diff --git a/docs/connection-gate-findings.md b/docs/connection-gate-findings.md
index f3b998b..0762d58 100644
--- a/docs/connection-gate-findings.md
+++ b/docs/connection-gate-findings.md
@@ -1041,3 +1041,18 @@ POPULATOR (who inserts into [ctx+0x19c8]) and its data source/channel. 29 funcs
 +0x14153e010, +0x141603fc0, +0x146f5a400 (store sites). This needs the LSX-response→map mapping
 or a decompiler. Live-memory + hook-call forcing (openfut-hook `probe`/force) is proven safe and
 reusable. Weigh vs [[project_direction_pivot]].
+
+### Correction/nuance (same session): the empty map is partly downstream
+
+Checked the LSX channel: FIFA's only GetSetting requests are IS_IGO_ENABLED, IS_IGO_AVAILABLE,
+ENVIRONMENT — NOT endpoint URLs. And no HTTP config fetch happens. So the endpoint config map is
+NOT populated by a request we're mishandling; in a live client it's filled from the online
+connection/login response (post-redirector). Therefore the empty map is (at least partly)
+*downstream* of "never connected", not a pure root cause. Net integrated picture across all
+runs: FIFA's frontend shows "connecting" but the online subsystem never initiates ANY network
+connection (no getaddrinfo, no redirector dial); every deeper object (connect states, session
+ctx, endpoint map) sits in its initial empty/dormant state because no connection ever happened.
+`nucleusConnectREST` is a config-lookup (empty), not a sender, so it can't be used to trivially
+trigger the dial. The true trigger — "start the online connection / dial gosredirector" — is
+upstream and never fires; that remains the M2 frontier. Forcing was valuable (safe; corrected
+the function's role; proved the endpoint map empty) but did not find a callable shortcut.