THE HARE SYSTEM OF PROPORTIONAL DELEGATION.
The single transferable vote, generally known as the Hare system, was
first invented by a Danish statesman, M. Andrae, and was used for the
election of a portion of the "Rigsraad" in 1855. In 1857 Mr. Thomas
Hare, barrister-at-law, published it independently in England in a
pamphlet on "The Machinery of Representation." This formed the basis of
the scheme elaborated in his "Election of Representatives," which
appeared in 1859.
He proposed to abolish all geographical boundaries by constituting the
whole of the United Kingdom one electorate for the return of the 654
members of the House of Commons. Each member was to be elected by an
equal unanimous number of electors. The method of election was therefore
so contrived as to allow the electors to group themselves into 654
constituencies, each group bound only by the tie of voluntary
association, and gathered from every corner of the Kingdom. The total
number of votes cast (about a million) was to be divided by 654, and the
quotient, say about 1,500, would be the quota or number of votes
required to elect a member. But some of the candidates would naturally
receive more votes than the quota, and a great many more would receive
less. How were all the votes to be equally divided among 654 members so
that each should secure exactly the quota? The single transferable vote
was proposed to attain this result. Each elector's vote was to count for
one candidate only, but he was allowed to say in advance to whom he
would wish his vote transferred in case it could not be used for his
first choice. Each ballot paper was, therefore, to contain the names of
a number of candidates in order of preference--1, 2, 3, &c. Then all the
candidates having more than a quota of first choices were to have the
surplus votes taken from them and transferred to the second choice on
the papers, or if the second choice already had enough votes, to the
third choice, and so on. When all the surpluses were distributed a
certain number of members would be declared elected, each with a quota
of votes. The candidates who had received the least amount of support
were then to be gradually eliminated. The lowest candidate would be
first rejected, and his votes transferred to the next available
preference on his ballot papers; then the next lowest would be rejected,
and so on till all the votes were equally distributed among the 654
members. Such was the Hare system as propounded by its author. The
electors were to divide themselves into voluntary groups; then the
groups which were too large were to be cut down by transferring the
surplus votes, and the smaller groups were to be excluded and the votes
also transferred until the groups were reduced to 654 equal
constituencies. These two processes, transferring surplus votes and
transferring votes from excluded candidates, are the main features of
the system. Mr. Hare's rules for carrying them out are drawn up in the
form of a proposed electoral law, and in the different editions of his
work the clauses vary somewhat. They are also complicated by an
impossible attempt to retain the local nomenclature of members. As
regards surplus votes it was provided that the ballot papers which had
the most preferences expressed should be transferred; still a good deal
was left to chance or to the sweet will of the returning officer, and
this has always been admitted as a serious objection. The process of
elimination is still more unsatisfactory. Mr. Hare was from the first
strongly opposed to the elimination of the candidate who had least first
preferences, and he therefore proposed that, in order to decide which
candidate had least support, all expressed preferences should be
counted. This involved such enormous complication that in the 1861
edition of his work he abandoned the process of elimination altogether
in favour of a process of selection. He then proposed to distribute
surplus votes only, and to elect the highest of the remainder,
regardless of the fact that they had less than a quota. He then
wrote:--"The reduction of the number of candidates remaining at this
stage of the election may be effected by taking out the names of all
those who have the smallest number of actual votes--that is, who are
named at the _head_ of the smallest number of voting papers, and
appropriating each vote to the candidate standing _next_ in order on
each paper. This process would be so arbitrary and inequitable in its
operation as to be intolerable. It might have the effect of cancelling
step by step more votes given to one candidate than would be sufficient
to return another.... Such a process disregards the legitimate rights
both of electors and of candidates." But the process of selection was
not proportional representation at all, being practically equivalent to
a single untransferable vote, and Mr. Hare finally adopted, in spite of
its defects, the "arbitrary and inequitable" process of elimination in
his last edition in 1873. And all his recent disciples have been forced
to do the same, because nothing better is known.
