Suara ke Kursi

The problem

Journalists, students and legislative staff reading the new election bill hit the same wall. Indonesia turns votes into seats with two rules few people can work through by hand. A party below the 4% national threshold gets no seats anywhere, even where it came first. Each of the 84 districts then shares out its seats by the Sainte-Laguë method, dividing party votes by 1, 3, 5, 7 and onward.

In 2024, 18 parties took part and 8 cleared the threshold. 17,304,303 valid votes produced no seat. PPP, a party already in parliament, missed the threshold by about 0.13 percentage points. The Constitutional Court held the 4% figure valid for 2024 only and asked lawmakers to set a new one before 2029 that keeps wasted votes low. That is a calculation, and no public tool performed it.

What I built

The seat maths lives in one small, checkable function

I kept the allocation engine apart from everything else. It takes party votes, district data and a set of rules, and returns the full result. It has no dependencies, no randomness and no clock. The same code runs in a plain script outside the browser, so anyone can check the numbers without trusting the interface.

The explanation shows the actual arithmetic

Every seat the engine awards records the full table of quotients at the moment it was won. The "show your work" view draws that record directly and never recalculates. For any district, you can step through the seats one at a time and read each division. The picture can't drift away from the maths, because it is the maths.

Seats move when the rules change

The main view is a 580-seat chamber. Drag the threshold from 4.0% toward zero and individual seats travel from one party's block to another. The engine recalculates all 84 districts in 0.23 milliseconds, so it can rerun on every animation frame. I drew all 580 seats in a single animation loop, which keeps the motion smooth on a phone. When the slider passes 3.87%, PPP enters the chamber.

It has to match 2024 before it can show a what-if

I set a rule that the app must reproduce the certified 2024 result before it offers any hypothetical, and the check blocks the build. The per-district vote table exists only in scanned election commission PDFs, and I haven't finished extracting it. So the header says plainly that the official result is not yet reproduced. National totals use certified figures, and the wasted-vote count matches Perludem's independent tally to the vote.

Every rule points to the law

Each rule cites its article in Law 7/2017, and each headline figure cites its election commission document. The citation opens in a panel you can reach by keyboard and read at leisure. The copy states outcomes in neutral terms, such as "ten parties won no seats".

Result

Suara ke Kursi is live, public and in Indonesian. It has four rule controls: threshold, threshold scope, divisor method and district map. Every setting is saved in the link, so an argument about a 2% threshold can be sent to someone. It models 580 seats, 84 districts and 18 parties across five views. Each view has a keyboard-reachable table with the same numbers. An automated audit found zero accessibility violations at three screen widths, in light and dark. I built it in about 28 hours over 2 and 3 September 2026.