Tuesday, 18 May 2010

Counting votes in the UK general election

This blog post is just to try and communicate something that won’t fit into 140 characters on twitter. It is a simplified version of events and some things have been left out.

  1. Voting
    • The electorate vote by hand onto paper ballots
    • electronic and mechanical voting and counting is not in use at general elections and has been widely unpopular when trailed or used in other elections. Electronic voting in particular has been shown to be unsafe.
    • Currently in our elections are first past the post and we make a mark (usually an X) in box for the candidate we wish to elect.
  2. Verification
    • Filled in ballots are counted and this count compared with the number of votes recorded as cast by the polling station.
    • This number is subtracted from the number of issued ballots and the unused papers checked and accounted for.
    • This count happens face up so that counting agents for all candidates can scrutinise the process and get a first set of electoral data.
  3. Sorting
    • Ballots are sorted into piles for each candidate.
    • A further pile of ballots that require adjudication because they are unclear, identify the voter, or potentially spoilt is made.
    • This process is again carried out face up to be scrutinised and for counting agents to get their second slice of electoral data.
  4. Counting
    • Piles are counted and sorted into bundles.
    • Bundles are counted.
    • Each of these processes is subject to scrutineering and checking.
  5. Adjudication and Reconciliation
    • The papers put aside in step three are adjudicated by the Returning Officer and he then informs the candidates/agents in sight of observers of rejections. His decisions can be objected to but not over-ruled. Accepted papers added to count.
    • The total numbers of papers for each candidate and the rejections are added up to ensure they equal the number from step 2.
    • Totalling is not restricted to those with physical access to the ballots as it can be seen, in fact bundle recounts can be done aloud.
  6. Result or recount
    • The result is made know to the candidates/agents at this point in the case of a close election they can ask for a recount.
    • Once the Returning Officer is happy that the result stands it is announced.
  7. If we used AV/STP
    • In the case of a future move to a redistributive preferential voting system after the first round of voting the second round could just start again from step three with re-allocation of surplus or next preference votes (depending on the voting system) done physically and using a method such as Hare. This means the process is transparent and obvious to counting agents and other observers.

How can this process be adapted for range voting while retaining the key points of paper ballots filled with a pencil, manual transparent counting of votes, no calculations beyond how many papers in a pile and how many piles do we have?

19 comments:

broken ladder said...

Score Voting and Approval Voting are objectively much simpler than STV in some objective ways (the following excerpt written by a Princeton math Ph.D. and voting methods expert, named Warren Smith).

A. Write a range voting computer program and an IRV computer program (preferably with error-checking of the inputted votes). The score voting program will be shorter and will run faster, assuming essentially any reasonable programmer does it. (This, called "Kolmogorov Complexity" is the standard objective metric used by scientists to assess "simplicity.")

B. Score voting runs on all today's (FPTP) voting machines without any modification (including non-computerized machines). IRV does not.

C. Voters experimentally make fewer ballot-invalidating errors when using score voting than when using IRV. Voters apparently find SV easier, as they screw it up less often.
http://scorevoting.net/SPRates.html

D. Not simple enough for you? Okay, score voting is a parameterized class of methods, with the parameter being the number of ratings. The simplest kind of score voting is called "approval voting." It has only two ratings, Yes and No. Approval Voting is absolutely the simplest major voting system reform possible. It requires no changes to ballot forms; all it requires is eliminating the "no-overvote" rule, thus actually simpifying the rules versus now.

-- http://scorevoting.net/Lorenzo.html

It is of virtually no practical benefit that you can count STV ballots by stacking them in piles in each round. You have look at every single ballot to see which pile to put them into in the first place - it's not like they magically stack themselves and then you can quickly count the number of votes for each candidate by just eyeballing those stacks.

You have to make little tick marks on paper or via some mechanical/electronic device.

It is true that with Score Voting and Approval Voting, this has to be done for multiple candidates per ballot, not just one. That is an increase in the time it takes to sum the values. But it is a linear increase in work, and precincts still have totals that they can announce. SV/AV are "additive" so you can just add up the precinct sub-totals and know who your winner is, in one round.

With STV (and its single winner form called Alternative Vote in the UK, and IRV in the USA) you cannot have precinct subtotals. (And it's possible for X to win an IRV election based on the ballots at every single precinct, but to lose when they are all summed together - bizarre.)