Mr. Hare's scheme has ceased to be of any practical interest, since it
is now generally admitted that electorates should not return more than
ten or twenty members. Moreover, it is admitted that the electors would
group themselves in very undesirable ways, and not as Mr. Hare expected.
And yet the only effect of limiting the size of the electorates is to
reduce the number of undesirable ways in which electors might group
themselves. Let us briefly note the different proposals which have been
made.
+1. Sir John Lubbock's Method.+--In his work on "Representation," Sir
John Lubbock says:--"The full advantage of the single transferable vote
would require a system of large constituencies returning three or five
members each, thus securing a true representation of opinion."
Three-seat electorates are, however, too small to secure accurate
proportional representation; with parties evenly balanced, for instance,
one must secure twice as much representation as the other.
The following rules are given to explain the working of the system:--
(1) Each voter shall have one vote, but may vote in the alternative for
as many of the candidates as he pleases by writing the figures 1, 2, 3,
&c, opposite the names of those candidates in the order of his
preference.
COUNTING VOTES.
(2) The ballot papers, having been all mixed, shall be drawn out in
succession and stamped with numbers so that no two shall bear the same
number.
(3) The number obtained by dividing the whole number of good ballot
papers tendered at the election by the number of members to be elected
plus one, and increasing the quotient (or where it is fractional the
integral part of the quotient) by one, shall be called the quota.
(4) Every candidate who has a number of first votes equal to or greater
than the quota shall be declared elected, and so many of the ballot
papers containing those votes as shall be equal in number to the quota
(being those stamped with the lowest numerals) shall be set aside as of
no further use. On all ballot papers the name of the elected candidate
shall be deemed to be cancelled, with the effect of raising by so much
in the order of preference all votes given to other candidates after
him. This process shall be repeated until no candidate has more than a
quota of first votes or votes deemed first.
(5) Then the candidate or candidates having the fewest first votes, or
votes deemed first, shall be declared not to be elected, with the effect
of raising by so much in the order of preference all votes given to
candidates after him or them, and rule 4 shall be again applied if
possible.
(6) When by successive applications of rules 4 and 5 the number of
candidates is reduced to the number of members remaining to be elected,
the remaining candidates shall be declared elected.
Objection is commonly taken to this method on account of the element of
chance involved in the distribution of surplus votes. Suppose the quota
to be 1,000, and a candidate to receive 1,100 votes, the 100 votes to be
transferred would be those stamped with the highest numerals. But if the
hundred stamped with the lowest numerals or any other hundred had been
taken the second choices would be different.
Strictly speaking, however, this is not a chance selection--it is an
arbitrary selection. The returning officer must transfer certain
definite papers; if he were allowed to make a chance selection it would
be in his power to favour some of the candidates.
Sir John Lubbock points out that the element of chance might be
eliminated by distributing the second votes proportionally to the second
choices on the whole 1,100 papers, and that it might be desirable to
leave any candidate the right to claim that this should be done if he
thought it worth while.
+2.--The Hare-Clark Method.+--The Hare system has been in actual use in
Tasmania for the last two elections. It is applied only in a six-seat
electorate at Hobart and a four-seat electorate at Launceston. The rules
for distributing surplus votes proportionally were drawn up by Mr. A.I.
Clark, late Attorney-General. The problem is not so simple as it appears
at first sight. There is no difficulty with a surplus on the first
count; it is when surpluses are created in subsequent counts by
transferred votes that the conditions become complicated. Mr. Clark
adopts a rule that in the latter case the transferred papers only are to
be taken into account in deciding the proportional distribution of the
surplus. Suppose, as before, the quota to be 1,000 votes, and a
candidate to have 1,100 votes, 550 of which are marked in the second
place to one of the other candidates. Then the latter is entitled to 50
of the surplus votes, and a chance selection is made of the 550 papers.
The element of chance still remains, therefore, if this surplus
contributes to a fresh surplus.
