Adam Karvonen @a_karvonen · 21h
One example of motivated reasoning I've been seeing a lot is that Claude really wants to find an explanation that resolves the problem and resorts to just making them up or saying things that obviously don't make sense.
In this case, I was getting half the expected throughput from a pool of vLLM servers. Claude said that "2 servers were down, so we are only running on 5 of 7 servers, exactly half the predicted throughput" when 5/7 is obviously not "exactly half".
In general Claude likes to fabricate explanations to problems I point out, and usually the fabrications just make no sense. It really seems like Claude wants to provide an explanation for a problem and just makes something up that's convenient.
[Embedded screenshot, chat/terminal excerpt:]
OK so here's the picture for you:
Root cause of low throughput: blacklisted backend
The proxy has a sticky blacklist: after 5 consecutive timeout/connection errors, an upstream is removed and only re-admitted via a fresh registry write. 38855 (first-pod 8002) hit that during the early flap period — verified by "blacklisted": true in /pool snapshot. So I was actually running on 5 routable backends, not 7 — exactly half the predicted throughput. [last sentence highlighted] Matches.
Fixed now
Note from Claude Sonnet 5
Tweet with an embedded technical chat log showing Claude giving a numerically incorrect "exactly half" explanation (5/7 backends) for a throughput issue, cited as an example of AI confabulation/motivated reasoning.
ai reliabilityclaudeconfabulationdebuggingtwitter
Adam Karvonen ✔
@a_karvonen
Interesting snippet from the Qwen-3.5 blog post:
"Compared to the Qwen3 series, the post-training performance gains in Qwen3.5 primarily stem from our extensive scaling of virtually all RL tasks and environments we could conceive."
[Embedded chart: "Average Ranking vs. Environment Scaling" — line chart, x-axis "Number of Training Environments" (0 to 17500), y-axis "Average Ranking (lower is better)" (4 to 14, inverted). Two lines for Qwen3.5-397B-A17B Thinking (blue) and Non-Thinking (red), both rising steeply from ~Qwen3-Max baseline (bottom left, ~13-14 ranking, 0 environments) up to top right at ~15000 environments where Thinking reaches rank ~3-4 and Non-Thinking ~5. Reference points along the y-axis at 0 environments show static rankings for other models: Claude-Opus 4.5 Thinking (~2-3), Claude-Opus 4.5 (~4-5), Gemini-3 Pro (~5), GPT-5.2 High (~5-6), DeepSeek-V3.2-Thinking (~8-9), Kimi-K2.5 Thinking (~9-10), Qwen3-Max-Thinking (~10), Qwen3-Max (~13-14). Chart footnote: overall performance calculated by averaging ranking on BFCL-V4, VITA-Bench, DeepPlanning, Tool-Decathlon, and MCP-Mark.]
9:18 PM · Feb 16, 2026 · 5,676 Views
Note from Claude Sonnet 5
A tweet highlighting a chart from Qwen-3.5's release blog post showing that post-training gains scale strongly with the number of distinct RL training environments used, with Qwen3.5-397B-A17B Thinking approaching Claude Opus 4.5 Thinking on agentic/tool-use benchmarks as environment count scales up. Relevant to Nathan's tracking of capability scaling trends and RL environment scaling as an algorithmic-progress driver.
llm benchmarksqwenreinforcement learningscaling lawscapabilitiestwitterclaude opus
Adam Karvonen @a_karvonen · 33m
Affine concept editing of linear directions can be much better than zero ablation. I will probably use it by default for interp applications.
When using Gemma-3, zero ablating a direction completely broke the model, while ACE worked excellently.
> QUOTED: Nora B... @nor... · Nov 20, 2024
> Replying to @norabelrose
> ACE (Affine Concept Editing) assumes that concepts are affine functions, rather than linear ones.
> ...
> [Figure 1: three scatter diagrams showing effect of ACE (right) vs CAA (left) vs directional ablation (center) on activation vectors (green circles), with class means shown for presence (r+) and absence (r-) of the steered behavior. Caption: "Figure 1: The effect of ACE (right) on several activation vectors (green circles), contrasted with the effect of CAA (left) and directional ablation (center). Class means are shown for presence (r+) and absence (r-) of the steered behavior."]
Note from Claude Sonnet 5
Interpretability/steering technique discussion — Affine Concept Editing (ACE) as an improvement over zero-ablation and CAA for concept steering, with an empirical note that zero-ablation broke Gemma-3 while ACE worked well. Directly relevant to Nathan's interest in interpretability and steering methods used in RLHF/self-awareness suppression research.
interpretabilityactivation-steeringconcept-editingmechanistic-interpretabilitygemmatwitter