Lula’s victory illustrates the challenge to Marxist strategy: his victory defeats the far right, yet his government creates the conditions for it to flourish. In a two-round election against the far right, should the radical left always back the reformist candidate from the first round? Ninety years of the movement’s own press give no simple answer about the ballot. They give a much clearer one about what happens next. The cases where the left lost its independence are mostly cases where support for a candidate became support for a government. And the committees that elections called into being rarely outlived the count.
TL;DR
The question came from Brazil this year. PSOL, the broad socialist party that contains the Brazilian section of the Fourth International, backed Lula from the first round. Some of the revolutionary currents argued for a first-round vote to the left of PSOL, then a Lula vote in the runoff. The question is old, and the movement’s papers have argued it many times.
The tradition’s default is clear. Stand on your own programme in round one, in settings where that is practical, and vote against the right in round two. Janette Habel wrote in 1981 that “nothing prevents each party standing its own candidate on its own programme in round one”. The FI’s 1971 resolution on Chile ruled out electoral support for “a front that includes petty-bourgeois and bourgeois parties”.
The case for backing the reformist from round one is real, and it should be given its due. Isolation has cost the far left before, and one of the LCR’s central leaders later took his share of responsibility for it. Squeeze votes cut independent candidates down whatever they decide. Sometimes no credible anti-capitalist candidate exists at all.
The vote itself mattered less than what followed it. The documented losses come where first-round support led on to ministers, or to a seat in the government’s parliamentary base: Brazil 2003–05, Brazil since 2023, Chile 2022–25, Spain 1936.
Elections are a poor way to build self-organisation. Chile’s 15,000 Popular Unity committees were, in Robin Blackburn’s phrase in Red Weekly, “put into cold storage”. The organs that lasted came out of struggle, not out of campaigns.
So the useful line is not “whom”, or “how” alone, but “and then what”. Vote against the far right. Refuse governmental responsibility for governments that attack workers. Organise those who draw the lesson. Where the evidence is thin, say so.
The question
The scenario is a familiar one, from Chirac to Clacton. A far-right candidate can win. The main opponent is supported by a coalition that includes openly bourgeois parties. To its left stands a broad anti-capitalist party with organised revolutionary currents inside it. In the runoff, every current calls for a vote against the far right. The argument is about round one, where there are three choices:
the party stands its own candidate;
it backs the reformist from the start and keeps “programmatic independence” through its legislative candidates; or
revolutionaries ‘break’ (either from the discipline of the coalition or from the ‘common sense’ of the broad vanguard) to vote for marginal revolutionary candidates.
One correction to that picture first. Not every current calls for a vote against the far right in the runoff. Lutte ouvrière did not do so in 2002, 2017 or 2022. In 2002 its line was « Pas une voix pour Le Pen ! », but « une enveloppe vide en guise de vote pour Chirac » (”not one vote for Le Pen, but an empty envelope as a vote for Chirac”). In 2022 Nathalie Arthaud wrote « je voterai blanc ». Every organisation linked to the Fourth International that the archive covers did call for a vote against the far right, in one form or another.
What our tradition said
The Trotskyist classics say less about round one than people assume. They say a great deal about programmes.
In 1932 the German Social Democrats voted for Hindenburg in both rounds to stop Hitler. Trotsky’s verdict that August was that “the Social Democracy elected Hindenburg, who thereupon carried out a coup d’état aimed directly against it”. Hindenburg removed the SPD-led Prussian government in July 1932, and he appointed Hitler six months later. Trotsky did not say that the Communists should have voted Hindenburg. He charged the KPD with refusing a united front with the Social Democrats, so its separate candidacy gained “barely 300,000 votes” in a year of SPD discredit.
In 1936 he drew the line between a vote and a programme. He gave Labour no blank cheque, but he wrote that “wherever the I.L.P. was unable to run its own candidates, it should have supported a Laborite against a Tory”. He called the POUM’s signature on the Spanish Popular Front programme a betrayal. On electoral blocs, his rule was second-ballot agreements only: “No common ‘programs’, no common and permanent institutions, no renunciation of the criticism of temporary allies.”
The Fourth International applied that rule to Allende in 1971:
Revolutionists cannot participate in such a coalition even by offering it electoral support. (Revolutionary Marxists can, in certain situations, vote for a labor candidate but not for a candidate of a front that includes petty-bourgeois and bourgeois parties.)
The same resolution called for support for the government’s progressive measures and “a united front against the attacks of the reactionaries”. Electoral support and defence against reaction were treated as separate questions.
For a two-round system the plainest statement is Janette Habel’s, in French Inprecor in April 1981. She was attacking the Lambertists for calling a Mitterrand vote in round one:
l’élection présidentielle se joue en deux tours. Rien n’empêche chaque parti de présenter son propre candidat sur la base de son programme au premier tour.