+3.--The Droop-Gregory Method.+--This method, advocated by Professor
Nanson, of the Melbourne University, is claimed to entirely eliminate
the element of chance. The Gregory plan of transferring surplus votes is
defined as a fractional method. If a candidate needs only nine-tenths of
his votes to make up his quota, instead of distributing the surplus of
one-tenth of the papers all the papers are distributed with one-tenth of
their value. Reverting to our former example, if a candidate is marked
second on 550 out of 1,100 votes, the quota being 1,000 and the surplus
100, then instead of selecting 50 out of the 550 papers, the whole of
them would be transferred in a packet, the value of the packet being 50
votes, or, as Professor Nanson prefers to put it, the value of each
paper in the packet being one-eleventh of a vote. Should this packet
contribute to a new surplus the third choices on the whole of the papers
are available as a basis for the redistribution. The packet would be
divided into smaller packets, and each assigned its reduced value. It
might here be pointed out that the use of fractions is quite
unnecessary, the value of each packet in votes being all that is
required, and that the-same process may be used with the Hare-Clark
method to avoid the chance selection of papers. The only real
difference is this: that when a surplus is created by transferred votes
Mr. Clark distributes it by reference to the next preference on all the
transferred papers, and Professor Nanson by reference to the last packet
of transferred papers only--the packet which raises the candidate above
the quota.
Which of these methods is correct? Should we select the surplus from all
votes, original and transferred, as Sir John Lubbock proposes; from all
transferred votes only, with Mr. Clark; or from the last packet only of
transferred votes, with Professor Nanson? Consider a group of electors
having somewhat more than a quota of votes at its disposal. If it
nominates one candidate only every one of the electors will have a voice
in the distribution of the surplus, but if it puts up three candidates,
two of whom are excluded and the third elected, Mr. Clark would allow
those who supported the two excluded candidates to decide the
distribution of the surplus, and Professor Nanson only those who
supported the last candidate excluded. Both are clearly wrong, for the
only rational view to take is that when a candidate is excluded it is
the same as if he had never been nominated and the transferred votes had
formed part of the original votes of those to whom they are transferred.
Whenever a surplus is created it should therefore be distributed by
reference to all votes, original and transferred. As regards these
surpluses, Mr. Clark and Professor Nanson have adopted an arbitrary
basis, which is no more than Sir John Lubbock has done; and they have
therefore eliminated the element of chance only for surpluses on the
first count. It may be asked, Why cannot all surpluses be distributed by
reference to all the papers, if that is the correct method? The answer
is that the complication involved is enormous. Yet this was the plan
first advocated by Professor Nanson, who wrote, in reply to a definite
inquiry how the Gregory principle was applied:--"I explain by an
example. A has 2,000 votes, the quota being 1,000. A then requires only
half the value of each vote cast for him. Each paper cast for him is
then stamped as having lost one-half of its value, and the whole of A's
papers are then transferred with diminished value to the second name
(unelected, of course). The same principle applies all through. Whenever
anyone has a surplus all the papers are passed to the next man with
diminished value." Now, the effect of this extraordinary proposal would
be that the whole of the papers would have to be kept in circulation
till the last candidate was elected, with diminishing compound
fractional values. In a ten-seat electorate a large proportion would
pass through several transfers, and would towards the end of the count
have such a ridiculously small fractional value that it would take
several millions of the ballot-papers to make a single vote! It is no
wonder that this method was abandoned when the complications to which
it would lead were realized.
A simple method of avoiding this complexity would be to treat
transferred surplus papers as if the preferences were exhausted. It must
be remembered that in all transfers a certain number of papers are lost
owing to the preferences being exhausted, and the additional loss would
be small. Thus at the first Hobart election 206 votes were wasted, and
this number would have been increased by two only. Every surplus would
then be transferred by reference to the next choice, wherever expressed,
on both original papers and papers transferred from excluded candidates.
