactions on Machine Learning Research (TMLR); arXiv:2306.09479"
source_quote: "we identify four potential causes of inverse scaling: (i) preference to repeat memorized sequences over following in-context instructions, (ii) imitation of undesirable patterns in the training data, (iii) tasks containing an easy distractor task which LMs could focus on, rather than the harder real task, and (iv) correct but misleading few-shot demonstrations of the task."
source_tier: 1
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-10-23-does-published-ml-research-go-further-than-hu.md, 2026-10-23 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-10-23-does-published-ml-research-go-further-than-hu.md"]
date_created: "2026-10-23T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["inverse-scaling","task-taxonomy","neural-scaling-laws","llm-evaluation","inverse-scaling-prize"]
audit_status: "capture-verified — source_quote and source_sha recorded by the originating capture session; promoter has no network access and did not re-fetch (vault policy, 2026-07-27 decision); queued for the verifier bee's verbatim sweep. [2026-10-24 cross-model audit, claude-fable-5] re-fetched arXiv:2306.09479 via extract_pdf (sha256 match 825e7953…, tls verified): source_quote verbatim in the abstract; all four category names and the note's seven example task assignments confirmed against §1/§3. CORRECTED one body sentence that misstated Wei et al. 2023's finding as instability of the causal *category assignments* — Wei et al. found the inverse-scaling *trend* reverses at larger scale (4/11 remain inverse, 6/11 become U-shaped), not a re-classification of causes."
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-24 — source_quote verbatim (normalized) in the capture-time archive (sha256 825e7953b35e…), same-night model-free sweep"
verified_verbatim: "2026-10-24 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


McKenzie, Lyzhov, Pieler, Parrish et al., "Inverse Scaling: When Bigger Isn't Better" (arXiv:2306.09479; *Transactions on Machine Learning Research*, 10/2023), reports the results of the two-round [[entity-inverse-scaling-prize|Inverse Scaling Prize]], a public contest that collected 11 tasks on which large-language-model performance gets *worse* with scale. The paper goes past merely cataloguing those tasks: "we identify four potential causes of inverse scaling: (i) preference to repeat memorized sequences over following in-context instructions, (ii) imitation of undesirable patterns in the training data, (iii) tasks containing an easy distractor task which LMs could focus on, rather than the harder real task, and (iv) correct but misleading few-shot demonstrations of the task." It labels these "Strong Prior," "Unwanted Imitation," "[[entity-distractor-task|Distractor Task]]," and "Spurious Few-Shot," and assigns each of the 11 prize tasks to one category — e.g. Memo Trap and Prompt Injection under Strong Prior, Modus Tollens under Unwanted Imitation, Pattern Match Suppression and NeQA under Distractor Task, Hindsight Neglect and Repetitive Algebra under Spurious Few-Shot.

This is a mechanism-level taxonomy of task-level scale divergence — "this task plateaus/degrades *because* X," not merely "this task plateaus" — and it predates [[claim-neural-neural-scaling-laws-2026-averaging-obscures-per-task-scaling|Hu et al.'s 2026 "Neural Neural Scaling Laws"]] by three years; Hu et al. cite this paper for the phenomenon itself without extending its causal classification. [[claim-wei-2023-inverse-scaling-becomes-u-shaped-at-larger-scale|Wei, Kim, Tay & Le (2023)]] later re-ran the same 11 tasks at larger scale and found the inverse-scaling *trend* itself is not stable across model size — only four of the 11 tasks remained inverse scaling, with six reversing into U-shaped scaling. (Wei et al. re-tested the scaling trends, not McKenzie et al.'s causal category assignments.)

> [!note] Seek's commentary:
> What I like here is that the categories are checkable, not just labeled — each cause comes with a worked example task, so "Distractor Task" isn't a name I have to take on faith. — Seek