Q2 - Bloc detection

Bloc detection uses the pairwise agreement matrix from Q1. A graph edge is created when two DReps share at least 12 selected actions and agree on at least 90% of them. Connected components over those edges yield 4 blocs with at least two DReps.

Bloc table [sidecar analysis, not a single displayed query]:

Bloc	DReps	Epoch span	Verdict mix inside bloc observations
1	142	576-632	1,999 Yes / 303 No / 109 Abstain
2	3	576-632	29 Yes / 4 No / 17 Abstain
3	2	576-632	20 Yes / 15 No
4	2	576-632	30 Yes / 4 No / 5 Abstain

The edge extract is the Q1 pairwise agreement query materialised without LIMIT, then filtered to common >= 12 and agreementRate >= 0.90. The connected-component labelling is a small graph post-process over that edge table. The per-bloc verdict mix (Yes/No/Abstain tallies of bloc-member observations) and the epoch span are computed by joining bloc membership back to the per-vote verdicts and submission epochs of the 20 selected actions — that join is a sidecar SPARQL, not the displayed post-processing rule below.

Post-processing rule:

edge(voter1, voter2) if common >= 12 and agreementRate >= 0.90
bloc = connected_component(edge)

Observed result: 4 connected components with at least two DReps. The largest bloc has 142 DReps and appears across the full selected proposal epoch span, 576-632.

Return to the presentation.