It might be provided, however, for greater accuracy that all papers
contributing to surpluses on the first count only should be transferred
in packets. Should these contribute to a new surplus, it should be
divided into two parts, proportional to (1) original votes and votes
transferred from excluded candidates, and (2) the value of the packet in
votes. Each part would then be distributed proportionally to the next
available preferences wherever expressed. To divide the packets into
sub-packets is a useless complication. The loss involved in neglecting
them would usually be less than one-thousandth part of the loss due to
exhausted papers.
Having now dealt with the main features of the different variations of
the Hare system, we may proceed to consider some details which are
common to all of them. A difference of opinion exists, however, as
regards the quota. Sir John Lubbock and Professor Nanson advocate the
Droop quota, which we have shown to be a mathematical error; Miss Spence
and Mr. Clark use the correct quota.
+The Wrong Candidates are Liable to be Elected.+--The Hare system may be
criticised from two points of view; first, as applied to the conditions
prevailing when it is introduced, and, secondly, as regards the new
conditions it would bring about. Its advocates confine themselves to the
first point of view, and invariably use illustrations based on the
existence of parties.
We readily grant that if the electors vote on party lines, and transfer
their votes within the party as assumed, the Hare system would give
proportional representation to the parties; but even then it would
sacrifice the interests of individual candidates, for it affords no
guarantee that the right candidates will be elected. The constant
tendency is that favourites of factions within the party will be
preferred to general favourites. This at the same time destroys party
cohesion, and tends to split up parties. Nor can this result be wondered
at, since the very foundation of the system is the separate
representation of a number of sections.
One reason why the wrong candidates are liable to be elected is that the
electors will not record their honest preferences if the one vote only
is effective. They will give their vote to the candidate who is thought
to need it most, and the best men will go to the wall because they are
thought to be safe. Mr. R.M. Johnston, Government Statistician of
Tasmania, confirms this view when he declares--"The aggregate of all
counts, whether effective or not, would seem to be the truer index of
the general favour in which each candidate stands, because the numbers
polled at the first count may be greatly disturbed by the action of
those who are interested in the success of two or more favourites who
may be pretty well assured of success, but whose order of preference
might by some be altered if sudden rumour suggested fears for any one of
the favoured group. This accidental action would tend to conceal the
true exact measure of favour in the first count." If this statement
means anything it is that the three preferences which are required to be
expressed should have been all counted as effective votes at the Hobart
election instead of one only; and this is exactly what we advocate. It
is also admitted that when two candidates ran together at the first
Launceston election the more popular candidate was defeated; and again
the _Argus_ correspondent writes of the recent Hobart election:--"The
defeat of Mr. Nicholls was doubtless due to the fact of his supporters'
over-confidence--nothing else explains it. Many people gave him No. 2
votes who would have given him No. 1 votes had they not felt assured of
his success."
A second reason why the wrong candidates are liable to be elected is
that the process of elimination adopted by all the Hare methods has no
mathematical justification. The candidate who is first excluded has one
preference only taken account of, while others have many preferences
given effect to. We have shown that this glaring injustice was
recognized by Mr. Hare, and only adopted as a last resort. Professor
Nanson admits that "the process of elimination which has been adopted by
all the exponents of Hare's system is not satisfactory," and adds--"I do
not know a scientific solution of the difficulty." To bring home the
inequity of the process, consider a party which nominates six
candidates, A, B, C, D, E, and F, and whose numbers entitle it to three
seats, and suppose the electors to vote in the proportions and order
shown below on the first count.
FIRST SECOND THIRD FOURTH
COUNT. COUNT. COUNT. COUNT.
7-vote ADEFBC ADEBC AEBC ABC
6-vote EFDACB EDACB EACB ACB
5-vote CEBDFA CEBDA CEBA CBA
4-vote BDFACE BDACE BACE BAC
4-vote DCEFBA DCEBA CEBA CBA
3-vote FBAECD BAECD BAEC BAC
It will be noted that F, having fewest first votes, is eliminated from
the second count, D from the third count, and E from the fourth. A has
then 13 votes, B 7, and C 9. If the quota be 9 votes, A's surplus would
be passed on to B, and A, B, and C would be declared elected. But D, E,
and F are the candidates most in general favour, and ought to have been
elected. For if any one of the rejected candidates be compared with any
one of the successful candidates it will be found that in every case the
rejected candidate is higher in order of favour on a majority of the
papers. Again, if the Block Vote be applied, by counting three effective
votes, the result would be--A 10 votes, B 12, C 9, D 21, E 22, and F 13.
