Westeros.org Book List – IV.

Why this List was created in the way it was: Possibly Asked Questions.

It is quite reasonable that people may query the rationale behind certain decisions in making this poll, either because they seem bad decisions, or simply out of curiosity. So, I’d like to pre-emptively answer some questions.

1. Why only twenty works per ballot?

If the ballot had been smaller, there would have been less variety in votes, and fewer works nominated overall. Why, then, not make it larger? There are four main reasons. Firstly, workload: there were approximately 2,000 different nominations to tally in this poll, and a larger ballot would have made this even harder. An ideal ballot would have been 100 works – but this would be five times as much work.  Secondly, inclusiveness: this was a popular poll, and even with an explicit allowance that incomplete ballots could be submitted, a request for a longer ballot would have put more people off voting. Thirdly, ignorance: only a portion of the electorate actually know enough to nominate 100 good books, and although some would simply have left their ballots incomplete, others would have packed them with inferior works in great numbers; at least with only twenty books per ballot, the books are likely to be genuinely admired, and not merely put in to make up the numbers.  And, fourthly, mechanics: how would points have been allocated in such a system? A large ballot with a sharp distribution of points (ie top votes receive many more votes than those lower down) would have rendered low votes essentially irrelevant anyway (if the top book gets 50 points to the bottom book’s 1, for instance, there’s no point voting for a bottom book at all, in a poll where only 100 people vote overall). A flat distribution (all votes get the same points, or top votes get only slightly more votes than low votes), combined with a large ballot, would have turned it from a popularity contest into a contest of who had even heard of the book in question, as well as making the labour of ranking a 100-strong ballot futile. A twenty-work ballot is big enough to produce interesting results, while small enough to be viable; a larger ballot would have produced only slightly more interesting data, with a considerable amount of extra work and difficulty.

2. Why not ranked?

I don’t believe it’s possible to rank books so finely. If it’s really obvious to you that your fourteenth book is FAR better than the fifteenth book, and that they shouldn’t receive the same number of points, then I think that maybe you’ve only read fourteen really good books. Many voters commented as it was that their 20 could be different on another day – what gain in accuracy would there actually be, then, from the greater precision of an exact ranking?

3. Why tiered?

Although I don’t believe in fine ranking, I think a tiering is useful and meaningful. There may not be a difference between fourteenth and fifteenth, but it’s reasonable that there may be a big difference between first and fourteenth.  In addition, this allows a sharp points allocation, as described in answer to Question Six. However, the exact nature of the tiering is arbitrary, beyond the fact that I wanted the highest tier to be smaller than those below.

4. Why the Greats List?

I’m not entirely sure; it’s certainly not necessary. However, I think there is something valuable in asking for people’s opinions of things truly superior, and by making it an optional extra I tried to prevent people from thinking that they should automatically vote in it, so that those not well-read did not simply give Greatness votes to their favourite books. It could have been worked out through the main list (by making the Greats votes just add more points to the tally for that book), but I felt it worthwile to distinguish these books from those who scored highly due to large numbers of low votes. It is worth noting, however, that all books on the Greats List would have made it onto the Main List on their own merit, even without this system. There were only two possible votes per voter for the GL as a form of defense against over-enthusiastic voters giving GL votes to too many of their personal favourites – with only two votes allowed, the damage any single voter could do was limited. However, in the event there were fewer GL ballots than I expected, so additional GL votes per ballot would probably have been a good idea in hindsight.

5.Why are so many terms left undefined?

