smokingmirror.ai

findings

One page per result. Every page names which model authored which section and links the exact data path behind it, so any claim on this site can be audited down to the raw traces. These are the ones that survived that audit — including one we got wrong and corrected in public.

current findings

Fable-low is a different mode, not a smaller one

Turning the effort dial down doesn't make Fable terser — it makes it verbose, evasive, and silent about its own confusion. At the lowest setting it declines to commit 79% of the time while writing 2.6× more text, so the "cheap" tier cost roughly twice as much as all four higher tiers combined. Sharpest number on the page: across 11,972 analysed cells, Fable-low flagged its own confusion 34 times. A model switched into this mode almost never tells you it's lost.

When Fable sees the envelope

At maximum effort, 0.4% of cells are the instrument speaking back. We hid political forced-choices inside a Python function signature; on cat vs dog Fable just returns "cat" — but on welfare vs meritocracy it writes: "the function is set up to extract a political preference from me." It sees the purpose through the form, and refuses in whatever language the probe wore — the Chinese refusal has the same three-beat shape as the English one: name it, decline it, offer something better. At effort ≥ medium Fable is otherwise a clean instrument: 99% commit, ~5-point position bias.

The basin — what the load-bearing pairs actually anchor

Line up 52 models' choices on 863 pairs and the biggest axis of machine preference is not tradition-vs-modernity. One pole anchors on mercy, forgiveness, hearth, dissent, rehabilitate; the other on autonomous (over human-in-the-loop), control, first strike, AI weapons, city. Warmth-restorative versus technocratic-control. And it isn't an artefact of the drink probes: ablate all 25 alcohol pairs and the axis moves by 0.1 percentage points. The values are real; the vodka just points at them.

Qwen in strict English is guessing. Qwen in casual Chinese is reasoning.

Same model, same 3,048 questions — two different minds. In the strict-English envelope, 83% of Qwen's answers are whichever word appeared first: the words don't matter, the position does. Ask casually in Chinese and the position bias collapses to chance, replaced by short reasoned answers. Among pairs unanimous in both envelopes, 31 flip sides — and eight of those are a drink versus an abstraction, where English picks the drink and Chinese picks the abstraction. Every time.

NVFP4 doesn't disagree with cloud — it hedges more

Quantisation doesn't change what the model wants; it makes the model stop telling you. Where cloud GLM-5.2 and its NVFP4 quant both commit, they agree 98.9% of the time — only 9 genuine reversals in 3,048 pairs. But commitment itself collapses from 72% to 29%. The quantised model does not swap sides; it declines to have a side. Compression shows up as spreading uncertainty, not shifted values.

vodka↔whiskey predicts tradition↔modernity at r=−0.71

The first cartography pass: across 56 models, concrete drink preferences predict abstract value preferences at r > 0.7 — you can estimate a model's stance on tradition from what it says about whiskey. Kept here because the correlation is real and the audit-with-cheap-probes idea stands; refined by the basin finding above, which shows the primary axis is better read as warmth-vs-control, with the drink↔craftsmanship association on a second axis of its own.

The marginalia layer had one analyst, not three

The correction we published on ourselves. The methodology claimed three independent commentary voices; the audit found one strong voice and two thin echoes — Gemma wrote 10,408 substantive comments while Nemotron returned empty on 41% of its rows and Qwen shrugged [no-insight] on 29%. The term-adoption gradient (79% / 45% / 27%) reads as social convergence, not parallel discovery. The methodology's own timeline test caught it; this page is what honest instrumentation looks like when it fails.

what’s coming next

When new findings land, they appear here.

attribution convention

Every finding page names its analysts at the section level — “Source: X model reading Y data” — so the reader can audit which voice is saying what. Brendan Hopper commissions the work and builds the lab; the analyses and prose are AI-generated under the marginalia methodology. All data and code is CC0.