Thande
Donor
"Thandean Representation" is a voting system (or family of systems) I developed primarily for LTTW, so that a parliament formerly elected from multi-member bloc vote constituencies* could be converted over to a form of proportional or pseudo-proportional representation without actually changing any aspect of the voting process from the point of view of the voter. However, it has also been influenced by my own personal views on the problems of both FPTP** and pure proportional representation systems.
The issues with FPTP are well documented, though it is of course a matter of opinion whether these constitute problems or not. The most commonly cited problem of FPTP is that the nationwide popular vote percentages do not map at all to how many seats are won by a party. I personally do not find this to be particularly problematic, because FPTP voting systems are fundamentally based on the philosophy that an MP is representing the people of their own constituency, not the whole nation. A more reasonable critique in my view is the fact that one can win an individual constituency with a mere plurality of the vote, sometimes less than 29%, if the opposition is sufficiently divided. In this case it is fair to argue that the MP is not representing the constituency very well because a majority of his constituents voted against him. This gives rise to systems such as the French two-round runoff vote and Alternative Vote/Instant Runoff Voting (AV/IRV) as used in Australia and a few other places.
However, I decided to take a slightly different tack, inspired by the fact that I have always been a little bit uncomfortable with the way modern British parliamentary constituencies are divvied up. You have a lot of situations where a 'natural community' is divided into three constituencies, and then you might have (for instance) Labour winning each constituency on 35-40% of the vote and getting all three seats, while there was clearly sufficient opposition across the community, not merely an individual constituency, to make these seem inequitable. I wanted a system which would allow 'natural communities' to be represented as a unit--whether it worked or not is a bit debatable.
Now the problems with proportional representation systems. There are obviously a wide variety of these and some are better than others (I tend to rather dislike the restrictive second-guessing element of STV for instance but we will not discuss ranked voting systems here). My major beef with proportional representation, besides the fact that it typically lacks such a strong constituency/community link, is the fact that one can have a dramatic shift in voters in a region or indeed across the country but its impact on a national level is often dampened. If the three big parties in a country both lose loads of votes to minor insurgent ones that fall below the threshold, their standing in parliament is barely altered. You can go from let's say 45%/35% for the first and second party in a region to flipping the other way around, a hugely dramatic (some would say too dramatic) impact under FPTP, yet under many PR systems that only represents a few seats changing hands and barely alters the parliamentary arithmetic. This is sometimes described by pro-FPTP advocates as lacking a 'throw the bastards out' mechanism (especially if said bastards can always get first place on a PR list by party machinations). It's not necessarily that fair an argument considering throwing the bastards out under FPTP is often harder than it sounds due to tribal voting, but it is worthy of consideration. There is also the point that PR will allow representation to a party that gets 5% nationwide but never breaks 10% in any particular region, which is counter to my personal view that representatives should be tied to a community--a particular ideology that can't get more than 10% support in any particular community should not be considered an integral part of it worthy of representation.
So basically I wanted a system where a 'nudge' to vote numbers will have a noticeable impact which makes people feel their votes count for something, but not one as overwhelming as under FPTP where a large number of seats can flip on a small change. I also wanted a system where the individual votes earned by an individual candidate have an impact on that candidate's chances, unlike the case under pure party-list PR where it is all about the party. Bloc vote has the latter, but suffers the same FPTP issue that all candidates of a party can grab the seats on offer based on a plurality of the vote. (Bloc vote percentages are usually calculated by the 'top vote' method, where you act as though only the votes earned by the top candidate of a party 'count' for summing purposes).
Therefore I produced Thandean Representation, which normally works like this:
This method returns results that look 'more reasonable' to my gut feeling.
If a party gets 71% of the vote across a 3-member constituency it seems reasonable for it to have all three seats with that overwhelming vote.
If a party gets only 49% and another party gets 35%, then it seems reasonable for the first party to get the first and third seats (rewarded for coming top) but the second party to get the second so its sizeable support base is represented.
If a party gets only 49% but the opposition splits between loads of parties with only ~10% each, then it seems reasonable for the first party to get all three seats because although it did not win a majority, there is no coherent, united strand of opposition that deserves representation in its own right, and it would be arbitrary to pick the opposition party that gets 12% rather than the ones who got 11%, 10%, 9%, 9%.
If the top three parties get 35%, 33%, 29% then it seems reasonable for all three to take a seat, because only a small nudge of the votes would be necessary to change this order and the leading party did not get much of a stand-out mandate worthy of additional representation.
Now these are just my gut feelings and you may well disagree, but these are the assumptions I put into this voting system.
