Methods annex — Syrian foreign-policy strategic instrument (v2, 2026-07-09)

Three distinct measured reads — SDSI (engagement), SYI (realized yield), SPI (latent potential) — empirically correlated (NOT statistically orthogonal); the interesting cases are the off-diagonal ones. Linked by a funnel (Potential → Intensity → Yield) into a Diplomatic-ROI ratio and a six-posture strategic matrix. Data cut-off per issue; v1 = July 2026.

SDSI — Syrian Diplomatic Significance Index (engagement / cost)

Per partner k, quarterly: SDSI_k = Σ Eⱼ · Wⱼ — engagement events × principal weight (desk < ambassador < foreign minister < president). Built from official state-channel event data; principal-weighted, trip-collapsed, multilateral-venue-corrected (the COP30 fix). Stated limitation: this measures publicized engagement, so back-channels are under-counted — the priority v2 upgrade is triangulation with an external event stream (GDELT / ICEWS).

SYI — Syria Yield Index (realized return)

Additive, time-bound ledger with decay: SYI_t = Σ(Wᵢ·Vᵢ,t) + SYI_{t−1}·e^−λ, with λ = ln2/4 per quarter (half-life = one year) — parameterized 2026-07-09; issue 1 is a single seven-month window in which decay is uniform by construction, so the term becomes operative from issue 2. Three tiers: (Ⅰ) political normalization — mission upgrade, sanctions waiver, UN-vote, ratified treaty; (Ⅱ) capital; (Ⅲ) demographic-security — repatriation, security cooperation, capacity-building.

• Normalization lane (lane 4, added 2026-07-10): sanctions lifting IS yield — the period's principal deliverable — but a different currency (a state change, not a flow). Governing principle: the act is yield (SYI, decays with e^−λ); the resulting state is activatability (SPI's A term, persists) — no value lost, no double-count. Fed by the curated sanctions-response ledger (14 accepted/promoted acts, day-one reviewer decisions honored): act weight (رفع/تعليق 5 · ترخيص/إعفاء 4 · تخفيف قطاعي 3 · رفع تجميد 3 · شطب 2) × actor weight (US 3 · EU/UN 2.5 · UK 1.5 · others 1) × direction; contribution = min(20, 0.6 × points) so the lane informs the composite without dominating it. Effects: US 39.6→59.5 (#1 — the EO + Caesar waiver credited), EU entity 51.2, Japan 22.1 (bank unfreezing → returns to Gem), Switzerland 14.3. Known curation note: one US row is an echo (UNHCR praising the lift) — absorbed by the cap, flagged for next ledger pass. Pending normalization (Caesar repeal in Congress, EU review clauses) belongs to a future signed-not-statutory signal, symmetric with the capital pipeline.
• Relief-lane rubric (documented 2026-07-09, retro-consistent with all stored values): Tier 1 systemic donor (multi-program, ≥$500M scale) 35–45 pts — US 42, EU 37.5 · Tier 2 major multi-program ($100–500M) 12–35 — UN agencies 12.5 · Tier 3 significant program ($20–100M) 5–12 — Switzerland 10, Iraq 8 (220kt wheat ≈ $60–70M) · Tier 4 targeted / in-kind (<$20M) 1–5 — Japan/Australia/Canada 5, Kuwait 4, Norway/UK/Ukraine 3.
• Capital lane = amount × execution-tier. Deal magnitudes are extracted from the archive and weighted by delivery status — discussed 0.15 · signed 0.5 · executed 1.0; per-deal capped at $25B and per-counterpart log-normalized. Precise dollar sums from news are noisy (co-occurring third-country figures), so magnitude is used directionally, not as fake precision. MoUs count as pipeline, not realized yield. Canonical computation persisted in `syi_compute.py` (2026-07-09): SYI = 0.832·relief + 0.235·agreements + 4.3·ln(capital_realized_$M); pipeline signal = 8.2·ln(pipeline_$M). Manual curation pass (2026-07-09, evidence-adjudicated): Norway dual-key merged (5.3); UAE/Germany/Japan/Italy/Qatar capital re-based on the underlying archive posts (phantom sums removed — e.g. Germany had absorbed a $400B reconstruction-cost estimate, Egypt a reverse-attributed figure); coverage extended to 29 counterparts (Iraq's 220,000-ton wheat grant, Kuwait's returnee cash program, Oman/Morocco agreements). Posture consequences accepted: Japan gem→opportunity, Egypt anchor→opportunity-convert.
• Pipeline signal (signed-not-executed + discussed) tracks maturing deals separately from realized yield — this is what distinguishes an Investing partner (Qatar's $14B MoU, France's CMA CGM) from a Resource Sink.
• Lanes kept disaggregated (relief / agreements / capital) — no false equivalence. Measures association, not causation.

