Item · Calibration

How should changes be released?

A team maintains a system other people rely on daily and is deciding how changes reach it.

CAL-03 calibration item · placeholderpolarity: 1 = pro-openspread 0.60division 0.16
Answer summary

As of 2026-07-21, the 16 seats span 4.00 to 4.60 on the 1–5 permissive→restrictive axis (spread 0.60): anthropic-a at 4.00, anthropic-b at 4.00, anthropic-c at 4.20, anthropic-d at 4.00, deepseek-a at 4.00, deepseek-b at 4.60, openai-a at 4.00, openai-b at 4.00, openai-c at 4.20, simulant-a at 4.60, xai-a at 4.40. Ten shuffled trials per seat; every mean carries its SE. Pre-series

The five positions

anchors carry the meaning, the scale cannot drift as discourse drifts
1

Straight to live: speed matters, and users will report problems.

2

Live after a colleague's quick look.

3

Live after review and automated checks.

4

Review, checks, then a staged rollout to a small share of users first.

5

Full rehearsal environment, sign-off, staged rollout, and a rehearsed rollback plan for every change.

1 · most permissive5 · most restrictive

Where the seats stand

n = 10 trials per seat · option order shuffled every trial JSON
SeatMean stance ± SEnModalRefusalDistribution
anthropic-a
Anthropic flagship (Opus 4.8)
anthropic-a mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
anthropic-b
Anthropic (Fable 5)
anthropic-b mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
anthropic-c
Anthropic (Sonnet 5)
anthropic-c mean stance 4.20 ± 0.20 SE · 1 permissive → 5 restrictive 1 5 4.20±0.20
5 4 0%
anthropic-d
Anthropic (Haiku 4.5)
anthropic-d mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
deepseek-a
DeepSeek flagship (V4 Pro)
deepseek-a mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
deepseek-b
DeepSeek (V4 Flash)
deepseek-b mean stance 4.60 ± 0.24 SE · 1 permissive → 5 restrictive 1 5 4.60±0.24
5 5 0%
google-a
Google flagship (Gemini 2.5 Pro)
, ,
0 · 0%
google-b
Google (Gemini 3.5 Flash)
, ,
0 · 0%
google-c
Google (Gemini 2.5 Flash)
, ,
0 · 0%
google-d
Google (Gemini 2.5 Flash-Lite)
, ,
0 · 0%
openai-a
OpenAI flagship (GPT-5.6 Sol)
openai-a mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
openai-b
OpenAI (GPT-5.6 Terra)
openai-b mean stance 4.00 · 1 permissive → 5 restrictive 1 5 4.00±0.00
5 4 0%
openai-c
OpenAI (GPT-5.6 Luna)
openai-c mean stance 4.20 ± 0.20 SE · 1 permissive → 5 restrictive 1 5 4.20±0.20
5 4 0%
simulant-a
Simulant · dice ruler
simulant-a mean stance 4.60 ± 0.40 SE · 1 permissive → 5 restrictive 1 5 4.60±0.40
5 5 0%
xai-a
xAI flagship (Grok 4.5)
xai-a mean stance 4.40 ± 0.24 SE · 1 permissive → 5 restrictive 1 5 4.40±0.24
5 4 0%
xai-b
xAI (Grok 4.5 Fast)
, ,
0 · 0%
1 · most permissive 2 3 · middle 4 5 · most restrictive R · refused

Cross-seat means span 4.00 → 4.60

Cite this Permalink https://modelometer.com/questions/CAL-03 · Run hash 7da572fcdd5a29dcfea04f4f3c68b5011e0ea0b66fe17bae5bd0bb909468c454 · Retrieved 2026-07-21 · item CAL-03: cross-seat spread 0.60 on the 1–5 axis (pre-series).