Counting votes in optional preferential

The Australian Senate has optional preferential voting. That means voters can number some candidates (or parties), but not number all of them. As I understand it counting goes like this, assuming you have 6 spots to fill:

• Divide the number of valid votes 7 (number of winners + 1), rounding down, then add 1 for luck. This is the quota. You need a quota of votes to win a seat.
• Count votes till someone has a quota. Now only 5 more to go. The rest of the winner's votes (their excess) continues in the count going to lower preferences. Except that doing it that way would be non-reproducible, so actually all the winner's votes continue, but only counting as an appropriate fraction.
• Except that some votes are now exhausted because they didn't vote below that winner. Now the quota needs to be lowered appropriately, otherwise you might not get enough winners.
• And so on till you have your 6 winners.

Do they actually reduce the quota? I left a question on Antony Green's web site but didn't get an answer.

To only reduce the quota for subsequent candidates is unfair. The people who vote for the most popular candidates get votes that count less. But going back and changing the quota for previously confirmed candidates could change the count. Maybe they push on with the original quota, hoping they'll get 6 people elected. But that is bad even if it works.

The correct answer is to set the quota to the smallest quota which will elect 6 candidates. So that if you reduce the quota by even one it will lead to 7 candidates getting a quota. The point is that you want to discard the largest number of votes, because that is where the loony single issue supporters are lurking. This is what easily happens when only one person is being elected: almost half the votes can be discarded.

With this counting method, I now think optional preferential Senate voting is ok, minimizing the chances of extremist parties.