With first-past-the-post there is a problem of vote splitting. Given two candidates A and B, suppose that A beats B by 55 to 45%. But a third candidate C has similar policies to A. If C also stands the votes become A 40%, B 45%, C 15%. Now B wins.
It is to avoid this kind of situation that various forms of "run off" voting are used. In voting theory there is the concept of the Condorcet winner: Imagine that just voting between A, B and C, there were three elections A/B. A/B and C/B. Since in these matches, A beats each other person, A is the Condorcet winner. Election systems are often designed so that the Condorcet winner (if one exists) is likely to be chosen.
If people act rationally and are fixed in their preferences then a first round with three candidates. Followed by a second round if no-one gets a majority will determine the Condorcet winner (if one exists)
But this makes assumptions, particularly that people don't change their preferences. A third candidate can:
- add to the rhetoric on one side of a political divide.
- make one of the two leading candidates seem more reasonable.
- act as a conduit for supporters of one party to move to another.
- delay the final choice, giving time for external events to erode the popularity of one of the main candidates.
- prevent a majority in the first round and give an impression that the leading candidate is actually failing (by failing to get over 50%).
Third candidates can have all sorts of effects on multi-round elections.
What this does not in any way prove is that the motivation of Agus was to change the outcome in favour of either of the two candidates. We don't know and we can't know that (and questions of internal motivation are therefore off-topic here).