If there are multiple rounds (which is a frequent occurrence with our IRV elections here in San Francisco, you have to count all the votes for the first round, and then announce the first-round totals, and then report to a central authority who sums up all the precinct total.

Then those totals are announced, telling the precincts who to eliminate, so they can know how to redistribute the votes. You then go to the pile of the eliminated candidate and redistribute his ballots to the other piles based on second-place rankings, while repeating the process of tick-marks or key presses on a counting machine.

You then announce your new ballot totals.

You then wait for the central authority to hear from all precincts and report back. Again.

You repeat this process until you have a "majority winner".

broken ladder said...

This process can take a long time. After our November 2008 elections in San Francisco, if you went to our city website, you would see this message FOR SEVERAL WEEKS.

"Due to the requirement that all ballots must be centrally tallied in City Hall and not at the polling places, the Department of Elections has not set a date for releasing any preliminary results using the ranked-choice voting method."

So don't tell me that Score Voting is tough to count. It's not. Any form of STV is tougher.

Oh, and IRV elections are more likely to lead to near-tie election debacles.
http://scorevoting.net/TieRisk.html

Clay Shentrup
San Francisco, CA
206.801.0484
clay@electology.org
http://www.electolog.org/

Tony Kennick said...

No they don't "magically stack themselves" we have humans that sort them, did you read the post before getting your copy and paste on?
We don't have "precinct subtotals" totals by wards in parliamentary constituencies only come from counting agents. Since all the ballots for an seat in a UK general election are counted together the issues you keep throwing up for AV don't exist.
The use of piles of paper that people can see on both sides of the counting table is a strength of the UK system and virtually anyone would want it kept in any voting reform.

Range Voting is a good theoretical system but needs either technology in the voting/counting process or would dramaticly increase counting time. Nothing you have said refutes this binary situation. As there are scores of reports about how the use of technology in voting opens it up to fraud or doubt, the UK does not want this, so in order to adopt range voting we would end up with very long counts.

kowey said...

In French elections (two round FPTP), I think they stick a slip of paper in an envelope (with the candidate's name on it).

At the very least, it seems like that methodology can be easily adapted to approval voting (stick as many slips of paper as candidates you approve of).

I guess it'd be trickier for range voting unless you want to start telling people they can use up to three slips per candidate or something. Fat envelopes.

kowey said...

Ugh, ambiguity. Let me rephrase that.

Think they stick a slip of paper (bearing the candidate's name) into an envelope. And then they drop the envelope into a ballot box.

I think by rights you're supposed to pick up slips for *all* the candidates (so that nobody knows who you're voting for), but I seem to remember people making a point by very conspicuously only picking out the slip they want. Could be wrong though

broken ladder said...

No they don't "magically stack themselves" we have humans that sort them, did you read the post before getting your copy and paste on?

Yes, humans have to count and sort them. That was exactly my point.

We don't have "precinct subtotals" totals by wards in parliamentary constituencies only come from counting agents. Since all the ballots for an seat in a UK general election are counted together the issues you keep throwing up for AV don't exist.

Are you saying that the normal FPTP ballots are all brought to a central location for counting, rather than having the counting agents count subtotals at individual polling stations and then report them to a central authority? That would seem highly unusual.

Even if it is true, the point remains that STV's non-additive nature presents yet another point of mathematical complexity to the method. You can't just sum the ballots like you can with FPTP or Score Voting and Approval Voting.

The use of piles of paper that people can see on both sides of the counting table is a strength of the UK system and virtually anyone would want it kept in any voting reform.

A) You haven't explained what benefit it offers, so what is your justification for claiming that it's a strength? Or that UK voters would feel some attachment to those per-candidate piles?

B) I thought the point of contention was the complexity/cost/feasibility of counting these respective methods. Your point here does nothing to change the fact that counting STV ballots is simply more complex. Even if having those piles has some benefits that you will go on to explain for us, that doesn't negate the complexity issue.

Range Voting is a good theoretical system but needs either technology in the voting/counting process or would dramaticly increase counting time.

As I explained, you have this completely backward.

Score Voting can be counted in a linear fashion, where you go ballot-by-ballot and add the scores for each candidate, then move on. This obviously takes longer than FPTP, but it's a linear increase in complexity. And in places that use voting machines, Score Voting can be tabulated on them with no upgrades.