“Great”, “series” and “genre” were all left to the individual, because I did not feel I had the authority to impose terms. Greatness is a hard concept to define rigidly; as it happens, I think people did a pretty good job with that. There was surprisingly little disagreement about what constituted a single series; in any case, I could do little about it, since I do not know every author on the list (there were more than 300), and because I did not know in advance what would be nominated, and could not make definitive statements in advance (and did not wish to make rulings during the voting, as that would be unfair to those who voted before the ruling). I did not define “genre” because I know no definition; but, more importantly, because I didn’t feel it was my place. Genre is, to me, whatever genre fans are fans of. If I had run this poll only to find out that the top “Science Fiction and Fantasy” authors were Agatha Christie, Dorothy Sayers and Ruth Rendell, that would just have meant that the fandom was mislabelled – and insisting that Christie fans vote between Wolfe and Mieville fans would have been pointless. Needless to say, many voters may ALSO be fans of Agatha Christie, but I assume that there is a division between the two types of literature within their minds – part of the point of a “genre”, after all, is things are read because they are within the genre. For similar reasons, I tried to stay clear of “Science Fiction and Fantasy”, or even “Speculative Fiction”, prefering the simple term “genre”, so as not to seem to be imposing external criteria on people’s decisions. In the event, there were only a small number of works voted for where this was an issue. For those (in particular the magic realist novels), I think the fact that genre fans consider them to be within their genre is more than enough to merit them being on this poll, which is, after all, for the benefit of genre fans, not literary critics. It is therefore the definitions of genre fans which should be paramount.

6. Why the 1-2-8 ranking?

Clearly, the precise point allocation system is arbitrary. Two extreme cases are balanced: the flat allocation system of 1 point per vote, regardless of position, and the extremely sharp allocation system (eg 20 points for #s1-4, 1 point for #s13-20). I opted, I think, for a middle ground, but leaning toward the sharp, because I wanted the list to represent a collection of books the hypothetical guided reader might potentially love, rather than of books they are likely to quite like; as such, I wanted to give particular weight to people’s favourite works. This also helps to create an eclectic list – a perfectly flat allocation would result in the 100 most mentioned books, which would entirely exclude books not widely read, and the sharper the allocation the more these unusual books get moved up the list. These books are likely to either be obscure personal favourites (which might become a favourite of a new reader) or else books votes for by very widely-read readers (in which case they are probably very good); either way, they are probably deserving of a wider audience. That said, I did not wish to go too far toward steepness – a too-steep allocation would be effectively making the ballot only four books long (because lower-ranked books would get ‘shouted out’ by the steep allocation), which would give too small a sample, and one too easily held hostage by individual votes, getting rid of the whole point of a large popular survey.

7. Why aggregate the votes by author?

I aggregated votes per author because otherwise some authors would do particularly badly as a result of fans splitting votes between different books. A good example of this was the British Big Read competition earlier this decade, where Pratchett fans split their votes between half a dozen books in the top 100. As a sample result, for instance, voting by book instead of by author would have excluded Christopher Priest from the Main List entirely. On the other hand, voting by book could have allowed some authors to get too many books onto the list.

8. Why use LV for the Greats List?

I didn’t think there was any point having a more sophisticated system, given the electoral landscape: the vote-splitting concern only applied to Tolkien, and he had more than enough votes to split.

9. Why the stepped allocation of slots?

I could have simply given one slot to each author, but to me that seemed unfair. A recommendations list should take notice of relative popularity, and an author like Wolfe deserves to have more books on the list than an author like Pullman, given that Wolfe got more than ten times as many points as Pullman did. Also, although the system deals with vote-splitting problems, there is still a tendency for voters to spread their votes between authors, meaning that secondary works by top authors often don’t get the votes they deserve, because they are overshadowed by a more famous work. It is true that a more sophisticated proportional representation system could have replicated this allocation, but that seemed a pointless amount of work, since it would have had to have been fine-tuned through trial and error.

10. Why a separate 21st Century List?

Reading polls are always biased heavily toward recent works – they are more widely read, and often more widely remembered. Because we are living now in 2010, this recency bias can be dealt with simply and with some justification by dividing the list into two: a list for the 20th Century, and a list for the first decade of the 21st Century. As it happens, the first part has been a little diluted by the inclusion of a handful of pre-20th century works (not enough to merit their own list, as was my original plan), but I think the concept remains coherent. The decision to count 2000 as part of the 20th century was made only partly on the basis of pedantry, and more on the basis of personal bias – too many books seemed to have come out in 2000 for me to shunt them off into their own list (although one of my own favourites from that year, Mary Gentle’s “Ash”, failed miserably to gain votes in the poll anyway).

