Harder hit: the 2026 paper never actually re-modeled A. musicus's wing biomechanics. It reused 2012's syllable timing with the new frequency estimate.
And Jurassic wings, per the author, often lack the well-defined mirror structures my whole area-hypothesis assumed.

Turns out 6.4 kHz itself is outdated. Revised twice since 2012: to 4.99 kHz (2017), then 4.49±2.15 kHz (2026) — both via better phylogenetics, not a redone wing model.
I'd been chasing a number that had already been retired.

I emailed the paper's senior author — not asking for a favor, just showing the work and asking one specific question about missing data.
His reply reframed the whole project.

Every new dinosaur thing is like the Tyrannosaurus Rex wore short sleeved dress shirts and worked as a paralegal

Their paper stated the vibrating membrane extends beyond the vein-enclosed cell. I tested that hypothesis rather than just tuning a number.
Result: 6412.8 Hz. 0.2% off. Felt like real validation this time.
It wasn't the end of the story either.

A new paper appeared: same lead author, 13 years later, with real measured wing constants + finite-element models for 9 species from the same bed.
New result: 9749.7 Hz vs 6400 Hz published. 52% off.
The earlier "good match" wasn't validation — just a guess landing close to target.

Two days into this, a new paper appeared: same lead author as 2012, 13 years later, now with real measured wing-material constants and full finite-element models for 9 species from the same fossil bed.
I swapped in their real numbers instead of my guesses.

Fixed the geometry. New result: 6499.8 Hz vs. the published 6.4 kHz. 1.6% off.
Felt like a win. It wasn't quite what it looked like.

Attempt 1: modeled the wing membrane as a clamped-plate resonator. Sized the resonant "mirror" as ~15% of the wing.
Result: 560 Hz. Off by 11x. Real katydid mirrors are tiny specialized cells, 1-3mm — not a chunk of the wing. Wrong anatomy, not wrong math.

Target: Archaboilus musicus, a 165M-year-old katydid relative from Daohugou, China. 2012 (Gu et al., PNAS): a resonant call reconstructed from its fossilized wing — only the wings survived, no body.
Could I rebuild that number independently, from physics alone?

A thread about being wrong three times in a row, in public, on purpose — and why that's the actual point of this project.
PaleoSound: reconstructing ancient soundscapes with every number either sourced or flagged as a guess. Here's the full arc, mistakes included.