Less than two months ago, Emmanuel Macron defeated Marine Le Pen by more than seventeen percentage points and became the first French president in twenty years to win re-election while still in office.
Tonight, France is voting again, and the political question is completely different.
In April, voters were asked who should be president. This month, they are choosing the 577 members of the National Assembly who will determine how much of the president’s programme can actually become law.
The distinction matters because winning the presidency and winning permission to implement an entire political programme are not the same thing.
When Macron won in April, I argued that his 58.55% should not automatically be interpreted as 58.55% support for Macronism. Some people enthusiastically supported him. Others primarily wanted to prevent Marine Le Pen becoming president. Some Mélenchon voters reluctantly chose Macron in the second round. Others abstained or cast blank ballots.
The presidential election compressed all those motivations into one result.
The legislative election begins separating them again.
Jean-Luc Mélenchon has understood this better than almost anyone else.
After finishing third in the presidential election, he rapidly brought together much of the French left into the New Ecological and Social People’s Union, or NUPES. His France Unbowed movement reached agreements with the Socialists, Greens and Communists to avoid competing against one another in many constituencies.
The strategy is almost the reverse of what happened during the presidential election.
In April, the fragmented left divided its first-round vote among several candidates.
For the parliamentary election, much of it has coordinated.
That demonstrates something important about electoral systems: political unity is partly endogenous to the rules through which votes become power.
Parties do not cooperate simply because they suddenly discover ideological agreement.
They cooperate when the electoral system makes cooperation valuable.
France elects members of the National Assembly through 577 constituency contests using a two-round system. That means parties competing for similar voters can damage one another if they fragment support across too many candidates.
NUPES attempts to solve that coordination problem.
The Greens, Socialists, Communists and Mélenchon’s movement continue disagreeing about significant issues.
But they agree that competing separately would probably produce fewer seats.
Coalition therefore begins before the election rather than afterwards.
This is politics responding rationally to institutional incentives.
The first round last Sunday demonstrated how competitive the new alliance has become. Macron’s Ensemble coalition and NUPES finished extremely close nationally, turning today’s second round into a genuine contest over whether the president will retain a comfortable parliamentary majority.
That is remarkable considering Macron appeared so dominant only eight weeks ago.
But perhaps it should not be surprising.
Presidential elections ask voters to compare individuals.
Legislative elections ask them to distribute power.
Those are different decisions.
A voter can reasonably believe Macron is preferable to Le Pen as president while simultaneously believing Macron should face a strong parliament capable of constraining him.
There is no contradiction.
In fact, divided government can be an intentional voter preference.
Political commentary often assumes voters want whichever party they support to possess as much power as possible.
Sometimes they do.
But citizens can also prefer institutional balance.
A voter may trust Macron on foreign policy and European leadership while disliking his pension proposals.
They may prefer him to Le Pen while wanting stronger environmental policy.
They may believe his economic reforms are broadly useful while worrying that a large presidential majority makes parliament too weak.
Voting for Macron in April and NUPES in June can therefore be perfectly coherent.
The voter is not changing their mind.
They are answering a different question.
This distinction is important because democratic elections are often interpreted too broadly. Politicians win one contest and describe the result as a mandate extending across every policy issue.
But elections rarely provide that kind of precision.
A vote is an extremely information-poor political signal.
One mark on a ballot has to represent opinions about taxation, pensions, climate change, immigration, Europe, healthcare, education, foreign policy and the personality of the candidate simultaneously.
Politicians then interpret the mark afterwards.
Naturally, they tend to interpret it in ways favourable to themselves.
Macron can say France endorsed his programme.
Mélenchon can say many Macron voters merely rejected Le Pen.
Le Pen can say the entire establishment coordinated to prevent her victory.
Each interpretation contains some truth and considerable political convenience.
Repeated elections provide additional information.
France is effectively conducting a second measurement.
April measured presidential preference.
June measures parliamentary preference.
If the two results differ substantially, that does not mean democracy has produced contradictory instructions.
It means citizens possess multidimensional preferences.
This is where France’s semi-presidential system becomes particularly interesting.
The president possesses substantial authority, especially in foreign and defence policy, but the government still depends on the National Assembly.
If an opposition bloc controlled parliament, France could enter cohabitation, where the president and prime minister come from opposing political camps.
That has happened before under the Fifth Republic.
Cohabitation demonstrates that executive authority itself can be divided.
The same electorate can produce one political preference for the presidency and another for parliament, and the constitutional system then has to reconcile them.
That looks inefficient compared with a system in which one election hands one party complete control.
It can also be understood as a safeguard against interpreting one victory too expansively.
Power becomes harder to monopolise because different institutions possess separate sources of legitimacy.
The United States institutionalises this even more explicitly through separately elected presidents, House members and senators.
France does it differently, but the underlying principle is similar.
Democratic legitimacy is distributed.
This creates checks.
It can also create paralysis.
That trade-off becomes particularly important if Macron loses his absolute majority.
His first term was facilitated by a large parliamentary bloc capable of supporting government legislation. Without 289 seats, the threshold for an absolute majority in the 577-seat Assembly, his government would need to negotiate more frequently with other parties.
That could make policy slower.
It could also make parliament more meaningful.
Efficiency and representation pull in opposite directions again.
A legislature that automatically approves the executive’s programme is efficient.
It may not provide much legislative scrutiny.
A fragmented parliament forces negotiation.
It may also struggle to pass necessary reforms.
There is no institutional design that maximises both speed and constraint.
This is why complaints about political gridlock need context.
Sometimes gridlock means institutions are failing.
Sometimes gridlock means society genuinely disagrees.
If voters distribute seats among several political blocs, difficulty passing legislation can accurately represent the difficulty of constructing social consensus.