STV, on the other hand, cannot be. Hence it seems to spur the adoption of (more fraud-prone) electronic voting machines.

Remember this headline?

The voting machines: Scotland switches to computerized voting - Scottish officials followed the advice of "experts" including the non profit "The Electoral Reform Society" and switched from hand counted paper ballots to computerized voting in May '07 to support STV, a form of Instant Runoff:
-- http://www.instantrunoffvoting.us/scotland.html

With IRV, every time a candidate is eliminated, you have to go back through and count the second-place preferences for the ballots of an eliminated candidate.

As there are scores of reports about how the use of technology in voting opens it up to fraud or doubt, the UK does not want this, so in order to adopt range voting we would end up with very long counts.

This is an argument in favor of Score Voting. Score Voting is easier to count without the adoption of complicated and/or fraud-prone technology.

This is not a contentious issue. It is just accepted as a point of fact by experts in the field.

broken ladder said...

Oh, how did I forget this link.
http://scorevoting.net/Complexity.html

Worried about the complexity of range voting?

It is astonishing to see government officials who are perfectly fine with making every citizen fill out complicated tax forms which can exceed 100 pages, tell us, "oh no, for your own good I have to protect you from the horrifying prospect of a 'complicated' voting system like Range Voting. You poor thing, you might not be able to handle the concept of scoring every candidate 0-to-99."

Get back to reality.

Australia and Ireland have both been successfully using STV voting since the 1910s and 1920s, and Australia for their elections for senators switched to reweighted STV in 1949 – which Ireland had been using all along.

Think about that. This was before the days of calculators and computers. In Australia, voting is compulsory (you get fined if you do not vote) and a full rank ordering of all the candidates is required on each ballot. That is harder for voters than range voting, with blanks (Xs) permitted, would be.

Then, after the votes have been collected, the counting proceeds. In reweighted STV, we first count all the top-rank votes, then find the candidate with the fewest and eliminate him. Steps of this nature (and another nature – declaring "winners" those candidates with above the "Droop quota" of top-rank votes) are repeated – and between these steps each vote is reweighted depending on how it voted for previous winners. Each cycle we redo the whole process with all the previous winner and loser candidates eliminated from all votes. In one Australian election there were 72 candidates and hence there were 71 such cycles to be performed, and each voter had to rank all 72 candidates in order.

That is a pretty complicated process. What really makes it tricky is that each vote has its own individual weight, which keeps changing throughout the process according to a formula involving multiplication, subtraction, division, and truncation to integers, and depending on the individual characteristics of that vote and what the set of previous winners is.

And all this has been going on, successfully, since the 1920s in the days before without calculators or computers, in Ireland and Australia.

So don't complain to us that range voting is complicated. It isn't. You just add up numbers to find sums – and if blanks are allowed you have to divide by the number of nonblank votes at the end to get averages. There is no vote reweighting or re-sorting. There are hardly any divisions or multiplications. Voters are allowed to leave lots of entries intentionally blank (i.e. fill them with an "X") or give them all zero scores. If the Australians and Irish could handle a much more complicated procedure – more complicated both for the voter and for the counters – than this in the 1920s and 1940s (and they like it – e.g. the Irish have been asked twice in referenda if they wanted to drop their voting system and return to the simpler plurality system, and both times they voted to keep it), then now, in the age of computers and calculators, we can handle a simpler procedure: range voting.

And not only that, range voting can be handled by today's plurality machines even without modification. (Try demo!) That instantly makes us better off than the Australians and Irish, who, in order to use voting machines, had to get special machines specially designed just for their kind of voting. That means the switch from plurality to range can be made comparatively painless.

And here's the bottom line: experimental fact. We point out that range voting has smaller spoilage-inducing error rates than either plurality or IRV voting. Experimentally. In other words, voters – at least in this respect – regard range voting as simpler than the present plurality voting system! So the thinking that range voting is "more complicated" is misleading: that complexity actually works in such a way as to reduce ballot-spoiling errors.

Tony Kennick said...

broken ladder:
Yes, humans have to count and sort them. That was exactly my point.
I am saying this is a good thing, you seem to think it is a bad thing. We have a system that while far, far from perfect is not going to get get dragged through the courts like the US presidential of 2000 or the Minnesota United States Senator of 2008. We would like to keep it that way.

