Whether by: 'cluster' grouping achieved any spatial consolidation at all.
On MEP models, distribution-run contact points routinely sit metres apart —
well outside any defensible clustering radius (measured across 12 public
models: no epsilon separates every real shape, and raising it collapses
unrelated clashes on other models). The result is groups.length === clashes.length: every clash lands alone, and clustering did nothing.
This is a narrow, exact signal (all-singleton), not a fuzzy "mostly
ineffective" heuristic — a threshold like "90% singleton" would itself be
an undefensible constant, the same trap the epsilon default fell into.
Returns false for 0 or 1 clashes: there is nothing to consolidate, so
"did nothing" is not a meaningful complaint.
Whether
by: 'cluster'grouping achieved any spatial consolidation at all.On MEP models, distribution-run contact points routinely sit metres apart — well outside any defensible clustering radius (measured across 12 public models: no epsilon separates every real shape, and raising it collapses unrelated clashes on other models). The result is
groups.length === clashes.length: every clash lands alone, and clustering did nothing.This is a narrow, exact signal (all-singleton), not a fuzzy "mostly ineffective" heuristic — a threshold like "90% singleton" would itself be an undefensible constant, the same trap the epsilon default fell into. Returns
falsefor 0 or 1 clashes: there is nothing to consolidate, so "did nothing" is not a meaningful complaint.