Steve's blog

Steve About
Steve's blog,
The Words of the Sledge

Subscribe to the RSS feed.


  • Home
  • Debian
  • PlanetDebian
  • Search PlanetDebian
  • Friends

  • Matthew Garrett
  • Jonathan McDowell
  • Jo McIntyre
  • Martin Michlmayr
  • Andrew Mobbs
  • Mike Pitt
  • Daniel Silverstone
  • Andy Simpkins
  • Neil Williams


    Friday, 22 April 2011

    Bias in the news

    Apologies for yet more politics; normal-ish service will be resumed shortly, I hope.

    I've just put the following into the "BBC News website feedback" form, but I doubt it will have any impact. Quoting it here for the record...

    The BBC news website appears to be showing a systemic bias over the AV referendum. For the last few weeks, there has been regular coverage of the referendum and it seems that there has been the usual blind adherence to "balance" - every story about AV has included some of the (incorrect, unscientific, even hate-filled) propaganda from the "no to AV" campaign without critique or analysis. It's difficult to see any justification for this except deliberate editorial bias, but I don't imagine that there's much that can be done about that.

    However, right now on the BBC News front page at I can see that there is a link "Referendum views" that points to yet another opinion piece from a "No" campaigner (Frederick Forsyth). I only see a single view here, not the plural "views" suggested. I don't see any positive view for AV promoted from the front page at all, nor any link from this puff piece to the other articles that have been written in the recent past. Looking further, I can see that there *has*, in fact, been a positive piece on the News Front Page today (from Billy Bragg) but there are no visible links to it any more. Both articles were posted/updated at the same time this morning (22nd April, 08:34 BST) yet now only the negative one remains. Very shoddy, and not at all what I would expect from the BBC.

    23:24 :: # :: /misc/politics :: 11 comments


    Re: Bias in the news
    Anonymous wrote on Sat, 23 Apr 2011 02:06

    I strongly support the implementation of better voting systems than our existing first-past-the-post system. However, reviewing the description in the articles you linked to, it sounds like "AV" refers to a system substantially similar to instant runoff voting (IRV). If so, that would make it significantly *worse* than our existing system. IRV ignores all of a voter's preferences other than their first choice, until that voter's first choice gets eliminated. This makes IRV actively harmful to both third-party and major-party candidates. Most people voting for a third-party candidate also have a preference between the candidates from the two major parties. Ranking your preferred third-party candidate first (your sincere preferences) causes your preference between the major-party candidates to get ignored, potentially allowing your least-preferred candidate to win. (Precisely the same problem occurs with the first-past-the-post system, but at least then voters *know* that voting for a third-party candidate means they can't express a preference among the other candidates; IRV provides the illusion of a preference but then ignores that preference.)

    By contrast, approval voting and Condorcet both avoid that problem, along with various others. Condorcet seems closest to ideal, while approval voting proves much simpler to explain and preserves most of the good properties.

      Re: Re: Bias in the news
      Chris Cunningham wrote on Sat, 23 Apr 2011 12:47

      That logic would only seem to apply if voters rank every candidate on the ballot rather than only those they actually wish to get into office. If you vote Green right now and actively do not want any of the 2½ major parties' candidates to get in, you can continue to rank only the Green candidate.

      However, if for some reason you would still rather that the Lib Dem candidate won than the Tory then you can give a lower rank to that candidate. You can threfore *both* vote with your conscience *and* do your part tactically.

        Re: Re: Re: Bias in the news
        Anonymous wrote on Sat, 23 Apr 2011 16:52

        I think you've missed my point. With IRV, ranking your preferred candidate first can cause you to get your least-preferred choice rather than your compromise choice. To use the three parties you mentioned as an example, if you want Green but you still prefer LibDem over Tory, then you order them [Green, LibDem, Tory]. Now, consider the two common cases. If Green continues to get too few votes to have a chance, then they get eliminated, and then your preference for LibDem counts and helps them beat Tory. However, you presumably voted for Green because you want them to *win*. If other Green voters also vote [Green, LibDem, Tory], then when you get enough Green voters, LibDem gets eliminated *before* Green; that only requires Green to get more first-place votes than LibDem, not more than Tory. The Tory can then still beat the Green. That result can occur even when a majority of voters preferred the LibDem over the Tory, because IRV ignores all of the preferences for LibDem over Tory except for the ones that rank LibDem first. Net result: because the Green voters voted sincerely and put their preferred candidate first, IRV eliminated the compromise candidate (their second choice) and left them with their least preferred candidate. So, the Green voters still end up having to vote tactically to avoid a bad result.

          Re: Re: Re: Re: Bias in the news
          Chris Cunningham wrote on Sun, 24 Apr 2011 00:19

          I very much doubt that this would actually occur. Second-choice votes are likely to be reciprocal to a fair degree. Nevertheless, if it turns out that neither the Lib Dems nor the Greens can make 50% of the vote then it is quite obvious that neither should be elected.

          Your example assumes that people will rank all the candidates rather than stopping at the ones they don't want (i.e. the Tories), which makes the system no worse than FPTP in that sense; it is still possible that a candidate for whom you express no preference at all is elected, so long as at least 50% of the electorate *do* give said candidate a ranking.

            Re: Re: Re: Re: Re: Bias in the news
            Anonymous wrote on Sun, 24 Apr 2011 04:39

            My example still holds if the Greens leave the Tory candidate unranked. But fair enough, in an elimination system you may want to leave unacceptable candidates entirely unranked. That doesn't let you express a preference among them, though.

            When you say "second-choice votes are likely to be reciprocal", you ignore the possibility that first-party voters will simply vote for their first-party candidate and ignore all others.