Are you saying that the normal FPTP ballots are all brought to a central location for counting, rather than having the counting agents count subtotals at individual polling stations and then report them to a central authority? That would seem highly unusual.

Even if it is true

I am saying, it isn't unusual it is normal and how dare you suggest I am am telling you anything other than the truth seen with my own eyes. ;-)

You said on twitter "This has absolutely nothing to do with what country you're in" and then dismiss important facts about how we do things.

In the recent UK wide general election parliamentary counts were run by the local authority each seat is in or mostly in. So for example Sheffield City County counted six seats for the UK parliament. If you would like to understand more about how we administer elections I can point you at all the guidance for election officials online.
Can I ask you how the distributed count in precincts is ensured to be be free and fair?
Take note that while there were 25 different parties in the last elections for the US House of Representatives there were 47 in the last UK elections so some parties are very small (and of course truly independent candidates are one man bands). These small parties can't put representatives in the near two hundred polling stations of a large rural constituency but can be at a central count for the seat.

Tony Kennick said...

You haven't explained what benefit it offers
I keep explaining it is transparent and obvious, all parties get to watch, nothing happens hidden inside a machine or on a calculator.

The fact that members of every party standing for election and other observers can watch the UK system from start to finish and can see the result is correct and fair without specialist equipment is a tremendous strength.

Score Voting can be counted in a linear fashion, where you go ballot-by-ballot and add the scores for each candidate, then move on
How can that happen in a transparent, checkable, repeatable way that can be subject to scrutineering and checking not involving machines in any way?

Thank you for mentioning Scotland 2007 by the way, although you use it as an argument against STV (which I am not arguing for) it is a key agument against electronic counting, do be careful though it was counting they did electronically, the ballots were still cast on paper (although there was at the same time an electronic voting trial in some places in England )

I love how you can both say "Score Voting is easier to count without the adoption of complicated and/or fraud-prone technology" and "then now, in the age of computers and calculators, we can handle a simpler procedure: range voting" and not see the dichotomy.

Finally, you keep using arguments against STV and AV against me, you will note I don't argue back. In fact your last comment from "Complexity" is almost entirely on that footing. I don't want to try and claim STV or AV or FPTP is batter than range voting, they are not, but as no-one has yet demonstrated how to count range voting in a transparent, checkable, repeatable way that can be subject to scrutineering and checking that does not involve machines in any way it is of no use.

I ask one last time for a HOWTO manual on running range voting for a 10 candidate election using paper ballots filled in with pencils.
Where the counting methodology is transparent, checkable, repeatable and can be subject to scrutineering and checking. That at no time has votes or calculations hidden in calculators, electronic tabulating machines, voting machines or anything else that reduces the ability for a party or candidate to see the election count happen in front of them. This system also needs to allow for recounts and ballot paper adjudication.
The answer should in no way draw comparisons with AV or STV as I will happily concede that range voting is on paper a better system than them and anything is better than FPTP.

Tony Kennick said...

kowey:
Interesting will have to look into that, it might just have one possible answer to this conundrum.

broken ladder said...

I ask one last time for a HOWTO manual on running range voting for a 10 candidate election using paper ballots filled in with pencils.
Where the counting methodology is transparent, checkable, repeatable and can be subject to scrutineering and checking. That at no time has votes or calculations hidden in calculators, electronic tabulating machines, voting machines or anything else that reduces the ability for a party or candidate to see the election count happen in front of them. This system also needs to allow for recounts and ballot paper adjudication.
The answer should in no way draw comparisons with AV or STV as I will happily concede that range voting is on paper a better system than them and anything is better than FPTP.


Essentially the same way as with FPTP. Say you have a hypothetical two-race (for the sake of simplicity/clarity) election for Dog Catcher and Restaurant Inspector.

The FPTP ballot might look like:
Dog Catcher: Jones (X) - Smith (_) - Douglas (_)
Restaurant Inspector: Adams (_) - Thomas (X) - Lewis (_)

So, going down that list, you add 1 to the tally sheet for Jones, and 1 to the tally sheet for Thomas.

With Score Voting it might look like:
Dog Catcher: Jones(10) - Smith (8) - Douglas (_)
Restaurant Inspector: Adams (0) - Thomas (10) - Lewes (0)

So, going down that list, you add 10 to the tally sheet for Jones, and 8 for Smith, and 10 for Thomas.

