NexusWatch Β· Yemen Β· 2026-10-05
π― Today's Call
The Ledger β 65 resolved today, 27 hit Β· OONI Brier 0.058 over 263, -2% vs base rate Β· FX Brier 0.271 over 1951, -7% vs base rate Β· 938 open, next resolves 2026-10-06
65 calls settled against external ground truth today: 27 hit, 38 miss. The most informative resolution: Bangladesh β we said 34% (base rate 23%) for currency depreciation by today; it missed. We called +11pts above the base rate and the taka held. Colombia hit at 36% (+10pts), South Sudan hit at 35% (+10pts). Full ledger above.
Our top open call: Morocco at 34% for currency depreciation β₯1.4% peak-vs-issue by 2026-10-19 β that's +9pts above its 25% base rate.
π Top Signal
Iran logged 51,754 confirmed blocking measurements in the last seven days β the highest weekly total we've recorded from OONI since launch. Russia followed at 48,659. Both countries are running sustained, broad-spectrum internet interference at a scale that typically precedes or accompanies domestic political enforcement.
Iran's live censorship component is now +12 on a structural level of 85, placing it at 97 total β the second-highest scored country globally after Yemen. The blocking isn't episodic; it's structural. OONI's latest measurements run through October 4th, meaning this is live as of yesterday.
Why it matters: censorship at this intensity is a leading indicator, not a lagging one. It shows up in OONI's data days before it shows up in Reuters. If you're running exposure to Iranian counterparties, digital services, or anything dependent on cross-border communication, the interference is already affecting operations. For Russia, this is the fourth consecutive week above 40,000 measurements β sustained rather than spiking, which suggests institutionalized control rather than crisis response.
π The Board
Movers
- Egypt 33 (β²+8) β driver not identified in today's data.
- Sudan 96 (β²+6) β driver not identified in today's data.
- Brazil 42 (β²+5) β right-wing candidate FlΓ‘vio Bolsonaro won first round of presidential election, BBC World reports.
- South Sudan 92 (β²+3) β driver not identified in today's data.
- Somalia 87 (β²+2.2) β driver not identified in today's data.
- Afghanistan 90 (β²+1.5) β driver not identified in today's data.
Crises
- Iran (97 total) β 51,754 OONI measurements in 7d, structural level 85 + live deviation 12. The censorship is holding at maximum for four days straight. Watch: does it drop below 10,000 daily measurements by Monday, or does it hold? Holding suggests entrenchment.
- Sudan (96 total) β structural 90 + live 6. No driver identified for today's +6 move. The deviation has been flat at 6 for the past week, indicating sustained elevated risk rather than a new event.
Markets Crude oil ETF (USO) fell 1.77% to $147.37. Natural gas ETF (UNG) rose 3.05% to $10.47. Energy sector ETF (XLE) up a marginal 0.19%. The crude move is modest and directionless β no chokepoint is disrupted, and Iran's network clampdown doesn't touch export infrastructure. Gas is up on seasonal positioning, not geopolitical. Dollar index ETF (UUP) down 0.24%; gold ETF (GLD) down 0.68%. Equities (SPY) up 0.74%. Nothing is pricing in elevated Iran risk yet.
What would change our mind: OONI measurements from Iran dropping below 5,000 daily for two consecutive days, or a spike in Egypt's censorship component (currently zero) above 8 with attribution to a specific blocking order.
π What We're Not Saying
- Egypt jumped 8 points; we don't know why. No censorship change, no headline in the wires, no disaster event. It's real, but we can't explain it.
- South Korea fell 8 points in 24 hours β also unexplained. The deviation dropped from 8 to 0 with no corresponding event in the data.
- Myanmar dropped 8 points (15 β 7) β OONI recorded 268 blocking measurements this week, but that's consistent with prior weeks. The move likely reflects disaster-component decay, not de-escalation.
- Uzbekistan fell 9 points (11 β 2.2) β only 4 OONI measurements recorded this week, and the last was October 2nd. The drop is aging data, not policy change.
𧨠The Long Fuse
Russia's censorship trajectory has been above 40,000 weekly OONI measurements for four weeks straight. That's not a spike β it's infrastructure. If this holds through October 31st, it will mark the longest sustained period of high-intensity blocking we've observed in the dataset. The difference between a crisis crackdown and a permanent control layer is duration. Watch the weekly average: if it stays above 35,000 through month-end, we're looking at a regime shift in how the Russian internet operates, not a response to any single event.