D, E, and F would therefore be elected. Thus we see that A, B, and C,
the favourites of sections within the party, are elected, and D, E, and
F, the candidates most in general favour--those who represent a
compromise among the sections--are rejected.
In practice, then, the Hare system discourages compromise among parties,
and among sections of parties; and therefore tends to obliterate party
lines. This has already happened in Tasmania, where all experience goes
to show that the Hare system is equivalent to compulsory plumping. In
every election the result would have been exactly the same if each
elector voted for one candidate only. The theory that it does not matter
how many candidates stand for each party, since votes will be
transferred within the party, has been completely disproved. Votes are
actually transferred almost indiscriminately. The candidates have not
been slow to grasp this fact, and at the last election handbills were
distributed giving "explicit reasons why the electors should give their
No. 1 to Mr. So-and-so, and their No. 2 to any other person they
chose."[7] Three out of every four first preferences are found to be
effective, but only one out of every five second preferences, and one
out of fifty third preferences. The first preferences, therefore, decide
the election.
The actual result is that, in the long run, the Hare system is
practically the same as the single untransferable vote. The whole of the
elaborate machinery for recording preferences and transferring votes
might just as well be entirely dispensed with. The "automatic
organization" which it was to provide exists only in the calculations of
mathematicians.
+A Number of Votes are Wasted.+--It is claimed for the Hare system that
every vote cast is effective, because it counts for some one candidate.
But unless every elector places all the candidates in order of
preference some votes are wasted because the preferences become
exhausted.
When a paper to be transferred has no further available preferences
expressed it is lost. In order to reduce this waste, a vote is held to
be informal in the six-seat electorate at Hobart unless at least three
preferences are given. Notwithstanding this, the number of such votes
wasted was 7 per cent, at the first election and 10 per cent, at the
second.
The effect of this waste is that some of the candidates are elected
with less than the quota. At the last Hobart election only three out of
six members were elected on full quotas, and at Launceston only one out
of four. The result is to favour small, compact minorities, and to lead
sections to scheme to get representation on the lowest possible terms.
The Droop quota, being smaller than the Tasmanian quota, would have the
effect of electing more members on full quotas, and it is often
recommended on that account. Indeed, Professor Nanson declares:--"In no
circumstances is any candidate elected on less than a quota of votes.
The seats for which a quota has not been obtained are filled one after
the other, each by a candidate elected by an absolute majority of the
whole of the voters. For the seats to be filled in this way all
candidates as yet unelected enter into competition. The matter is
settled by a reference to the whole of the voting papers. If any
unelected candidate now stands first on an absolute majority of all
these papers he is elected. But if not, then the weeding-out process is
applied until an absolute majority is obtained. The candidate who gets
the absolute majority is elected. Should there be another seat, the same
process is repeated. If an absolute majority of the whole of the voters
cannot be obtained for any candidate, then the candidate who comes
nearest to the absolute majority is elected." It will be seen that
Professor Nanson proposes to bring to life again all the eliminated
candidates, in order to compete against those who have less than the
quota. The proportional principle is then to be entirely abandoned, and
the seats practically given to the stronger party, although the minority
may be clearly entitled to them. The vaunted "one vote one value" is
also to be violated, because those who supported the elected candidates
are to have an equal voice with those still unrepresented. And finally,
the evil is not cured, it is only aggravated, if an eliminated candidate
is elected.
+The Hare System is not Preferential.+--The idea is sedulously fostered
that the Hare system is a form of preferential voting, and many people
are misled thereby. The act of voting is exalted into an end in itself.