A little more work, but from an election integrity standpoint, nothing is significantly different. No calculators are needed. As I pointed out, STV is much more complex and was used in Australia in the 1910's, without calculators or computers of course.

In fact, because it's more difficult/detectable if you spoil a Score Voting ballot, we think it may BETTER resist fraud. For instance, say a Florida Republican decides to add a vote for Douglas if he sees that a FPTP ballot was cast by a black man. Now the ballot is "spoiled", so his vote for Jones doesn't count. That's hard with Score Voting. He could give Douglas a 10, since the voter left Douglas blank, but if Douglas was Jone's strongest opponent, then this would look very strange, and would make the fraud slightly more obvious (especially with statistical analysis across multiple ballots).

A bigger point is that fraud is a relatively minor concern compared to bad voting methods.

Bayesian Regret calculations show that Score Voting increases election representativeness as much in comparison to FPTP as FPTP increases it relative to random totally undemocratic selection. That is, Score Voting is as big an improvement as the invention of democracy.

Fraud may occasionally tilt elections to the wrong winner in close elections, but the net effect of it is very small in comparison with the net effect of having bad voting methods such as FPTP. So I think your priorities are objectively out of order.

Here's a page on the relative importance of various voting reforms. It entails a lot of estimation, but generally based on sound evidence.
http://ScoreVoting.net/RelImport.html

broken ladder said...

I love how you can both say "Score Voting is easier to count without the adoption of complicated and/or fraud-prone technology" and "then now, in the age of computers and calculators, we can handle a simpler procedure: range voting" and not see the dichotomy.

There is no dichotomy.

1) Score Voting is easier than STV/IRV to hand-count, because it is simpler.

2) In today's society, people are much more exposed to sophisticated tools like computers and calculators, and generally have to use their logic and reasoning skills more often, e.g. the secretary who used to have to type on a typewriter now has no navigate a host of computer applications and occasionally deal with technically oriented error messages, or the 5-year-old who learns to program the DVR to record his favorite cartoon.

The combined point is, if people from the days before the Internet/television/smartphone could handle a fairly complicated system like STV/IRV, then people in the much more technologically advanced world of today should certainly be able to handle a simpler method like Score Voting.

I am saying this is a good thing, you seem to think it is a bad thing.

No, I never said it was a bad thing to manually count ballots. You suggested somehow it was beneficial to have those ballots in piles. My point is that each individual ballot still has to be inspected in order to tabulate them. So I don't see what these piles do for you. And you can put Score Voting ballots in piles too - so I must be missing important detail about these piles that makes them more helpful in FPTP elections.

You said on twitter "This has absolutely nothing to do with what country you're in" and then dismiss important facts about how we do things.

Let me be clear. That STV requires central tabulation is an mathematical certainty, and has nothing to do with what country you're in. You didn't deny that, but instead argued that it didn't matter because the ballots are already centrally tabulated under the FPTP system. Now you are seeming to backpedal to a position that central tabulation "isn't unusual". My point remains, if you have some polling stations which are currently able to pass up subtotals, that will effectively end under STV. That's worse for integrity.

broken ladder said...

In the recent UK wide general election parliamentary counts were run by the local authority each seat is in or mostly in. So for example Sheffield City County counted six seats for the UK parliament.

You're not answering the question. Does the counting always happen in one central location per election, or are sub-totals reported to a central authority for simple summing?

Can I ask you how the distributed count in precincts is ensured to be be free and fair?

I never claimed that it is guaranteed to be free and fair. The point is that, whatever probability there may be for fraud at the polling-site level, central tabulation introduces additional fraud because of e.g. chain-of-custody issues. Problems like these:
http://www.youtube.com/watch?v=PKQEQ7qHvgM

Further, a fraudster at a central tabulation site has access to more ballots than a fraudster at polling site or precinct. And as cynical as it may sound, some of the fraud at local levels is canceled out; but the higher up the aggregation hierarchy you go, the less impacting that cancellation effect is, and thus the more likely that the fraud changes the election outcome.

Take note that while there were 25 different parties in the last elections for the US House of Representatives there were 47 in the last UK elections so some parties are very small (and of course truly independent candidates are one man bands). These small parties can't put representatives in the near two hundred polling stations of a large rural constituency but can be at a central count for the seat.

