How does Condorcet compare with BC-STV?

First, a fuller description of the former BC-STV proposal:

  1. The proposed system would have elected 85 representatives, as at present, but from 20 electoral districts, not 85, electing from two to seven representatives per electoral district (depending on population density — higher numbers of representatives per district in higher-density areas, minimizing the geographical size of the districts).
  2. Voters vote in a single round, voting for a minimum of three candidates, ranking them in order of preference:  1, 2, 3, etc.
  3. Given that multiple candidates are to be elected, the number of votes needed for each to be elected is reduced accordingly.  This threshold is the “Droop quota”, as:  1 + (valid-ballots) / (1 + number-of-seats).

Counting can involve several rounds.  The count begins by counting each ballot’s first-preference choices:

  1. If no remaining candidate has at least as many votes as the election threshold (Droop quota), the candidate having the least number of votes is eliminated, as for IRV, reallocating this candidate’s ballots, as for IRV.  This is repeated until:
  2. If at any point the votes for any candidate equals or exceeds the election threshold, that candidate is elected.  In the event that the required number of representatives are now elected, the process is done, otherwise
    1. the “surplus” ballots, being any votes in excess of the election threshold for the elected candidate(s), are re-allocated to the respective next-choices of the elected candidates’ ballots.
    2. This is done based on a weighting factor, being the percentage of surplus above the election threshold, and the portion of the elected candidates’ ballots corresponding to each respective ballot’s “next” candidate.  So, let’s say 100 votes are needed to win, and Candidate A receives 200 votes. This gives us a weighting factor of 100/200 = 50% indicating the portion of votes that can be transferred to other candidates.
    3. If a given other candidate B is the “next” candidate on, say, 50 of Candidate A’s ballots, that’s 50/200 of the vote, so 50 * 50% votes are transferred to candidate B.  Candidate B would receive 25 more votes.  We proceed thus for the other un-elected, un-eliminated candidates.
  3. … we repeat this process until all required positions are elected.

The first general observation is that, as for IRV / AV, a ballot’s full list of preferences is not necessarily used.

If we consider BC-STV as a single-member system:

  1. The Droop quota becomes 1 + (ballots/2), i.e.:  50% + 1. This is a simple majority.
  2. Given that in this case we are electing only one member — once a candidate is elected there are no remaining positions to be elected, so there is no reallocation of the elected-candidate’s “surplus” ballots;  for a single-member BC-STV the weighting factor computation is irrelevant.
  3. If, after any given counting round, there is no winner, the lowest-count candidate is eliminated, and his or her votes reallocated according to their respective next preferences.  This is IRV.  (See Why not IRV?)
  4. For a single-member election BC-STV devolves exactly to IRV.

We can also use Condorcet/Ranked-Pairs as a multi-member system, to elect, say, n members:

  1. We take the n most-preferred candidates from the Ranked-Pairs outcome.
  2. There’s no “weighting” calculation, no “surplus” votes to be reallocated anywhere, and no progressive elimination.
  3. In one fell swoop we identify — and elect — the required number of members, having holistically considered all preferences from all ballots.

While I’m not enthusiastic about proportional representation in general, multiple-representation schemes such as BC-STV diminish some of my concerns.

Of particular favourable note here, is that votes are cast for individuals (who might or might not be identified with a given party), and are not explicit votes for a party.  This keeps the elected representatives fundamentally responsible to the voters, NOT their respective parties.

Though I do think that BC-STV is, to all practical intents and purposes, a dead issue in BC for many years to come, if it turns out that the grand consensus of the population is for multi-member electoral districts, a Condorcet/Ranked-Pairs multiple-member election wouldn’t be a bad alternative.

Bookmark the permalink.