Report #02
Pakistan Civic Education · Democracy & Civil–Military Relations

Win — But Not Too Much

The real lever in Pakistani politics is not who is allowed to win an election — but how strong a winner is ever allowed to become.

In one line

Stop asking whether the establishment ‘brings’ a leader to power, and start watching the size and dependency of the mandate it permits. Do that, and a pattern appears across 2018, 2022 and 2024 that no single election can reveal.

1A sharper question

The usual debate runs in a circle: did the establishment bring Imran Khan, or did he earn his support? The earlier report showed the honest answer is both. This report asks a sharper question — one that explains far more once you see it: not who is allowed to win an election, but how much they are allowed to win.

Once you watch the magnitude of a mandate rather than only its owner, a pattern emerges that a single result can never show you — and it applies not just to Imran Khan, but to almost every civilian leader Pakistan has produced.

2The core idea: control the magnitude, not the winner

To an unelected power, a strong winner and a weak winner are not two sizes of the same thing. They are opposite kinds of problem. So the real lever is not victory or defeat — it is the size and dependency of a victory.

Two winners, two completely different problems
A commanding mandateA managed (thin) mandate
What the PM can doAmend the constitution; assert civilian supremacyBarely pass a budget without permission
Who they depend onNo one — governs in their own rightSmall coalition allies who can be influenced
How removableVery hard to dislodgeOne withdrawal of support away from falling
Use to an unelected powerA rival to be fearedA partner that can be steered — or dropped

So the most useful question is not “who did they back?” It is: who were they willing to let become strong — and who did they keep weak?

3Why a thin coalition is a leash

The mechanism is simple once named. You do not need to defeat a popular leader to control him. You only need to ensure his win is just short of a majority, so that governing requires partners — and the partners are people you can reach.

Real popularityA genuine wave you did not create
Capped at a pluralityLargest party, but short of a majority
Forced into coalitionNeeds small allies to govern
Allies are leverageKing’s-party partners can be influenced
Removable on demandPull the allies, and the government falls

The dependency is not a side-effect of the win. For a managed mandate, the dependency is the point.

42018 — the friendly cap

When the relationship is warm, magnitude-control looks like quiet fine-tuning rather than tampering.

Largest party
PTI won more seats than anyone else…
Short of a majority
…but not enough to govern alone
Coalition-dependent
Took office leaning on small allied parties

The documented facts: PTI emerged the largest party with 116 of the 272 directly elected National Assembly seats, short of the 172 needed for a majority in a house of 342, and took office through a coalition with the MQM-P, the PML-Q and other smaller parties; and the Election Commission’s Result Transmission System (RTS) failed on election night, during the count. The interpretation — that the margin was deliberately held down to keep him dependent — is a contested allegation, not a proven fact.

The fair counter-view: Pakistan’s geography produces a plurality, not a majority, for any party whose vote is regionally uneven — so falling short needs no conspiracy. And an RTS crash is as consistent with ordinary technical failure as with design. What is not in dispute is that the blackout destroyed transparency at the worst possible moment.

52022 — pulling the thread

A leash only matters when it is pulled. The dependency built into a thin coalition is also its removal switch — and in 2022 it was flipped.

Built-in dependencyA government leaning on small allies
Friction growsThe PM asserts more independence
Allies peel awayCoalition partners switch sides
No-confidence voteThe numbers collapse
RemovedOusted by 174 votes, 10 April 2022

The vulnerability created in 2018 became the method of removal in 2022.

The fair counter-view: a no-confidence motion is a constitutional, opposition-driven process — the Supreme Court ruled on 7 April 2022 that blocking the vote had been unconstitutional and ordered the Assembly to hold it — and PTI had genuinely alienated some of its own allies: the Jamhoori Watan Party, most of the Balochistan Awami Party and the MQM-P all left the coalition in the last days of March 2022. PTI calls the removal establishment-engineered; others see ordinary parliamentary politics. Both readings draw on the same events.

62024 — the hostile cap

By 2024 the relationship had reversed — but the objective did not change. The goal was still to keep him short of a commanding mandate. Only the method turned crude, because the pressure now ran against the public mood rather than with it.

Same objective, cruder instrument
What happenedThe effect
Party dismantled, leaders jailedThe machine for converting votes was broken up
Election symbol strippedCandidates forced to run as scattered independents
Public backs PTI anywayPTI-backed independents become the largest bloc
Disputed count (Form 47)Largest bloc, yet denied a majority — a rival coalition governs

The fair counter-view: the Election Commission disputes the rigging characterisation and points to the tribunals, which by April 2026 had dismissed 242 of the 246 post-election petitions they had decided; the specific seat figures are contested estimates. What stands without dispute is the shape of the result: the largest bloc was kept out of power — and once the Supreme Court’s Constitutional Bench set aside, on 27 June 2025, the July 2024 judgment awarding PTI the reserved seats, the coalition that had kept it out held a two-thirds majority.

7The pattern across all three

Place the three moments in a row and the lesson that no single election can teach you appears at once.

One constant goal, three different instruments
201820222024
RelationshipAlignedSouringHostile
MethodQuiet cap to a pluralityCoalition pulled awayDisputed count (alleged)
How visibleLow — looked organicMedium — looked parliamentaryHigh — looked like tampering
The objectiveNo commanding mandateNo commanding mandateNo commanding mandate
  • The lever is magnitude, not identity. The decisive question was never just who won — it was how strong any winner was permitted to be.
  • A built-in dependency is a built-in removal switch. The thin 2018 coalition and the 2022 ouster are the same fact seen at two moments.
  • The objective survived a total reversal of the relationship. An aim that holds whether a leader is friend or enemy is structural, not personal.
  • This pattern predates Imran Khan. Zulfikar Ali Bhutto was removed by General Zia’s coup in 1977; the governments of Benazir Bhutto and Nawaz Sharif were dismissed by presidents invoking Article 58(2)(b) in 1990, 1993 and 1996; and the comfortable two-thirds majority Nawaz Sharif won in 1997 ended in General Musharraf’s coup on 12 October 1999. The names change, the limit does not.

8How to think about this differently

The payoff is a habit of mind, not a verdict on one man.

  • Generalise beyond one leader. Watch how the strong mandates of every party have fared over time, not just the one you support or oppose.
  • Separate record from interpretation. The coalition maths, the RTS failure, the no-confidence vote and the 2024 seat count are on record; that magnitude was deliberately controlled is an argument built on them, credible but not proven.
  • Beware the theory that fits every outcome. If strong-win, thin-win and loss can all be explained by the same hidden hand, the theory must be held loosely.
  • Always ask who benefits from the telling. ‘He was leashed from day one’ and ‘he was a free agent betrayed’ each serve someone’s politics.

This report is an educational explainer for the general reader. It summarises documented events and clearly-attributed allegations as of early 2026, and does not endorse any political party. Where claims are disputed, that dispute is noted in the text. Contested figures are estimates, not settled facts.