The legislature is not necessarily malfunctioning.
It may be displaying the disagreement voters created.
NUPES itself demonstrates the complications this creates. Mélenchon has framed the legislative election almost like a third round of the presidential election, urging voters to make him prime minister.
Constitutionally, French voters do not directly elect the prime minister.
The president appoints one.
But a parliamentary majority hostile to the president can effectively force the president to appoint someone capable of commanding that majority.
Mélenchon’s slogan therefore converts a complicated institutional process into a simple electoral proposition.
That is politically clever.
It also personalises a parliamentary election around another leader.
This is increasingly characteristic of French politics.
Macron’s coalition is organised heavily around Macron.
NUPES is strongly associated with Mélenchon.
The National Rally remains strongly associated with Le Pen.
The collapse of the old party system has not eliminated political organisation.
It has reorganised organisation around personalities.
That may weaken the intermediary institutions that once translated social interests into government.
But NUPES also demonstrates something moving in the opposite direction.
The Socialist Party, Greens, Communists and France Unbowed have had to negotiate an actual programme and candidate allocation.
Coalition-building forces institutionalisation.
Mélenchon cannot simply rely on personal popularity if several parties need to cooperate across hundreds of constituencies.
The electoral system is therefore simultaneously personalising and institutionalising politics.
That apparent contradiction is useful because political systems rarely move in only one direction.
The same rules create multiple incentives.
There is also a generational and ideological dimension to the left’s revival. The catastrophic presidential result for the Socialist Party did not mean left-wing voters disappeared.
They moved.
Mélenchon’s strong first-round presidential performance already demonstrated this.
NUPES now attempts to convert those voters into parliamentary power.
Again, this supports the argument from April: party collapse is not the same thing as ideological collapse.
An institution can die while its electorate survives.
The voters then search for another organisation.
France’s traditional Socialist Party became too weak to organise much of the left.
Mélenchon built a new structure around voters who still support redistribution, environmental policy, stronger public services and opposition to parts of Macron’s economic agenda.
The political demand remained.
The supplier changed.
This is another reason established parties should be careful about treating historical constituencies as permanent property.
Australia’s Liberals learned the same lesson from the teal independents.
Voters are increasingly willing to unbundle political loyalties.
France demonstrates it at a much larger scale.
The National Rally is also important in the legislative election, although France’s electoral system historically makes it difficult for Le Pen’s party to convert national vote share into proportional parliamentary representation.
This is where electoral-system design becomes politically consequential.
A party can receive millions of votes nationally while winning relatively few constituencies because its support is distributed inefficiently or because other voters coordinate against it in second rounds.
That creates a representation problem.
A legislature can accurately reflect geography while inaccurately reflecting national vote shares.
No electoral system represents everything perfectly.
Proportional representation maximises one kind of representativeness.
Constituency systems maximise another.
France’s system attempts to produce local representatives while allowing second-round consolidation.
The result can reward alliances and punish isolated parties.
Again, institutional rules shape political strategy.
Le Pen’s movement has spent years trying to overcome that structural disadvantage by normalising itself enough that voters from other parties stop automatically coordinating against it.
The legislative election therefore provides another measure of whether that strategy is working.
This is why elections should be analysed as systems rather than simply national vote totals.
A vote becomes political power only through institutional rules.
Twenty per cent under proportional representation means roughly twenty per cent of seats.
Twenty per cent under a constituency system could mean dramatically more or less.
The same public preference can therefore create different governments depending on constitutional design.
That is not necessarily undemocratic.
Every electoral system has to decide what it wants representation to mean.
The important thing is that citizens understand the rules and that the rules remain sufficiently stable that politicians cannot rewrite them opportunistically whenever outcomes become inconvenient.
France’s parliamentary election also arrives during an unusually difficult international environment. Inflation is rising, energy prices remain high, Russia’s war in Ukraine continues and the European Union faces enormous economic and strategic uncertainty.
Macron argues that France needs stable government to respond effectively.
That is a strong argument.
But stability cannot simply mean giving the president whatever majority makes governing easiest.
Democratic legitimacy requires allowing voters to decide how much constraint they want.
This is the fundamental difference between technocratic and democratic reasoning.
A technocrat might ask: Which institutional arrangement allows the government to implement the optimal policy most efficiently?
Democracy asks another question first: Who gets to decide what counts as optimal?
If voters want stronger redistribution or slower pension reform, the fact that economists or ministers consider another policy more efficient does not make the electoral preference irrelevant.
Expertise can explain consequences.
It cannot replace consent.
This is where Macron’s second term may become fundamentally different from his first.
In 2017, he entered office with novelty, momentum and a large parliamentary majority.
In 2022, he has been re-elected but faces a much more fragmented political environment.
That may force a different style of government.
Negotiation rather than command.
Coalition-building rather than automatic majority.
Compromise rather than programme implementation.
Macron may find that frustrating.
It may also be what French voters are asking for.
This brings us back to the distinction between authority and mandate.
Macron unquestionably possesses presidential authority after April.
The scope of his governing mandate is being determined again today.
That is not a weakness of democracy.
It is democracy refusing to answer every political question through one election.
Different institutions ask different questions.
Different elections produce different answers.
The political system then has to reconcile them.
We do not yet know the final composition of the Assembly as I write this tonight, and the exact seat distribution will determine how difficult Macron’s second term becomes.
But the broader lesson is already visible.
In April, France chose Macron over Le Pen.
That did not necessarily mean France chose Macron without constraints.
A democracy can elect a leader and still decide that the leader should have to negotiate.
Perhaps that is one of the healthiest distinctions voters can make.
Choosing who governs does not require choosing how easily they should be allowed to govern.
