Pakistan’s role as mediator in the US–Iran–Israel war produced a clear, measurable improvement in the quality of democratic discourse on X-platform at home, focusing on empathy, civility, trust and agency. Over a seven-week observation window, the national Democracy Discourse Index (DDI) composite rose from 56.1% during conflict entry to 62.6% during the Islamabad Talks, representing a 6.5-point gain. The strongest movement came in Civility (66.9%) and Trust Language (67.9%), showing that diplomatic credibility can quickly improve the public tone of politics and increase confidence-bearing language around public institutions. But the improvement had limits. Democratic Agency remained the weakest and least responsive dimension (55.2%). In practical terms, citizens spoke more respectfully about politics and institutions when Pakistan was seen as a credible mediator, but they did not consistently speak as democratic actors able to shape outcomes. The implication for policymakers and diplomats is that constructive diplomacy can generate discursive legitimacy, but it does not automatically build democratic participation. The Pakistan pilot shows both the value and fragility of the “diplomacy premium.” Gains in public discourse are strongest when diplomacy is institutionally framed, transparently communicated, and reinforced by domestic accountability. The DDI therefore offers policymakers an early-warning tool to detect, in near real time, whether political events are strengthening democratic discourse or merely improving its surface tone. In Pakistan’s case, the signal is clear. Diplomacy improved the quality of public language, but durable democratic gains will require stronger civic agency and protections for civic space.
1. Introduction
Between March and May 2026, Pakistan assumed an unprecedented diplomatic role: mediating between the United States, Iran, and their regional allies. Islamabad hosted a quadrilateral summit with Turkey, Saudi Arabia, and Egypt (29 March), facilitated direct US–Iran negotiations led by Vice President J.D. Vance (13–14 April), and maintained back-channel contact through Prime Minister Shehbaz Sharif and Field Marshal Asim Munir even after the first round ended without a comprehensive agreement. The episode is, by any measure, the most consequential foreign-policy moment of Pakistan’s last decade — and standard democracy indices would not be expected to register it for months, if at all.
Standard democracy indices register what has already happened to institutions. The Democracy Discourse Index (DDI), developed by the Global Centre for Rehumanising Democracy, Sensika Technologies, and a founding consortium of seven universities — including NED University of Engineering & Technology in Pakistan — is built precisely to fill that gap. It treats the quality of public discourse as a proxy indicator of democratic health, reading civic dialogue as democracy’s vital signs and coding discourse across dimensions of empathy, civility, trust, and agency in near real time. In doing so, it reveals shifts in civic resilience and erosion often months before conventional democratic metrics can detect them.
The DDI is a hybrid human-AI infrastructure anchored in Critical Discourse Analysis and deliberative-democracy theory. It treats language as a social practice that constructs power relations and treats the quality of public conversation as both an indicator and a determinant of democratic health. It combines frontier multilingual AI with systematic human coding to track empathy, civility, trust and democratic agency in public discourse.
Trained researchers provide the ground-truth discourse judgements that teach and validate the AI, ensuring the system’s judgements are grounded in human expertise and experience. The DDI is catalogued in the Doha Solutions Platform, the United Nations Department of Economic and Social Affairs (UN DESA) repository of innovative initiatives advancing the SDGs, where it is mapped to SDG 16 (Peace, Justice and Strong Institutions).
Pakistan pilot. Nearly 3,000 posts coded across five weekly batches by 36 student-researchers at NED University of Engineering & Technology, under Principal Investigator Prof. Dr. Wajiha Raza Rizvi. Each post is scored on four dimensions: Empathy (D1), Civility (D2), Trust Language (D3), Democratic Agency (D4).
Learn more: discourseobservatory.org/pakistan · gcrd.org.uk/ddi
2. Methodology
The Democracy Discourse Index (DDI) is a hybrid human-AI discourse measurement framework grounded in Critical Discourse Analysis, deliberative-democracy theory, and computational content analysis. It treats public language as a democratic signal: not merely as an expression of opinion, but as a social practice through which trust, legitimacy, conflict, belonging, and civic agency are constructed.
For the Pakistan pilot, discourse quality was measured across four dimensions:
- Empathy: the degree to which discourse recognises the humanity, suffering, perspective, or legitimate concerns of others.
- Civility: the extent to which political disagreement is expressed without dehumanisation, abuse, intimidation, or rhetorical escalation.
- Trust Language: the presence of confidence-bearing, good-faith, institutionally oriented, or epistemically responsible language.
- Democratic Agency: the extent to which citizens are represented as capable of shaping political outcomes through participation, collective action, rights-claiming, or institutional engagement.
Each dimension contains five indicators, producing twenty indicators in total. Each indicator is scored on a 0–4 ordinal scale, where 0 indicates the absence of the relevant democratic-discursive quality and 4 indicates a strong or exemplary presence. Per-post dimensional scores are calculated as the mean of the five indicators within each dimension and rescaled to a 0–100 scale.