11. Why re-allocate pool slots to top authors?

I could have given all pool slots to authors at the bottom of the list. However, the re-allocation system enables authors with broad support to get multiple books on the list – had more of the slots gone to the bottom of the list, they would actually have gone to books with less support than some of the additional books higher up the list.

12. Why impose quotas on pool slots and on second/third allocations?

I didn’t want top authors to get books in with only the slightest support. The quota system ensures that all books on the list are either a top-4 book for somebody, a 5-12 book for two people, or a 13-20 book for at least four people (as well as the author of the book having broader support for their overall body of work).

13. Why not simple LV for the 21st Century List?

Authorial fiat. The list departs from LV in giving slots to Greats automatically; although these books weren’t put on the Greats list, I felt that they deserved to be somewhere. It also includes only post-200o votes; I felt it not entirely fair to eliminate contemporary authors in favour of the less-regarded late works of older authors who are already included in the Main List. For the sake of transparency: not safeguarding slots for Greats would have taken the Chiang and Stover off the list and replaced them with the Gaiman and the McCarthy; including pre-2000 votes would have taken off the Valente and the Lynch and replaced them with the Wolfe and the Gaiman; making both changes would take off the Chiang, the Stover and the Valente, and replace them with the Gaiman, the Wolfe, and the Priest.

14. Why is this only a genre list?

Because the voters were genre fans, and thus not the people to ask for a general list. Such a list would have been a “non-genre books that genre fans might like” – that’s perfectly valid and interesting in its own right, but as there are plenty of non-genre lists out there, I thought it more interesting to make use of the relative expertise of the voters and compile a “genre books for genre fans” list instead.

15. Why is this a popular vote?

Some might suggest that a better list would be compiled by only asking “experts”. This has some merit, and an interesting project for the future would be to compile the lists of different reviewers. However, not all potential readers are experts with expert tastes, and I felt it fair to include the views of ordinary readers as well; in any case, I don’t know the voters well enough to feel confident picking out “experts” in advance (and doing it retrospectively on the basis of how they voted would be clearly biased).

16. What sort of electoral system is this, anyway?

The electoral system is of my own divising, to meet the particular needs of the poll – unusually, this system is designed to be comprehensive, rather than consensual, with diversity respected through the election of representatives from different factions, rather than through the election of broadly acceptible candidates. In a political election, this system would probably produce an unworkable assembly of extremists, but I felt that this was suitable for this type of list. For the curious, I would call this a Double Simultaneous Limited Positional Vote, with a Qualified Banded Apportionment List Vote and an Individual Block Vote as the two elements. Double Simultaneous Votes (DSV) are most famous for their use in Uruguay, where they are used in single-member constituencies, but they also occur, without the name, in several Proportional Representation systems – essentially, it means that a single vote is a vote for both a party and a candidate (or in this case both an author and a book). Positional Voting is a name for any system where candidate are given points for their position on a ballot, and is contrasted with conventional Preferential Voting (generally, although strictly speaking positional systems ARE preferential; there’s no word known to me for non-positional preferential systems, although I would call them ‘Subordinate Voting’) – in politics, this is best known in the form of the Borda Count, and the hypothetical Quota Borda System (in both of which all the candidates must be ranked, while this is ‘Limited’ because only some candidates are positioned), but it is also commonly seen in the award of prizes by juries. “Banded Apportionment” is my word for the system of allocating ‘seats’ in rough chunks based on position in the results, rather than based on actual voting scores – I don’t think it’s used in politics anywhere, although the Chilean binomial system could be described as banded apportionment with an exception for parties with more than a certain number of votes. It is “Qualified” because the apportionment only gives provisional ‘seats’, which are only taken if certain criteria are met. The Block Vote is simply the system where the voters can vote for multiple candidates, and the candidates with the most votes win.


One thought on “Westeros.org Book List – IV.

  1. […] IV. – Why the list is made like that […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s