The ballot proves nothing

In June 2567, Thailand chose two hundred senators without holding an election.

Around forty-six thousand people applied. They then voted for each other. No member of the public cast a vote, and none was entitled to.

Establishing what happened in that process has since required an investigation subcommittee, more than seven hundred witnesses, and a case file exceeding eighty thousand pages.

That is the argument of this article, and it is a point about design rather than about anybody’s conduct. An election that has to be reconstructed from testimony, at that scale, is one whose own records contain nothing decisive. The ballot proves nothing, because it was built so that it could not.

How it was supposed to work

The intention was defensible and worth stating fairly.

Thailand’s designers faced a real problem. An elected Senate becomes a second chamber of party politics. An appointed Senate is a patronage instrument. So the 2017 Constitution tried a third thing: let people select from among themselves, by occupation, so that the chamber represents expertise rather than parties.

Twenty occupational groups. Two hundred seats, ten from each group. Applicants register in the group matching their background and are selected upward through three levels — district, then province, then national.

At each level there are two rounds. First you vote within your own group. Then comes the safeguard: การเลือกไขว้, cross-selection. The groups are divided into four lines of three to five groups each, and in this round you may not vote for yourself or for anyone in your own group. You must vote for strangers in another occupation.

That second round is the anti-collusion device, and it is the interesting part of the whole system.

Why the safeguard points the wrong way

Cross-selection defeats exactly one kind of bloc: one confined to a single occupational group. If twenty accountants agree to vote for each other, the round in which they cannot vote for accountants destroys the arrangement.

It does nothing whatever to a bloc that spans groups. For a coordinated set of entrants distributed across several occupations, cross-selection is not an obstacle. It is a mechanism, because it converts an advantage held inside one group into votes delivered into another.

The safeguard is orthogonal to the threat. It was built against the arrangement that is easiest to imagine and hardest to organise, and it is indifferent to the arrangement that is harder to imagine and much easier to organise.

That is a criticism of a design, not an allegation about anyone. It would be true if the 2567 selection had produced a chamber nobody complained about.

The property that makes it unauditable

Now the deeper problem, and it is the one this site keeps arriving at from other directions.

In an ordinary election, the electorate and the candidates are different populations. That separation is what makes electoral offences detectable at all: something has to pass between the two groups, and the passing can leave a trace — money, a list, an instruction, a witness on the other side of the transaction.

Here the voters are the candidates. Every elector is a competitor and every competitor is an elector.

Which means the observable behaviour of collusion and the observable behaviour of legitimate association are the same behaviour. People in a professional group who know each other, travel together, share a region and vote for one another are doing something the system explicitly invites. A coordinated bloc does the identical thing and leaves the identical record.

And the arithmetic removes what little signal might remain. With roughly forty-six thousand applicants spread across twenty groups and hundreds of districts, the number of voters in any single district-level group ballot is very small — frequently double figures, sometimes fewer. In a contest that size, a handful of coordinated entrants is decisive, and their effect on the result is statistically invisible. There is no turnout anomaly to find, no improbable margin, no discontinuity.

So there is nothing in the ballot to audit. Not “hard to detect”. Nothing to detect, by construction.

This is the legibility problem this archive has now hit six times. Rules bind when the behaviour they govern is observable by another person. Article 134 Nobody hears you speed found it in the horn that nobody hears. Article 142 found the benign version — a market that works because performance happens in public. Here it is load-bearing for the constitutional order, and it fails.

Which is why it takes eighty thousand pages

Follow the consequence.

If wrongdoing cannot be established from the electoral record, it can only be established from outside it — from testimony, phone records, money movements, travel, accommodation, and people willing to say what they were told and by whom.

That is not an election petition. It is a conspiracy investigation, and it has to be conducted with the tools of one.

Which is exactly what has happened. The joint Election Commission and Department of Special Investigation subcommittee reported gathering evidence from more than seven hundred witnesses across a file of over eighty thousand pages. Separately, reserve candidates have pressed the DSI to advance parallel lines of inquiry under secret-society and money-laundering provisions rather than electoral law.

Read that as a structural verdict rather than a scandal. When the ordinary machinery of election scrutiny is useless, the state reaches for organised-crime statutes instead. The design forced that, whatever the outcome turns out to be.

What is alleged, and what has actually happened

This section is reporting, not conclusion. Nothing below is established.

In July 2568 (2025), the central investigation and inquiry subcommittee — a joint EC and DSI body chaired by a Deputy Secretary-General of the Election Commission — resolved to refer 229 accused persons to the Election Commission for consideration. Thai reporting of the resolution described these as 138 sitting senators and 91 executive committee members of the Bhumjaithai Party and associated network.