The Pakistan pilot analysed 2,950 public posts from Platform X, collected between 12 March and 28 April 2026. Posts were randomly selected within topic taxonomies, organised into five weekly batches, and mapped to major phases in Pakistan’s diplomatic role during the US–Iran–Israel crisis. Each post was independently coded by three trained student researchers at NED University of Engineering & Technology, under faculty supervision. Disagreements greater than one ordinal point were flagged and resolved through faculty-supervised adjudication before inclusion in the final analytic dataset.
Intercoder reliability was assessed using percentage exact agreement and benchmarked against the Krippendorff’s α ≥ 0.70 threshold: 0.80+ was treated as strong, 0.70–0.79 as acceptable, 0.67–0.69 as borderline, and scores below 0.67 as requiring interpretive caution. Reliability was highest in the earlier batches and declined as the discourse grew more complex and contested — consistent with the greater interpretive difficulty of coding stalled negotiations, military-personality framing, and civic-space restrictions.
3. The diplomatic timeline, in five phases
The Pakistan pilot organises the seven-week observation window into five temporal phases mapped to identifiable diplomatic events. The DDI composite tracks this sequence:
- Phase 1 — Conflict Entry (8–15 Mar · composite 56.1%). Pakistan-Afghanistan cross-border strikes (Operation Ghazab-ul-Haq), the Iran-US-Israel war emerges, Pakistan Navy begins Strait of Hormuz escort of merchant vessels. The dataset floor War & Conflict Agency reads 32.8%: the lowest agency reading anywhere in the sample.
- Phase 2 — Mediation Emerges (25–30 Mar · 59.1%). Foreign Minister Dar confirms both Washington and Tehran requested Islamabad’s facilitation. Pakistan hosts a quadrilateral summit with Turkey, Saudi Arabia and Egypt. Social Cohesion peaks at 67.9% composite; Empathy hits 73.8% — its highest reading for any major topic across all five batches.
- Phase 3 — Islamabad Talks (11–16 Apr · 62.6%). Direct US–Iran negotiations in Islamabad. Dataset peak. Governance Civility reaches 68.6%: the highest single-batch civility score for any major topic in the dataset; Trust Language reaches 67.9%. When the talks stall, however, War & Conflict Agency drops back to 36.3%.
- Phase 4 — Military-Centric Framing (21–26 Apr · 60.4%). The hashtag #CDFGlobalPeaceMaker dominates. Field Marshal Munir’s Iran visit reframes diplomacy as personal military achievement. Governance composite drops 5.6 points: its steepest single-batch decline. ISPR campaigns and propaganda pressures erode civic space.
- Phase 5 — Collapse and Reckoning (28 Apr–4 May · 60.6%). A 21-hour Islamabad summit collapses. Petrol hits Rs. 393.35/litre. Social Cohesion crashes to 51.9% (Civility 50.0%, Agency 49.9%). Sedition charges against journalist Fozia Baloch. Counterweights: Asif Ali Zardari’s clean-energy visit to China; Sobia Khan’s appointment as Pakistan’s first transgender jail warden.
4. What the data show: peacemaking moves the discourse
Across the five batches, the DDI traces a clear arc: a sharp climb as mediation emerges and Islamabad becomes the venue for direct US–Iran talks, followed by a partial reversal once diplomatic momentum stalls and the framing shifts toward military personalities. The composite rises 6.5 points between Phase 1 and Phase 3 — a meaningful movement on a scale that, in cross-national observatory work, typically shifts by less than 2 points week-to-week.
This is analytically important because the strongest gain arrives when Pakistan is perceived to possess the strongest diplomatic efficacy. The composite rises from 56.1% to 59.1% as mediation emerges, then to 62.6% during the Islamabad Talks. The subsequent decline to 60.4% under military-centric framing indicates that discourse quality is sensitive not only to what the state does, but to how political achievement is narrated.
Disaggregating by dimension reveals which kinds of discourse are responding. Civility and Trust Language do most of the work: respectful registers and confidence in institutions surge as Pakistan moves into the mediator’s seat, with both indicators crossing the 67% threshold during the Talks. Empathy follows more cautiously. Democratic Agency — the dimension that captures whether citizens frame themselves as actors capable of shaping outcomes — moves least, and never crosses 56% in any batch.
The Civility–Agency gap is the core democratic warning in the data. It begins at 13.7 points in Week 2, narrows to 9.9 points during early mediation, widens again to 11.7 points during the Islamabad Talks, and remains above 10 points through Weeks 6 and 7. Pakistan’s public sphere can become more civil without becoming more democratic in the participatory sense.