Now, I have run the results of the 2010 and 2015 UK general elections through it in order to produce something to discuss. This was easier said than done. Unlike bloc vote councils elections which are easily converted as they usually start out with mostly 3-member wards, the general elections are of course single member FPTP constituencies. So how do you convert that? I combined 3 constituencies at a time to make a new larger 3-member constituency and then used all the OTL votes cast. There are a number of problems with this approach:
So this is very much an approximation. Realistically the UK would be split anew by the Boundary Commission into more natural 3-member constituencies and people would vote differently. However, I think it is still interesting to look at, and may even give some insights into below-the-radar trends with significance for considering the OTL elections.
Therefore, periodically I'm going to post studies and maps for individual parts of the UK here and how their elections would have gone differently under Thandean Representation. (This was partly inspired by Ares96's FPTP Sweden scenario which you should all look at).
But where to begin...?
*Definition of bloc vote for those unaware of it: you have a constituency/district/etc with N seats in it, the voters generally have N votes each, and can cast them for whichever candidates they want regardless of party, with the highest N candidates being elected. Used, among others, for many council elections in England and Wales, some state legislature elections in the USA, and formerly more common than not in many nationwide elections in the UK, Canada and USA.
**First-past-the-post, AKA Single Member Plurality or 'Most Votes Wins'; single member constituencies/districts where the candidate with the most votes wins, regardless of whether that represents a percentage majority or not. Currently used for general elections in the UK, Canada, India, Nigeria, etc. and for most elections in the USA.
The issues with FPTP are well documented, though it is of course a matter of opinion whether these constitute problems or not. The most commonly cited problem of FPTP is that the nationwide popular vote percentages do not map at all to how many seats are won by a party. I personally do not find this to be particularly problematic, because FPTP voting systems are fundamentally based on the philosophy that an MP is representing the people of their own constituency, not the whole nation. A more reasonable critique in my view is the fact that one can win an individual constituency with a mere plurality of the vote, sometimes less than 29%, if the opposition is sufficiently divided. In this case it is fair to argue that the MP is not representing the constituency very well because a majority of his constituents voted against him. This gives rise to systems such as the French two-round runoff vote and Alternative Vote/Instant Runoff Voting (AV/IRV) as used in Australia and a few other places.
However, I decided to take a slightly different tack, inspired by the fact that I have always been a little bit uncomfortable with the way modern British parliamentary constituencies are divvied up. You have a lot of situations where a 'natural community' is divided into three constituencies, and then you might have (for instance) Labour winning each constituency on 35-40% of the vote and getting all three seats, while there was clearly sufficient opposition across the community, not merely an individual constituency, to make these seem inequitable. I wanted a system which would allow 'natural communities' to be represented as a unit--whether it worked or not is a bit debatable.
Now the problems with proportional representation systems. There are obviously a wide variety of these and some are better than others (I tend to rather dislike the restrictive second-guessing element of STV for instance but we will not discuss ranked voting systems here). My major beef with proportional representation, besides the fact that it typically lacks such a strong constituency/community link, is the fact that one can have a dramatic shift in voters in a region or indeed across the country but its impact on a national level is often dampened. If the three big parties in a country both lose loads of votes to minor insurgent ones that fall below the threshold, their standing in parliament is barely altered. You can go from let's say 45%/35% for the first and second party in a region to flipping the other way around, a hugely dramatic (some would say too dramatic) impact under FPTP, yet under many PR systems that only represents a few seats changing hands and barely alters the parliamentary arithmetic. This is sometimes described by pro-FPTP advocates as lacking a 'throw the bastards out' mechanism (especially if said bastards can always get first place on a PR list by party machinations). It's not necessarily that fair an argument considering throwing the bastards out under FPTP is often harder than it sounds due to tribal voting, but it is worthy of consideration. There is also the point that PR will allow representation to a party that gets 5% nationwide but never breaks 10% in any particular region, which is counter to my personal view that representatives should be tied to a community--a particular ideology that can't get more than 10% support in any particular community should not be considered an integral part of it worthy of representation.
So basically I wanted a system where a 'nudge' to vote numbers will have a noticeable impact which makes people feel their votes count for something, but not one as overwhelming as under FPTP where a large number of seats can flip on a small change. I also wanted a system where the individual votes earned by an individual candidate have an impact on that candidate's chances, unlike the case under pure party-list PR where it is all about the party. Bloc vote has the latter, but suffers the same FPTP issue that all candidates of a party can grab the seats on offer based on a plurality of the vote. (Bloc vote percentages are usually calculated by the 'top vote' method, where you act as though only the votes earned by the top candidate of a party 'count' for summing purposes).
