Interactive Data Explorer

Epiphenomenon@Trove — Dataset

Explore the underlying data: threat discourse frequencies, Australia's structural profile, and WWI/WWII pre-war signals.

Peak sino-ppm
647
1905 — Yellow Peril peak
Peak russo-ppm
1,275
1950 — Cold War onset
Documents indexed
55M+
Trove digitised press 1900–1954
Corpus years
55
Annual ppm 1900–1954
Sino-referential discourse — Trove digitised press
Mentions per million words, 1900–1954
Trove
1904–1905 spike: Yellow Peril panic coincides with the Anglo-Japanese Alliance and Russo-Japanese War — first demonstration of an Asian military power defeating a European one. 1950 rise: Korean War and PRC recognition debates drive discourse to post-war high. The 1910s–1940s trough reflects editorial displacement by World War coverage, not the absence of threat framing.
Russo-referential discourse — Trove digitised press
Mentions per million words, 1900–1954
Trove
1943 peak (762 ppm): Soviet Union as wartime ally — discourse at maximum, but largely positive. 1948 spike (1,268 ppm): Cold War crystallises; the ally becomes the existential threat in a single year. 1950 peak (1,275 ppm): Korean War drives russo-ppm to the dataset maximum; Sino and Russo fears fuse for the first time.
Sino vs. Russo discourse comparison
Both series overlaid, 1900–1954
Methodology
Raw keyword counts (sino_raw, russo_raw) normalised by total annual word count to produce per-million-word frequencies. Sino terms include: "China", "Chinese", "Sino-", "Celestial", "Yellow Peril" and related compounds. Russo terms include: "Russia", "Russian", "Soviet", "Bolshevik", "USSR" and related compounds. Source: National Library of Australia Trove digitised newspaper corpus.
Peak sino-ppm
35.5
2021 — China Threat peak
Baseline (2008)
0.13
sino-ppm before the turn
Rate of change
127×
2008→2021 multiplication
Russo 2022
17.4
ppm post-Ukraine invasion
Sino-referential discourse — Australian Parliament (Hansard)
Mentions per million words, 2006–2025
Hansard
2017 inflection: First significant rise driven by foreign interference debates. 2018–2019: Huawei 5G exclusion, foreign donation legislation. 2020–2021: COVID-19 origins debate, trade coercions, diplomatic rupture drive discourse to 127× the 2008 baseline. Post-2021: Partial diplomatic restoration under Albanese government produces discourse retreat, but not to pre-2017 levels.
Sino vs. Russo — Parliamentary discourse comparison
Per million words, 2006–2025
Structural observation: Sino and Russo discourse have historically been anti-correlated in Australian parliamentary records — when one dominates, the other recedes. The 2022–2023 period is exceptional: for the first time in the dataset, both series spike simultaneously (Ukraine invasion + sustained China tensions). This co-activation of both threat discourses is structurally unprecedented in the Hansard record.
Methodology
Raw keyword counts extracted from Parliament of Australia Hansard records (House of Representatives and Senate combined). Normalised by total session word count. Years with fewer than 40 files excluded from trend analysis. 2005 and earlier not shown due to digitisation coverage gaps.
Current SHI (2026)
+0.5
Near-zero, slightly positive
Historical minimum
−8.25
1922 — Lore-Archive dominant
Historical maximum
+2.5
1975 — Vietnam-era peak
Alarm threshold
+3.0
Not yet breached historically
Australia — Shield Hypertrophy Index, 1895–2026
SHI = Shield NV − mean(Lore NV, Archive NV)
CAMS
Structural interpretation: Australia has maintained a consistently negative SHI throughout most of its recorded history — an unusually Lore-Archive dominant society. The recent positive drift (2015–2026) toward zero and above reflects AUKUS commitments, defence spending escalation, and intelligence-sharing intensification. Australia has not historically breached the +3.0 alarm threshold, but the current trajectory is the most sustained positive movement in the 130-year record.
Australia SHI — Recent period detail (2000–2026)
Showing the contemporary structural shift
Methodology
Node Values (NV) computed from CAMS scoring of Australian institutional structure across eight nodes (Helm, Shield, Lore, Stewards, Craft, Hands, Archive, Flow) on four dimensions (Coherence, Capacity, Stress, Abstraction). SHI = Shield NV − mean(Lore NV, Archive NV). Positive SHI indicates Shield dominance over the knowledge-memory complex; negative indicates Lore-Archive dominance. Source: aus_annual.csv, CAMS ensemble scoring 1895–2026.
1935 SHI (mean)
+7.0
Ensemble mean — critical breach
1935 SHI_min
+5.25
Pessimistic bound — still robust
Lead time
4 yr
Robust signal to war (1935→1939)
Weimar nadir
−8.9
1929 — cultural peak / political collapse
Germany — Shield Hypertrophy Index, 1925–1945
CAMS5 ensemble mean · Alarm (+3.0) and Critical (+5.0) thresholds marked
Pre-War
Without-hindcasting finding: Germany's SHI crosses the +5.0 critical threshold at ensemble mean in 1934 (five-year lead time to war outbreak). The signal becomes robust — clearing +5.0 even at the pessimistic bound (SHI_min = +5.25) — in 1935 (four-year lead time). This finding is validated across five independent scoring passes and survives the strictest methodological constraints applied in this project.

Weimar paradox: The deeply negative SHI of the 1925–1932 period (nadir: −8.9 in 1929) reflects the extraordinary cultural and intellectual productivity of Weimar Germany — a society whose Lore and Archive nodes dominated its institutional structure precisely during its most politically vulnerable period.
Germany SHI trajectory — year-by-year
CAMS5 ensemble mean values, 1925–1945
Methodology — Five-scorer ensemble
Germany, UK, Russia, and Norway each scored independently by five separate CAMS passes. Node Values averaged at the Coherence/Capacity/Stress/Abstraction level before Node Value computation. Inter-scorer variance reported as V_min/V_max envelope. Signals classified as Robust (SHI_min > threshold), Partial (mean > threshold, SHI_min ≤ threshold), or Ambiguous (envelope width > 8 units). Full methodology: see the Without Hindcasting working paper.