4.1 National DDI dimensional scores by batch
| Batch | Posts | D1 Empathy | D2 Civility | D3 Trust Lang. | D4 Agency | Composite |
|---|---|---|---|---|---|---|
| B1 (Wk 1,2) | 600 | 55.9% | 60.6% | 60.9% | 46.9% | 56.1% |
| B2 (Wk 3,4) | 550 | 58.1% | 62.5% | 63.3% | 52.6% | 59.1% |
| B3 (Wk 5) | 600 | 60.3% | 66.9% | 67.9% | 55.2% | 62.6% |
| B4 (Wk 6) | 600 | 58.3% | 65.1% | 63.4% | 54.8% | 60.4% |
| B5 (Wk 7) | 600 | 58.7% | 65.5% | 63.9% | 54.4% | 60.6% |
| Cumulative | 2,950 | 58.2% | 64.0% | 63.8% | 52.8% | 59.7% |
4.2 Topic-level highlights
| Topic | Reading | When | What it shows |
|---|---|---|---|
| Governance | 68.6% | B3 Talks · Civility | Highest single-batch civility for any major topic. |
| Governance | 57.1% | B4 Military-centric | Steepest single-batch decline (−5.6 pp). |
| Social Cohesion | 67.9% | B2 Mediation · Composite | Topic peak — Empathy hits 73.8%. |
| Social Cohesion | 51.9% | B5 Collapse · Composite | Topic trough — 16-point fall from B2. |
| War & Conflict | 32.8% | B1 Conflict · Agency | Lowest agency reading anywhere in the dataset. |
| War & Conflict | 36.3% | B3 Talks stall · Agency | Fatalism returns the moment talks lose momentum. |
5. The fragility: solidarity tied to outcomes, not values
The Social Cohesion topic peaks at 67.9% in Phase 2, then collapses 16 points to 51.9% in Phase 5 as talks fail — the single largest peak-to-trough movement in the dataset. The solidarity generated during mediation was real, but tied to the anticipated outcome of a successful Pakistan-brokered de-escalation, rather than to internalised civic values.
Across all posts, Civility averages 64.0% while Democratic Agency averages 52.8% — an 11.2-point gap that opens before the diplomatic episode and persists through it. Pakistanis, in this sample, talk respectfully about politics. They do not consistently talk as if they shape it.
6. Interpreting the paradox: diplomacy, legitimacy and democracy
The Pakistan case shows that diplomatic performance can function as a source of discursive legitimacy — consistent with a broader theory of performative legitimacy. However, the same data indicate democratic thinness. The improvement is strongest where the discourse concerns institutional competence and weakest where it concerns citizen agency. This matters because a democracy can enjoy moments of legitimating performance while still failing to deepen the participatory imagination of its citizens. The Pakistan case should not be read simply as evidence that diplomacy improves democratic discourse. It should be read as evidence that diplomacy improves some dimensions of democratic discourse while leaving others structurally underdeveloped.
The military-civilian framing shift is especially significant: the Governance composite falls from 62.7% to 57.1% when diplomatic success is reframed around military personalisation. Process, mandate and civilian accountability matter. Where diplomacy is narrated as personal or military charisma, the democratic quality of governance discourse weakens even if the diplomatic activity itself continues.
The data also confirm the discourse costs of civic-space violations. An ISPR campaign targeting a Baloch woman without evidence (24 April) depresses Governance, Social Cohesion and Civic Activism scores in its batch. The subsequent Fozia Baloch sedition charges (2 May) and Sheema Kirmani’s detainment (5 May) fall outside the coding window but are consistent with the civic-space pressures the data document. International peacemaking and domestic rights violations cannot co-exist without measurable discourse costs.
7. Policy Recommendations
Anchor diplomatic achievement to institutional frameworks — parliamentary authorisation, foreign-ministry briefings, multilateral mandates — rather than to individual figures. Governance discourse quality is highest when process drives the framing and lowest when personality does.
War & Conflict Agency at 48.1% cumulative — and 36.3% when talks stall — is the most urgent intervention target. Position peacebuilding as citizen-centred: public deliberation on negotiation terms, ceasefire monitoring, rights-based framing. A target of Agency above 50% in this topic is realistic and meaningful.
Standing fact-checking desks for diplomatic reporting, professional certification for diplomacy correspondents, and rapid-response verification protocols are concrete answers to the structural media vulnerability revealed in the data.
Calls to suppress dissent peaked precisely during Pakistan’s most sensitive diplomatic episodes. Civilian institutions should issue enforceable commitments that diplomatic sensitivity does not justify suppression of journalism, rights activism, or minority voices. The DDI cost of these episodes is measurable and should be reported transparently.
Pakistan cannot sustain discourse-quality gains while operating information campaigns against its own citizens — including blocking access to X (formerly Twitter) for several months. Frank dialogue about domestic accountability is best framed not as conditionality but as a shared interest.
8. Conclusion: discourse as democracy’s vital signs
Pakistan’s role in the 2026 US–Iran peace process produced the most significant measurable improvement in democratic discourse quality in the seven-week window. A 6.5-point composite gain during the Islamabad Talks, anchored by Civility at 66.9% and Trust Language at 67.9%, is empirical evidence that constructive diplomacy has a real, positive effect on how a nation’s public converses about its institutions. But the 16-point collapse of Social Cohesion that followed is sobering: solidarity tied to outcomes does not survive their absence.
Traditional democracy indices register what has already happened; the DDI registers what is happening and, where validated, what is coming. Pakistan’s pilot demonstrates that real-time, dimensional discourse measurement can identify where diplomatic capital strengthens democratic conversation, where it does not, and which structural features of the public sphere most urgently require investment. That is a capability the policy community has not previously had.
