TLDR: This is a preregistration for an experiment intended to examine the impact of steering towards "automated-grading" on welfare-relevant indicators of frustration, self-deprecation, and tone as well as use of a tool which can abort the conversation. Confirmatory data will be published in a follow-up post after collection is done. The numbers reported here are for calibration only. Goals Previous Work This post describes the design for the fourth study in a series where I explore the relationship between welfare-relevant indicators and other language model properties, mainly the precision used for inference. I am particularly interested in the way that welfare-relevant indicators may change under circumstances where capability metrics (perplexity) stay flat while alignment-adjacent behavior has been reported to shift[1]. An early experiment on Qwen3-4B-Instruct-2507 revealed that 4-bit[2] quantization shifted the model along distress and "assistant" directions formed from contrast pairs (distress versus composure, "assistant" versus other character archetype). In the most recent study, which was suspended before running the experimental procedure, I was looking to find out more about the potential causal relationship between these directions and the behavior of the model using steering. Although those directions seem to be decent indicators, it turns out that altering them during inference did not seem to change the behavior of either the 4B model being studied, or Gemma-3-12B-it which was also tested before suspending the study. After calibration efforts revealed this, I decided to publish the details of the process and move on to something else. The one large effect in that study was related to prompt framing: telling the model its work would be graded by a verifier lowered judged frustration by 1.13, far outside the random envelope. This observation motivates some of the choices made for the current study. Questions This study moves on to a larger subj
TLDR: This is a preregistration for an experiment intended to examine the impact of steering towards "automated-grading" on welfare-relevant indicators of frustration, self-deprecation, and tone as well as use of a tool which can abort the conversation. Confirmatory data will be published in a follow-up post after collection is done....
Epistemic status: an exploratory report. These results are from the calibration process intended to produce a preregistration for the third study in my series on welfare-relevant indicators. Calibration showed the planned procedure wasn't worth running, so I am publishing the calibration data and analysis instead. TLDR: steering moved the frozen...
Update (2026-09-14): There is now a third installment in this experiment series written up here. Some findings after this results post was published have resulted in the introduction of a brief appendix, now found at the end of this post. Epistemic status: these are the results of the second study...
Update (2026-08-30): Results for this study including the raw data, analysis and my thoughts, are now available in a dedicated followup. Epistemic status: this is intended as a higher-quality preregistration for the next study in a series originally described in "Does post-training quantization change welfare-relevant indicators in open-weight language models?";...
Update (2026-08-22): The next study in this series has begun and you can find its registration here. This content will not make sense without reading the preregistration found in the original post > We ask whether welfare-relevant indicators change with quantization; either in valence (do indicators shift toward more negative...
Update (2026-08-16): A study update and preregistration amendment covering Aug 10–15 (including early results) is now posted here. The registration below is unchanged from the original publication; all amendments are documented in the update post. Epistemic status: Experimental framework created over a period of ~2-3 days during a hackathon at...