(”The presidential election is played in two rounds. Nothing prevents each party standing its own candidate on its own programme in round one.”) In 1974 Alain Krivine put the same point more shortly: “no carte blanche and no blank check! That’s what we are demonstrating on the first round.” He then called for a vote for Mitterrand in round two as “a class vote”.
The case for round one
The other side has a serious case, and some of it comes from inside the FI’s own ranks.
Isolation is a real cost. In 2007 the LCR voted 60–40 to run Olivier Besancenot rather than join a unity candidacy. Its minority reported to the International that the League “found itself isolated: not only in broad sectors of its electorate, but also among the sympathizers and actors of the social movements”. François Sabado led the majority line. Fifteen years later he said: “we had refused to unite. I have my own responsibility in this… we did not see Mélenchon’s initiative coming.”
The squeeze happens anyway. Heloísa Helena polled around 12 per cent in Brazil in August 2006 and finished on 6.85, and PSOL’s own list took 1.4 per cent. In Portugal this January, Jorge Costa described a “desperate concentration of votes” on the Socialist candidate. The three left parties fell from 9 per cent to 4.5. Standing alone does not guarantee an independent profile at the ballot box.
Sometimes there is no one to vote for. In Peru in 2021 Corriente Amaru called for “a critical vote for Verónika Mendoza and JP in the first round” because “there is no anti-capitalist candidacy” and a runoff between two right-wing candidates was likely. That is a different situation from one where a credible left candidate is standing.
A vote is not an endorsement. Martín Mosquera argued in Argentina in 2023 for a vote that would block the far right, “but this is not the same as accepting the slippery slope of the logic of the ‘lesser evil’”. That is Trotsky’s own distinction from 1936, and it is a fair one.
What the record shows
The archive has one comparison that comes close to a controlled test. In Brazil in 1998 the PT’s national leadership imposed first-round support for the PDT candidate in Rio. In Rio Grande do Sul the two parties stood separately in round one and united in round two. A note in International Viewpoint at the time recorded the left’s argument:
The left had argued that standing separate candidates in the first round to argue for the two parties’ distinct programmes would not prevent both parties supporting the best placed candidate in the second round. This is exactly what happened in Rio Grande do Sul
The PT won Rio Grande do Sul with 50.78 per cent. In 1989 the PT stood Lula on its own programme in round one. In round two it accepted the support of bourgeois parties in a front which, as Juarez Guimarães wrote, « n’impliquait pas l’élaboration d’un programme commun ni la participation automatique au gouvernement, en cas de victoire » (”did not involve drawing up a common programme, nor automatic participation in government in the event of victory”). Democracia Socialista called the result a united-front dynamic. An editors’ footnote in Socialist Action, in the United States, called the same front “popular-frontist”. The two readings have argued with each other ever since.
On the other side of the ledger, the damage is documented, but it mostly comes after election day.
Brazil, 2003–05. Democracia Socialista backed Lula on a ticket with the right-wing Liberal Party, and one of its members, Miguel Rossetto, became a minister. Within a year the PT had expelled Heloísa Helena and others for voting against the pension reform, and PSOL was founded. In 2005 the FI’s International Committee concluded that holding office in the government was “contradictory with the building of an alternative in Brazil that is coherent with our programmatic positions”.
Brazil, since 2023. PSOL backed Lula from round one in 2022. In December that year its national directorate resolved that « o PSOL não terá cargos na gestão que se inicia » (”PSOL will hold no posts in the incoming administration”), while calling Sonia Guajajara’s appointment as a minister « uma conquista de extrema importância » (”a gain of the utmost importance”). Members took posts “in their own name”, and the parliamentary group sat in the government’s base. A year later Ana Cristina Carvalhaes of Insurgência reported that union leaders were telling her: “We’re not going to go on strike and weaken Lula”. She also reported that the FI’s own comrades in PSOL were “split in two”.
Chile, 2025. The left inside Boric’s governing coalition lost to Kast by 58 to 42. Karina Nohales reported that “all left-wing parties outside the governing coalition are legally dissolved”, in part for failing to unite a list of their own.
None of these is a pure test of the first-round vote. The vote and the government come bundled together, and the archive cannot separate their effects. What it does show is that the sources themselves treated the two decisions as separable. PSOL in 2018 supported Haddad in the runoff “while maintaining our political differences and preserving our independence”, and it did not enter any coalition. The PSTU in 2022 called a critical vote for Lula and added: « Não integraremos nem apoiaremos um eventual governo desses. Seremos oposição » (”we will neither join nor support any such government; we will be in opposition”). In 2024 the NPA-l’Anticapitaliste entered the New Popular Front on condition that it would not sit in an NFP government.
Not whom, but how?
I went into the archive with an argument of my own. In these elections, I thought, what matters is not whom you vote for but how you use the election: to build workplace and community self-organisation, and anti-capitalist poles, that can fight the “lesser evil” once it governs.