The most elaborate provisions are now suggested by Professor Nanson to
allow the elector to express his opinion only as far as he likes. The
simple and practical method in use in Tasmania of requiring each elector
to place a definite number of candidates in order of preference is
denounced as an infringement of the elector's freedom. Why force him to
express preferences where he does not feel any? The Professor has
therefore invented "the principle of the bracket." If the elector cannot
discriminate between the merits of a number of candidates he may bracket
them all equal in order of favour. Indeed, where he does not indicate
any preference at all, the names unmarked are deemed equal. Therefore,
if he does not wish his vote transferred to any candidate, he must
strike out his name. It is pointed out that a ballot paper can thus be
used if there is any kind of preference expressed at all, and the risk
of informality is reduced to a minimum. All the bracket papers are to be
put into a separate parcel, and do not become "definite" till all the
candidates bracketed, except one, are either elected or rejected; the
vote is then transferred to that candidate. And as bracketed candidates
will occur in original papers, surplus papers, and excluded candidates'
papers at every stage of the count, the degree of complication in store
for the unhappy returning officer can be imagined.
The whole of these intricate provisions are founded on a patent fallacy.
Preferences are not expressed in the Hare system, as in true
preferential voting, that they may be given effect to in deciding the
election, but simply in order to allow the elector to say in advance to
whom he would wish his vote transferred if it cannot be used for his
first choice. The elector is allowed to express his opinion about a
number of candidates, certainly, but after being put to this trouble
only one of his preferences is used. And which one is used depends
entirely on the vagaries of the system. The principle of the bracket
illustrates this fact; if the elector has no preference the system
decides for him. If his first choice just receives the quota the other
preferences are not even looked at. Again, of all the electors who vote
for rejected candidates, those who are fortunate enough to vote for the
worst (who are first excluded) have their second or third preferences
given effect to, and few of their votes are wasted; but the votes of
those who support the best of them (who are last excluded) are either
wasted or given to their remote preferences. In Mr. Hare's original
scheme, for instance, the votes of the last 50 candidates excluded would
have been nearly all wasted, unless some hundreds of preferences were
expressed.
Another claim on which great stress is laid is that by the process of
transferring votes every vote counts to some one candidate. This means
nothing more than that the votes of rejected candidates are transferred
to the successful candidates. Where is the necessity for this? So long
as each party secures its just share of representation and elects its
most favoured candidates, there is no advantage gained by transferring
the votes. Miss Spence even declares that "every Senator elected in this
way will represent an equal number of votes, and will rightly have equal
weight in the House. According to the block system, there is often a
wide disparity between the number of votes for the highest and the
lowest man elected." Surely the mere fact of transferring votes till
they are equally distributed does not make all the successful candidates
equally popular! On the contrary, it is very desirable to know which
candidates are most in favour with each party.
+Ballot Papers Must be Brought Together for Counting.+--This is a
practical objection to the Hare system, which puts it out of court for
large electorates. If the whole of Victoria were constituted one
electorate, as at the Federal Convention election, the transference of
votes could not be commenced till all the ballot papers had come in from
the remote parts of the colony, two or three weeks after the election.
On this point Professor Nanson writes:--"In an actual election in
Victoria this 'first state of the poll' could be arrived at with the
same rapidity as was the result of the recent poll on the Commonwealth
Bill. In both cases but one fact is to be gleaned from each voting
paper. The results from all parts of the colony would be posted in
Collins-street on election day. These results would show exactly how the
cat was going to jump. The final results as regards parties would be
obvious to all observers, although the result as regards individual
candidates would be far from clear. But this, although of vast
importance to the candidates themselves, would be a matter of small
concern to the great mass of the people." These remarks are based on the
assumption that the electors vote on strictly party lines, which a
reference to Tasmanian returns will show is not usually the case. Few
will be disposed to agree that a knowledge of the successful candidates
is a matter of small moment.
FOOTNOTE:
[7] _Hobart Mercury_