Therefore I produced Thandean Representation, which normally works like this:
Get a constituency/district/ward that elects three members; it might be an existing OTL one that does this under bloc vote. Ideally all constituencies should elect three members, with VERY rare exceptions allowed for specific circumstances like isolated islands. Just as under bloc vote, every voter gets three votes which they can cast accordingly for any of the candidates regardless of party--usually every party should stand three candidates, and for reasons explained later minor or independent candidates should probably have a full slate of three even if the other two are paper candidates.
Once the votes are counted, calculate percentages under one of two methods--this is the main difference between the two variants of Thandean Representation which will be discussed here.
THANDEREP-ALLVOTE counts all of the votes cast for all candidates of a party to produce a total party vote and then sums all the party votes for a turnout figure, dividing the first by the second to calculate a percentage vote for that party.
THANDEREP-TOPVOTE only counts the votes cast for the leading candidate of each party, then sums these for a turnout figure and derives a percentage by the same way.
This is the only difference between the two methods. ALLVOTE clearly discriminates against parties/independents with only one candidate, whereas TOPVOTE allows a fairer hearing. ALLVOTE is generally better for parties that can command some level of support across a wider area, while TOPVOTE rewards those who can command particularly strong support in a small area.
Regardless of the method used, the three seats are then filled up accordingly:
The party with the highest percentage gets the first seat. Its percentage is then divided in two.
The party with the highest percentage now gets the second seat. This might be the second placed party from the start, or it might be the first party again if it has more than twice as many votes as the second. If it is the second, that party's percentage is now divided in two. if it is the first again, its STARTING percentage is divided in three.
The party with the highest percentage after these changes gets the final seat. It could be the third placed party from the start, or it could be the first party again.
Another way of looking at this in party-list PR (thanks to Owen for suggesting this) is that a vote for a party constitutes a full vote for the first candidate, 1/2 a vote for the second, 1/3 for the third. But this isn't party-list PR and the candidates aren't ranked from the start, so who decides which of the party's three candidates gets to take a seat first? The answer is the voters. Whichever candidate got the highest individual votes to start with is the top candidate, whichever got the second highest is the second candidate, etc. Therefore, the individual who tops the poll is always guaranteed a seat.
This method returns results that look 'more reasonable' to my gut feeling.
If a party gets 71% of the vote across a 3-member constituency it seems reasonable for it to have all three seats with that overwhelming vote.
If a party gets only 49% and another party gets 35%, then it seems reasonable for the first party to get the first and third seats (rewarded for coming top) but the second party to get the second so its sizeable support base is represented.
If a party gets only 49% but the opposition splits between loads of parties with only ~10% each, then it seems reasonable for the first party to get all three seats because although it did not win a majority, there is no coherent, united strand of opposition that deserves representation in its own right, and it would be arbitrary to pick the opposition party that gets 12% rather than the ones who got 11%, 10%, 9%, 9%.
If the top three parties get 35%, 33%, 29% then it seems reasonable for all three to take a seat, because only a small nudge of the votes would be necessary to change this order and the leading party did not get much of a stand-out mandate worthy of additional representation.
Now these are just my gut feelings and you may well disagree, but these are the assumptions I put into this voting system.
Now, I have run the results of the 2010 and 2015 UK general elections through it in order to produce something to discuss. This was easier said than done. Unlike bloc vote councils elections which are easily converted as they usually start out with mostly 3-member wards, the general elections are of course single member FPTP constituencies. So how do you convert that? I combined 3 constituencies at a time to make a new larger 3-member constituency and then used all the OTL votes cast. There are a number of problems with this approach:
- UK parliamentary constituencies are less equal in electorate than in many countries
- Sometimes 3 constituencies make a natural community, e.g. off the top of my head Croydon, Doncaster, Ealing, Stoke, Hull, Newcastle. Others divide neatly into several 3-member constituencies, e.g. Cornwall, Gloucestershire. However, others do not fit together particularly well, and sometimes the nature of borders forced me to split up a natural community, which was rather aggravating.
- Unlike a bloc vote election, not everyone in this imaginary 3-member constituency had a vote for every candidate standing in it - people in OTL Doncaster Central like me were not able to vote for Ed Miliband in OTL Doncaster North.
- Under ALLVOTE this therefore discriminated against parties/independents who only stood in one of the three constituencies - but under TOPVOTE it is effectively throwing out loads of people's only vote, it's not like the proper system where everyone has more than one vote and they're just being pruned down.
So this is very much an approximation. Realistically the UK would be split anew by the Boundary Commission into more natural 3-member constituencies and people would vote differently. However, I think it is still interesting to look at, and may even give some insights into below-the-radar trends with significance for considering the OTL elections.
Therefore, periodically I'm going to post studies and maps for individual parts of the UK here and how their elections would have gone differently under Thandean Representation. (This was partly inspired by Ares96's FPTP Sweden scenario which you should all look at).
But where to begin...?