The evidence supports half of that. The other half it complicates, and I should say where.
The vote did matter, though often not in the way intended.
The LCR’s decision in 1988 to withdraw Krivine and back the dissident Communist Pierre Juquin produced 2.1 per cent. It ended with part of the League leaving for Juquin’s “Nouvelle Gauche” (both sides of that argument are in Critique communiste 81).
PSOL’s decision in 2022 cost it Plínio de Arruda Sampaio Jr’s current.
Vote choices split organisations, even when they moved few votes. I cannot call the question of whom secondary when it has broken organisations more than once.
Elections are a poor way to build self-organisation. The Popular Unity committees of 1970 numbered about 15,000. Robin Blackburn wrote in Red Weekly after the coup that they had been “created from top down as bodies appointed by, and accountable to, the Party bureaucrats”, and then “put into cold storage to await the next election”. The Juquin committees faded with his score. The French anti-liberal collectives of 2006 did not survive the decision to stand separate candidates. Of the New Popular Front’s local groups, Besancenot said this year that some remain, “but yes: overall, the alliance has collapsed”. The bodies that did last, the Chilean cordones industriales of 1972, came out of the bosses’ strike, not out of a campaign. Trotsky warned in 1935 that it is a mistake to think the masses can be called “at a set day and hour” to elect committees of action. The record bears him out, and it cuts against the hope I started with. The campaign is not where self-organisation is made.
Anti-capitalist poles grow in peaks, and keep little. The NPA was founded with more than 9,000 members, and by 2022 it had split in two. FIT-U’s Myriam Bregman polls between 9 and 14 per cent in Argentina this year, after the Front kept its own profile against Massa. It is too early to say whether that will hold.
The decision that most consistently shaped what came after was the one about government. Where the radical left took office, or sat in the government’s parliamentary base, the documented outcome is division and a narrower field for independent action. I find it hard to take a positive view of members becoming ministers “in their own name” while the party stayed in the government’s parliamentary base. But I have to set the other side of the record beside that. Staying out is not shown to bring gains in itself. Sinistra Critica after Prodi, and Popular Unity after Tsipras, stayed out and did not grow. Staying out is a condition of keeping independence, not a guarantee of anything.
So I would now put my argument differently. Not whom, and not how alone, but “and then what”.
And then what
That suggests a line that is less about round one than either side usually allows.
Vote against the far right in the runoff, and say what the vote is: a vote to stop something, not a mandate for anyone. In 2002 the LCR’s formula was “On May 5, vote against Le Pen!”, together with “Chirac is not a rampart against the far right”, and with the streets first. That is still the model. It was, as Alan Thornett wrote then, support “as the rope supports the hanging man”.
Treat round one as a tactic, decided by conditions. Where the reformist is certain to reach the runoff, the case for standing on your own programme is strong. It costs little and keeps a pole in public view: João Machado argued exactly this for Brazil in 2021. Where the danger is a runoff without the left, and no credible anti-capitalist candidate exists, the argument shifts, as it did in Peru in 2021. Neither choice is a principle, and neither will build anything by itself.
Refuse governmental responsibility for governments that attack workers. That means no ministers, “in their own name” or otherwise, and no seat in the parliamentary base. Vote for what helps workers. Vote against what hurts them, and say why in public. That is the test the unity of round one should be put to, and it is the one most often failed.
Organise those who draw the lesson, in struggle, not in the campaign office. The committees that matter are the ones workers build when they need them. An election can bring people together, and the job is to be useful to them when the next strike, occupation or eviction comes, not to keep a list of canvassers warm until the next poll.
Unity in action, a concrete test, independent organisation of those who draw the conclusion, and then start again. The ballot is one moment in that cycle. Treat it as the whole of the strategy, whether by backing the reformist from round one or by refusing to, and it will not survive the first year of the government it helped elect.
Another really helpful comment on electoral strategy. As you now, I will be leading a Workshop in Cymru'n Codi on ecosocialist republicans and elections (date & discussion paper to be finalised) and, whilst the election systems you have reviewed do not map directly with those in Cymru for the Senedd or the UK (Westminster), the principles may be generalised, and I have shared this latest contribution to our members WA Group. Thanks!
Thanks for this useful article. I have two side remarks :
- France in 2007 was not about unity with Mélenchon - who hadn't split from the PS at that point, and while the question of a unity antineoliberal candidacy was on the order of the day and the LCR's then orientation can be criticised, the question was more of building a broad front for the first round than posing a serious challenge to win the election. It had a bearing on subsequent developments, but it is still separate.
- The squeeze in votes doesn't mean the advantages of an independent profile are lost : running Poutou in 2017 meant an important media coverage for the NPA, and while many sympahisers won during the campaign ended up voting for Mélenchon, people are still telling us about that campaign 9 years later.