The provisions cited were sections 36 and 62 of the Organic Act on the Acquisition of Senators B.E. 2561, enforced through section 70 — section 36 prohibiting collusive, dishonest or unfair conduct in the selection, section 62 prohibiting supporting or conniving at breaches. Section 77(1) permits the Commission, where a political party is implicated, to petition for that party’s dissolution. Reporting also referred to section 113 of the Constitution, which prohibits senators from being under the mandate of a political party.

The Election Commission publicly stated that reports it had prosecuted 229 people were untrue and that the file remained at its first stage. The EC Secretary-General recused himself, having served as Director of the Senate Selection and therefore being an interested party.

As at August 2026 the file is still not concluded. The Commission has indicated its consideration would finish at the end of this month — and has been explicit that finishing consideration is not the same as voting on the case, which requires a separate meeting. Reporting has noted delays attributed to incomplete evidence.

Those accused are entitled to the presumption of innocence, several have denied the allegations publicly, and no court has determined any of it. If and when the Commission votes, this article requires updating before it can be published at all.

The term “สว.สีน้ำเงิน” — blue senators — is in general use in Thai media, blue being the party’s colour, to describe senators observed to be aligned with a major party. iLaw has documented the concentration of successful applicants in particular provinces, noting that Buriram returned the largest number of any province at fourteen. Concentration is a fact about the results. It is not proof of anything and is not offered as such.

Why any of this matters beyond the case

Here is the part that survives whatever the Commission decides, and it is the answer to why this is a democratic question rather than a procedural one.

Under the 2017 Constitution the Senate is not a debating chamber. It is a gatekeeper.

It participates in confirming appointments to the independent agencies — the bodies that regulate elections, investigate corruption, and adjudicate constitutional questions. And constitutional amendment cannot proceed without a threshold of Senate support, which is the mechanism by which the charter has remained unamended in the ways that matter.

So the chamber selected by the least auditable method in the system holds the veto over changing the system, and a hand in appointing the people who police it.

One consequence of that deserves stating plainly and without any accompanying insinuation. The Election Commission is among the bodies whose membership the Senate has a role in confirming. The Election Commission is also the body now adjudicating allegations about how that Senate was selected. That is a feature of the constitutional architecture as drafted. It is not an accusation against any commissioner, and nothing here suggests any commissioner has behaved improperly.

Article 76 Nothing lasts long enough to build found that Thailand has never let a constitution live long enough to accumulate the conventions that make one work. Article 11 111 dissolved parties: how Thailand made disbanding politics routine found party dissolution normalised as a routine instrument.

This is the same disease at a different site. A rule was written to keep parties out of the Senate. The rule cannot see whether parties are in the Senate. So the question of whether the chamber is legitimate cannot be settled by looking at what the chamber did — only by a criminal investigation, whose outcome will itself be contested, and which will be adjudicated in part by institutions the chamber helped to staff.

Whatever the answer, the country cannot get it from the ballot. That is what was actually built.

The honest counter-argument

It deserves a hearing and this site’s rule is to give it one.

The alternative designs are all worse in obvious ways. A directly elected Senate duplicates the lower house and is captured by parties openly rather than covertly. An appointed Senate is capture by definition. A Senate selected by professional bodies hands the power to whoever controls those bodies.

The designers were not solving an easy problem and there is no clean solution to it. It is also true that some senators selected in 2567 have no alleged connection to anything, and that treating the entire chamber as illegitimate is both unfair to them and analytically lazy.

And a system being unauditable does not establish that it was abused. That is precisely the point of this article and it cuts both ways: the design cannot prove collusion, and it cannot disprove it either. Everyone selected under it — including anyone entirely innocent — now carries a suspicion the system gave them no way to dispel.

That is the strongest thing that can be said against the design, and it does not require anybody to be guilty.

---

Common misconceptions

“Thais elected this Senate.” No member of the public voted. Roughly 46,000 applicants selected from among themselves.

“229 senators have been charged.” No. An investigation subcommittee referred 229 accused to the Election Commission for consideration in July 2025. The Commission publicly corrected reports that charges had been brought.

“The cross-selection round prevents collusion.” It prevents blocs confined to one occupational group. It is not an obstacle to a bloc spread across groups.

“If it happened, the results would look strange.” With very small numbers of voters in each district-level group ballot, a coordinated minority is decisive and leaves no statistical signature.

“The case will settle it.” It may determine criminal liability. It cannot make the selection method auditable, and the next selection runs on the same rules.