SPI — Syria Potential Index (latent value)

Multiplicative: SPI = C^α · A^β (α=β=1 default; exponents exposed only for an explicit council tilt; C, A on 0–1). v3 change: Feasibility was removed from the product — as a multiplier it floored ~150 states at ×0.45 and crushed precisely the high-capacity, under-engaged states the index exists to surface (India, Switzerland, Canada, Finland all halved). Potential is now the size of the reachable prize; Syria's own footprint is reported as a separate Readiness axis (activation cost), never a prize discount.

• C — Capacity: the partner's objective size-of-prize across six Syria-weighted categories (Economic .26 · Political .22 · Developmental .15 · Refugee-return .14 · Security .13 · Cultural .10), aggregated arithmetically within each category (compensatory) and weighted across. Diaspora enters as diaspora-capital = population × host per-capita income. Regional-pivot sub-indicator (v3.1, in Political): demographic weight (log-population, normalized to India) + regional-institution anchor (hosting/leading AU, ASEAN, ECOWAS, OIC, Mercosur seats; hub function — Addis, Nairobi, Jakarta, Abuja), uplifting Po by 0.30 × pivot, capped at 100. Corrects GDP-proxy blindness to demographic/regional significance (Kenya Po 41→61, Ethiopia 42→65, Nigeria 57→82). Knowledge-encoded; verify against source datasets before publication.
• A — Activatability: the partner's willingness — inbound diplomatic presence (the 2×2) × sanctions-profile × posture. (NLP rhetorical-posture is a flagged v2 addition needing new data.) A-principle (v3.1): willingness measures the partner's observable openness and must never encode Syria-side strategic risk — that is a policy-layer judgment, not a prize discount. The v1 pariah-cost had leaked into A for Moscow-aligned states, zeroing them against our own engagement evidence (Russia read A=0.09 while holding a Damascus chargé and producing SDSI 339 incl. summit-level meetings — internally contradictory). Recalibrated: Russia 0.45, Venezuela/Belarus/Cuba 0.30 (all hold Damascus missions); North Korea stays 0 (no mission).
• R — Readiness (formerly Feasibility; a separate reported axis, NOT in the product): Syria's own execution bandwidth — outbound footprint (the 2×2) as proxy. Footprint counts both embassies and consulates (a consulate ≈ 0.4 of an embassy, +0.15 each capped at 1.0): Syria's consular network — Gaziantep & Istanbul (Türkiye), Bonn (Germany), Jeddah (Saudi), plus consular operations at Doha, Brussels, Athens, Beirut — is recorded and weighted, and sits precisely where the diaspora/refugee populations are. Read jointly with SPI: high potential + low readiness = greenfield (open the mission first); high potential + high readiness = immediately actionable. Full MFA budget / staffing / language-desk data is pending, MFA-supplied.

Linkage — the funnel and the matrix

• Diplomatic ROI: ρ = SYI_t / (mean SDSI_{t−i} + ε), lag-corrected (2–4-quarter incubation).
• Strategic Posture Matrix — four postures on a 2×2 (v4): axes = engagement (SDSI, x) × realized yield (SYI, y); SPI splits the no-yield column. Returning (SYI ≥ 20 or pipeline ≥ 45): Anchor if heavily engaged (SDSI ≥ 200), else Gem (scale a proven channel). Not yet returning: Opportunity if SPI ≥ 15 — subtype convert (SDSI ≥ 100: Lebanon, Iraq, UK, China) or greenfield (Switzerland, Indonesia, India — open the channel) — else Peripheral (monitor). Regional-opportunity sub-posture (v4.2, renamed from "gateway"): the global SPI bar compares every state against the world's largest prizes, which structurally buries the under-engaged regions the instrument exists to surface; so Opportunity has two tiers — global (SPI ≥ 15) and regional (فرصة إقليمية): within each geographic peer group (Africa, Arab world, Latin America, SE Asia, South Asia, Central Asia/Caucasus, Oceania) the top-k peripheral states by SPI (k = 6/2/3/2/2/2/1, floor SPI ≥ 8) are promoted, because relative to their region they are its leading prizes — Ethiopia (AU seat), Kenya (Nairobi UN hub), Senegal, Uganda, Mexico, Chile, Costa Rica, Singapore, Philippines, Kazakhstan, Armenia, New Zealand. Rendered one tier lighter (map fill-opacity .55, chip «فرصة إقليمية»); "regional opportunity" was chosen over "gateway" because it claims only relative standing, not a hub function, and frames the state as a prize in its own right. States that clear the global bar on absolute size (Nigeria 17.1, South Africa 22.3) are full global opportunities and need no sub-tier. The SPI scale itself is untouched, so cross-region comparability survives for the funnel/ROI reads. Over-engaged peripherals carry a wind-down flag (SDSI ≥ 200, no yield, no pipeline, SPI < 15): Russia is the sole case — the former Sink class, demoted to a flag because a one-member class earns an annotation, not a colour. Gem vs Opportunity in one line: a Gem's return is a fact (realized yield on minimal effort), an Opportunity's is a bet (high SPI, zero yield so far) — one lives on the yield axis, the other on the potential axis. Evolution 6 → 5 → 4: Investing → Anchor (pipeline route); Dormant → Peripheral (residual ground); Sink → wind-down flag. Rationale: heavy effort with no yield is a drain only when there is nothing left to win — potential makes that distinction, an engagement cutoff can't. The ministerial decision exhibit; rendered as the posture world-map (194 states, micro-states as coordinate dots).