            Also, your comment about "50% of the vote" doesn't make it at all clear what you mean by that. Consider the following concrete example:

            49% of voters: Tory 25% of voters: Lib Dem 26% of voters: Green, Lib Dem

            In these results, IRV will first eliminate Lib Dem, then eliminate Green, and declare Tory the winner. However, it seems fairly clear from these results that Lib Dem should win, as 51% of voters would prefer Lib Dem to Tory.

            IRV claims to make third-parties more viable by allowing third-party voters to still express a preference between the two major parties, but if the third party ever actually *does* become viable then this situation arises. The results get even worse and more unpredictable when multiple third parties become viable.

            I've provided a concrete, plausible example where IRV produces a fairly self-evidently worse result than Condorcet, and arguably worse than FPTP since IRV more strongly encourages sincere third-party votes but breaks with viable third parties. Can you provide a concrete example where IRV does better than Condorcet? I've never seen such an example.

              Re: Re: Re: Re: Re: Re: Bias in the news
              Anonymous wrote on Sun, 24 Apr 2011 04:39

              Sorry, my example came out incorrectly formatted. Let's try that again:

              49% of voters: Tory

              25% of voters: Lib Dem

              26% of voters: Green, Lib Dem

              Re: Re: Re: Re: Re: Re: Re: Bias in the news
              Chris Cunningham wrote on Mon, 25 Apr 2011 18:19

              That's a rather contrived example which presupposes a lack of voter education. In reality, "opposition" votes will certainly be reciprocal to a good degree; it is this reciprocity in opposition votes which is precisely AV's strength, as in sufficient numbers it guarantees that one or the other of the opposition parties will win even if the party in power has a plurality of the votes.

              Re: Re: Re: Re: Re: Re: Re: Re: Bias in the news
              Anonymous wrote on Tue, 26 Apr 2011 08:44

              I think you're strongly underestimating the number of blindly partisan voters. That won't change overnight. Many voters will ignore the new system and vote as they always have, and many more will understand the new system but not bother to find out about candidates other than their preferred party's candidate.

              Apart from those, I think you've also underestimated the number of voters with a mindset of "I'd take the other major party before one of those crazy third parties".

              I do agree that the example looks contrived, but then I've never seen a voting system example that *doesn't* look contrived; humans are bad at random, and tend to construct nice neat examples. That said, I do think this scenario accurately reflects the failure mode of IRV. Even if many Lib Dem voters listed Green as a second choice, it would take almost all of the Lib Dem voters doing so to change the results in the example I gave. And also, the situation gets worse (and more chaotic) if you add an equivalent third party splitting up the Tory vote, or multiple parties doing so. Fundamentally you'll still get the spoiler effect any time the number of first-choice votes for a major party falls below the number of first-choice votes for a third party, with neither of them having a strict majority.

              Fundamentally, IRV seems quite prone to strategic voting, since only your first preference counts.

              Various articles I've seen about the use of IRV in Australia suggest that the voters there have already figured out that the spoiler effect still applies, and that third parties still find it quite challenging to get voters to rank them first due to a (correct) perception that IRV requires strategic voting.

              To quote (one of many articles making the same points about IRV): "Before a third party is competitive, the effect of IRV is equivalent to a plurality system in which all supporters of minor parties are somehow convinced to abandon their principles and vote for the "lesser of two evils." Yes, those voters get the satisfaction of knowing they voted for the party and the candidate they truly prefer, but their first choice is eventually eliminated and has no effect on who actually wins. After a third party is competitive, on the other hand, the effect if IRV is equivalent to a plurality system in which many voters are somehow convinced to forget about strategy and vote sincerely. As most intelligent voters know, that would wreak havoc with the stability of our political system."

              In any case, I don't mind if you disagree with my point that IRV seems like a net loss. It just bothers me to see IRV claimed as a win for third parties. I mostly just wanted to make the point that IRV still has several major flaws, and that better systems exist which don't have those flaws and take all preferences into account rather than just the first preference.

              I'll also reiterate one previous point: I've provided an example where IRV produces a self-evidently worse result than Condorcet. I've never seen an example of a set of voting results where IRV produces *better* results than Condorcet; at best, it sometimes gives the same answer. Given the known property of voting systems that no perfect system can exist (all systems must violate at least one of a set of desirable properties), obviously some such example must exist where Condorcet produces an arguably undesirable result, but I've never seen any such example (realistic or otherwise).

      Re: Re: Bias in the news
      Steve wrote on Sun, 24 Apr 2011 21:49

      First of all, thanks for actually arguing on the *merits* of the different voting systems - that's quite a refreshing change!

      The article at

      includes a very good mathematical treatment of the voting systems, and was passed on by friends of mine.

      Yes, AV/IRV *can* possibly produce the results you're describing, but only in quite unlikely circumstances. Compared to FPTP (which *very* often produces an undesirable result) I'm prepared to live with that.

      Condorcet voting options would be a much better answer mathematically, but I don't believe they're likely to be explicable for normal voters.

        Re: Re: Re: Bias in the news
        Anonymous wrote on Mon, 25 Apr 2011 09:22

        It seems like the situation I mentioned will arise any time a third party becomes sufficiently viable to displace a major party. Given that I very much want to see that happen, IRV seems like a major problem.

        On the other hand, IRV at least helps third parties get a much better show of support, which will help with ballot access problems and hopefully help amplify support in future elections (due to people realizing they're not alone in their sanity).

    Re: Bias in the news
    RichiH wrote on Sat, 23 Apr 2011 08:15

    It would help if you told people what the AV issue actually _is_. Not everyone is from the UK ;)


    Your Comment

    URL/Email: [http://... or mailto:you@wherever] (optional)
    Title: (optional)
    Anti-spam:Select the first of the following words and enter it in the "Human" box
    quotes quickie quack quailed quarry quills queasily quads quires quenches
    Save my Name and URL/Email for next time