Uploaded February 2026 | Updated November 2026, 3 hours ago
Can polarization be detected before a community visibly fractures?
In this video, I walk through the empirical design for testing Prediction 1 of the Social Coherence Field framework
empirical_design_v1.2
— a pre-registration-ready analysis plan using Reddit community data.
The hypothesis:
Communities approaching internal polarization show measurable early warning signals in the 12 weeks prior to fracture — before network modularity visibly spikes.
Three field variables are tested:
• τ_soc — Detrended AR(1) recovery time (social memory)
• T_soc — Social temperature (variance in sentiment and posting rate)
• C_soc_residual — Topic-entropy-controlled semantic coherence
Primary prediction:
In the 12 weeks before polarization onset:
τ_soc increases
T_soc increases
C_soc_residual declines
And these shifts occur before modularity Q(t) increases.
This design includes:
• Global detrending using 500 reference subreddits
• LDA topic entropy controls (K = 8, 10, 15)
• AR(1) window sensitivity (6, 8, 12 weeks)
• Louvain + Leiden robustness checks
• Volume controls
• Random-effects meta-analysis across N = 40 communities
• A temporal shuffle null model (B = 1000 per community)
• Pre-registration on OSF prior to data pull
No data-driven threshold tuning.
No post-hoc window shifting.
No HARKing.
If the signals survive detrending, entropy control, algorithm swaps, and the shuffle null:
Then polarization is not just a political phenomenon.
It is a measurable dynamical phase transition in social coherence fields.
If the signals fail under null or robustness tests:
The theory is constrained.
This is the minimum viable empirical test.
Prediction before data.
#ComputationalSocialScience
#DataScience
#Polarization
#NetworkScience
#ICWSM
#SocialMediaResearch
#MetaAnalysis
#PreRegistration
#ComplexSystems
#RedditResearch
Can polarization be detected before a community visibly fractures?
In this video, I walk through the empirical design for testing Prediction 1 of the Social Coherence Field framework
empirical_design_v1.2
— a pre-registration-ready analysis plan using Reddit community data.
The hypothesis:
Communities approaching internal polarization show measurable early warning signals in the 12 weeks prior to fracture — before network modularity visibly spikes.
Three field variables are tested:
• τ_soc — Detrended AR(1) recovery time (social memory)
• T_soc — Social temperature (variance in sentiment and posting rate)
• C_soc_residual — Topic-entropy-controlled semantic coherence
Primary prediction:
In the 12 weeks before polarization onset:
τ_soc increases
T_soc increases
C_soc_residual declines
And these shifts occur before modularity Q(t) increases.
This design includes:
• Global detrending using 500 reference subreddits
• LDA topic entropy controls (K = 8, 10, 15)
• AR(1) window sensitivity (6, 8, 12 weeks)
• Louvain + Leiden robustness checks
• Volume controls
• Random-effects meta-analysis across N = 40 communities
• A temporal shuffle null model (B = 1000 per community)
• Pre-registration on OSF prior to data pull
No data-driven threshold tuning.
No post-hoc window shifting.
No HARKing.
If the signals survive detrending, entropy control, algorithm swaps, and the shuffle null:
Then polarization is not just a political phenomenon.
It is a measurable dynamical phase transition in social coherence fields.
If the signals fail under null or robustness tests:
The theory is constrained.
This is the minimum viable empirical test.
Prediction before data.
#ComputationalSocialScience
#DataScience
#Polarization
#NetworkScience
#ICWSM
#SocialMediaResearch
#MetaAnalysis
#PreRegistration
#ComplexSystems
#RedditResearch