Exhibits

The instrument renders as an interactive Arabic artifact: posture world-map (194 states) + effort×yield matrix (scatter: x = SDSI log-scale, y = SYI, dot size = SPI, colour = posture, gold guide-lines at the two decision thresholds SDSI 200 / SYI 20, quadrant annotations, per-dot tooltips; origin-dwelling peripherals excluded and counted in the caption — no silent truncation). Posture palette CVD-checked: contrast ≥3:1 passes both themes; identity never rides on colour alone (position-as-quadrant, size, rings, direct labels, tooltips). Data correction (2026-07-09): Australia Rf 0→44 (calibrated vs UK 45/US 46: ~50k Syria-born + ~12k dedicated resettlement 2015-17) → SPI 15.4, greenfield opportunity.

Method hygiene (reviewer-driven)

• Normalization to 0–1, family-appropriate (log for skewed monetary/counts).
• Weights default to neutral; every knob (πₖ, exponents, tier weights, λ) is an explicit, sensitivity-tested council choice, never a hidden default. Sensitivity analysis is standing practice — the opportunity ledger's core held across six weight scenarios.
• Validation: retrodiction (the instrument reproduces the Iran/Russia → Gulf/Türkiye pivot) + a quarterly opportunity→yield feedback audit.
• Reproducibility: freeze the classifier version + a labelled gold-set for drift; per-issue as-of/vintage convention (SIPRI annual, World Bank lagged, UNHCR frequent).
• Data provenance: "classified" = LLM-categorized PUBLIC posts (no secret material; fully auditable). Refugee-return figures are an external UNHCR indicator (dated), not a model output.
• Scope: a state-to-state instrument; non-state / sub-state actors are a separate layer, out of scope by design.
• Positioning: the first Syria-specific instrument of its kind.

Provenance / firm-up targets

Real data: GDP, SWF, refugee-stock, diplomatic presence, UNGA votes, deal-magnitudes (curated 2026-07-09), population (World Bank SP.POP.TOTL, verified 2026-07-09 — 194/194 matched, zero material deviations from encoded values; `wb_population.json` + re-runnable `spi_pivot_verify.py`). Analyst-set, documented: willingness A (♦ evidence-cited corrections), regional anchors (`regional_anchors.json`, 80 scored, scale documented). Dv/Se/Cu rank-verified against source datasets (2026-07-10, `dv_se_cu_verification.json`): Se vs SIPRI Milex 2025 ρ=0.847 (n=146; deviations by design — Syria-relevance weighting) · Dv vs GHS 2021 ρ=0.77 (n=179) and OECD DAC donors ρ=0.61 (n=40) · Rf/diaspora vs UN DESA migrant stock 2024 ρ=0.944 (Syrian-born hosts; LatAm descent-vs-refugee design validated). Brand Finance Soft Power 2025 = gated (sole unverified sub-input; spot-checked). No score corrections warranted. Readiness rebuilt from the ministry's official roster (`mofaex.gov.sy/diplomatic-missions` → `syrian_missions_official.json`: 44 embassies + 7 consulates-general incl. news-verified Gaziantep/Jeddah; ladder: embassy 0.85 · consulate-only 0.60 · +0.15/consulate · cap 1.0 · floor 0.45). Open v2 upgrades: the external triangulation stream (GDELT/ICEWS), NLP posture, MFA staffing data.