As I said after Australia's 2022 election (https://grampsgrumps.blogspot.com/2022/05/who-to-eliminate-in-stv-preferential.html), we should eliminate candidates who can't win against any remaining candidate. The 2025 election reinforces this. There have been several cases where one candidate has a good lead but can't win against either of the next 2 candidates. The result is then determined by who finishes 2nd. This causes 2 problems:
- The voters who backed that leading candidate are deprived unnecessarily of their right to influence the winner with their preferences.
- Voters may feel that they have to vote strategically to get their preference between expected 2nd and 3rd candidates counted. This means that they don't get to indicate the policies they really prefer, and might even prevent their preferred candidate winning if she has more support than expected. Avoiding strategic voting is one of the key advantages of STV over 1st past the post.
Eliminating the candidates who can't win is easy by computer if the votes are digitised,
If forced to do it by hand you can make a lot of progress this way. If there are N candidates then draw up an NxN table. For each vote, for each pair of candidates X and Y: if X has a higher preference then add 1 to his square against Y and subtract 1 from Y's square against X. We can have lots of people doing this at once on different lots of votes, then add the tables at the end. When all votes completed, we eliminate every candidate that only has negative numbers. We can also eliminate groups of candidates at the bottom who can't combine to get ahead of any other candidate. Then repeat. No redistributing preferences needed. Hopefully at the end there is one candidate who wins. If not then we do have to see who can't win. And if we end up with X beats Y and Y beats Z and Z beats X, or worse, then we have to do something more arbitrary, like eliminating the one with the lowest top preferences, as now, or the most bottom preferences, as I somewhat prefer.