Indra and Spain's vote count: persistent fraud suspicions

Indra and the recount: the suspicion that won't fade

There is a paradox defining Spain's electoral system: the official result of general elections depends on private software, but the actual recount happens on paper. Minutes are filled out by hand at each polling station, signed by observers, and scanned. Indra’s computer only sums them up. However, distrust has continued to grow since the PSOE went from 80 to 130 seats in two years, according to calculations circulating online. In that interval, part of the electorate stopped believing in statistical chance and began looking for the mechanism.

The trigger for this debate is not judicial evidence, but an accumulation of indications that have never reached the courts. The results website crashed on May 28 and again on July 23. Transparency officials requested reports on those failures, but Interior Minister Marlaska did not provide them, according to circulating tweets. Pre-Indra electoral audits are not public in Spain, although they are in other countries, also per social media claims. And the final recount, according to several testimonies, is limited to cross-checking doubtful minutes against the company’s program. None of this constitutes proven fraud. But it is not a transparent system either.

What the law says about the vote count

The LOREG (General Electoral System Organic Law) establishes that the general scrutiny is carried out by the Provincial Electoral Board. Political parties can be present, but they are not obliged to audit every minute. That distinction—being able to do so versus being required to—is the core of the discussion. Those who defend the system point out that any party can request a table-by-table recount and that data is published on the Ministry of Interior website. Critics respond that the burden of proof cannot fall on the parties, because they are both judge and party to the same mechanism.

The most repeated argument is that minutes are public. Thousands of agents take photos at each polling station. Comparing those photos with official data would, in theory, be as simple as putting two columns into a spreadsheet. If the numbers match, there is no fraud. If they don’t, there is a complaint. The problem is that no one has systematized that comparison on a large scale. Not because it is impossible, but because no one with legal standing to report it has done so.

The case of Junts votes appearing in other districts

One of the most cited episodes is the appearance of votes attributed to Junts in constituencies where the party was not running: Madrid, Andalusia, Canary Islands. The error has been explained as a software allocation failure, not manipulation. But the episode fueled the thesis that the system can make mistakes without anyone detecting them until the error is too big to ignore. How many small errors go undetected?

The institutional response has always been the same: the minute prevails. If the minute says one number and the computer says another, the minute wins. That is true on paper. The doubt is who verifies that the computer didn’t change the number before the minute reached the Electoral Board. And this is where sealed envelopes, uncertified minutes, and testimonies from agents who claim to have seen discrepancies without anyone correcting them come in.

Why there are no electoral fraud complaints in court

Spanish legislation limits standing to challenge an electoral result. An individual citizen, an association, or a law firm cannot file a complaint. Only political parties that ran in the election can do so. This restriction, which exists to prevent massive challenges and frivolous litigation, becomes the central argument for those who suspect foul play: the only ones who could report fraud are the very ones benefiting from the system.

The objection is reasonable. If a small party doesn’t gain representation, why would it spend resources challenging the result? If a large party gets seats, why would it question its own legitimacy? The result is a balance where no one has incentives to audit. And in the absence of auditing, suspicion becomes certainty for part of the electorate and conspiracy theory for the other.

The precedent no one forgets: PSOE primaries

The most cited episode as proof that the system is manipulable is not a general election, but a primary. In 2017, the PSOE primaries where Sánchez regained leadership were challenged by sectors within the party itself. The management committee controlling the organization at the time was accused of favoring one candidate. That was not electoral fraud in the legal sense, but it demonstrated that internal counting mechanisms can tip the scales.

That precedent is now used as an analogy: if irregularities can occur in a party with identified and registered members, what can happen in a system with millions of votes and opaque software? The comparison is debatable, because the scale and controls are not the same. But the question about software opacity remains unanswered.

How to verify the count without depending on Indra

The most repeated proposal is simple: publish images of all minutes on the official website, alongside the data the system attributes to each table. If the photos and numbers match, distrust dissolves. If they don’t, there is objective proof. It doesn’t require changing the law, creating a new body, or hiring more civil servants. It only requires that scanned minutes be accessible.

Another measure would be making previous software audits public, as is done in other countries. And a third: allowing agents from all parties, not just those winning seats, to access the final recount. None of these measures have been implemented. The system remains the same. And as long as it stays the same, the shadow of pucherazo (electoral fraud) will continue to fuel theories that can neither be proven nor entirely debunked.

---

How many counting errors are tolerable before a system loses trustworthiness? The question has no technical answer. It has a political one. And that, for now, no one has wanted to give.