Caveats – What’s Wrong with this List?
All selection systems are flawed. A number of things must be noted regarding this one:
- The treatment of novel series is to a degree arbitrary; whether books are counted individually or as a single series depends firstly upon voting patterns (whether multiple books from the same series are voted for repeatedly) and secondly upon subjective conceptions of how united the series is, and the extent to which earlier books must be read before later – as I have not read most of the series nominated, this judgement may be questionable.
- Because of this lumping of works into series, and because voters could not vote for multiple books from the same series, those authors who write multiple standalones will have an advantage over those who wrote only a single long series. However, if readers had been free to vote for multiple books from the same series, there would likely have been an even greater bias towards writers of successful series.
- It is possible for single voters to pack their ballots with votes for a single author. In some ways this is fair (by doing so, they forego the chance to vote for other authors), but it is worth noting that in a small handful of cases authors have made the list as a result of vote-packing by individuals, not by broad support. Future polls might include a vote-shrinking system whereby multiple votes for one author were allowed, but with diminishing points allocated to each additional vote. In any case, this feature has not overly distorted the results.
- A major structural tension arises from the problem of giving attention to great but obscure works. The large number of points given to top-tier votes goes some way to remedying this, but the poll remains structurally biased against those authors who are too obscure for ‘populist’ readers to have read, yet not quite good enough to make it onto the top twenty of the ‘expert’ reader. As we might expect the top twenty of a well-read voter to be of better quality than that of a narrowly-read voter, this means that many works have gone un-nominated, or been given only a handful of votes, despite being of higher quality than some listed works. This problem cannot be addressed without a far longer ballot, which would have brought its own problems.
- There are errors in the tabulation. If you think about it, that shouldn’t be too surprising – it’s a grea big table 100 columns wide and over 600 rows long! I have endeavoured to minimise errors (I have, for instance, checked the total points for each voter, so that the correct number of votes has been registered for each), but without triple-checking, by hand, every one of the 6000 cells against the 2000 nominations, I can’t guarentee that some votes haven’t been put into the row above or below the one they should have gone into. In fact, I know that this has happened, since one work has ended up with 0 points, when it should actually have had 1 point. No doubt there are a handful of other minor errors that I haven’t noticed. However, I don’t believe that any of these errors are likely to have had any effect on the results.
- This poll was the product of voters at a single internet forum, dedicated to the series “A Song of Ice and Fire”, by George R.R. Martin. Although it represents a good cross-section of tastes, selection bias is inevitable.