Okay, say your party is too small to have an election observer at a polling location, and you can either:

1) Have the ballots counted at that polling station, where the Conservative and Labor and LibDem observers will at least (presumably) see to it that your party's votes aren't fraudulently given to either of the other two "competitive" parties. Then the sub-totals are (publicly) reported to the central tabulation authority, or

2) Have those ballots physically delivered to the central location, ala Butch and Hoppy style from the video above.

Granted, UK elections probably are significantly more sane than that New Hampshire election in the Youtube video -- but I'm not sure that the election observer argument is sufficient to show that central tabulation is better. This one specific issue is admittedly harder for me to address without living in the UK and observing UK elections.

broken ladder said...

One of the people in my group mentioned some interesting things about your system.
===

There was a report by the Electoral Commission that included a section
on the central counting of votes vs. precinct counting. Apparently,
this is entirely a matter of custom. In Britain, a Returning Officer
is appointed for each constituency (perhaps multiple constituencies),
and is charge of conducting the entire election. The Electoral
Commission report mentioned that election officials in continental
Europe thought it was really strange that in Britain precinct-level
counting is not used, while in Britain, they think it really odd that
counting would be done at the precinct level. But it also noted that
one returning officer had used precinct-level counting for one
election, and that it had seemed to go OK.
Some constituencies, particularly those in Sunderland, make a big
effort to complete their count in the minimum amount of time. So the
BBC will show vehicles with a police escort rushing to the counting
center, and then runners bringing the ballot box into the tables used
for counting.
One feature of the central county is that it makes it a very public
process, so that the candidates are expected to attend. The
candidates are brought onto a stage, and then step forward when their
vote total is announced. At one time, party affiliation of candidates
was not officially recognized, and in the BBC coverage the returning
officer would announce, "Joe Smith ... 2048 votes" and in the brief
pause the BBC announcer would quickly interject "Liberal Democrat" or
whatever party.

Tony Kennick said...

So, going down that list, you add 1 to the tally sheet for Jones, and 1 to the tally sheet for Thomas.

Except we don't do use tally sheets for counting, if you re-read section 4 counting above, piles are counted into bundles (of 25, 50, 100 whatever), bundles are counted.
I'll say that again, as you keep saying you don't understand what the bundles are for, in FPTP Ballots are sorted into piles for each candidate, these piles are counted into bundles, those bundles are counted to give a total (with the left over small pile of ballots not big enough to make a bundle.

Now the party counting agents who come along from the political party watching from the other side of the table use tally sheets as they don't have physical access to to the ballots and will take that information away to their party huddles, but the counting staff use physical multiples.
each individual ballot still has to be inspected in order to tabulate them
Nope as you already know who all the votes in that bundle are for.

So, going down that list, you add 10 to the tally sheet for Jones, and 8 for Smith, and 10 for Thomas.
Tally sheets are on the face of it are far more open to fraud, over or under tallying for example and they don't have the transparency for candidates and their agents at counts that I keep saying is important.

say a Florida Republican decides to add a vote for Douglas if he sees that a FPTP ballot was cast by a black man
Is that a feasible in your elections?
I am horrified that that is possible.
How can someone know who cast a ballot and have access to it?

Score Voting is easier than STV/IRV to hand-count, because it is simpler.
I don't care how much easier more difficult is compared with two other voting systems, I want to know about Score Voting's openness and transparency at the count.
Both STV/IRV can be counted using relatively simple extensions of the piles of paper ballots system I have outlined for FPTP

I never claimed that it is guaranteed to be free and fair
Then I am not interested, I am only interested in methodologies that maintain or improve a free and fair election and count.

Tony Kennick said...

a fraudster at a central tabulation site has access to more ballots than a fraudster at polling site or precinct

The ballot boxes are unsealed and the count proceeds in full view of the candidates, their agents, their counting agents, the police, the press and independent observers that is a lot of people to have to get onside to commit fraud.

While there is nothing wrong in letting the scrutiny of the major parties secure your share of the vote, if you want to isn't entirely abhorrent I don't want a system where you have to.

ala Butch and Hoppy style
I didn't watch the video about New Hampshire, so not sure who Butch and Hoppy are but the reason I am happy with the chain-of-custody is the police escort you mention while talking about Sunderland doesn't just happen there, local constabularies take electoral security very seriously and are heavily involved in ensuring the integrity of the ballot box.
This is especially true if ballots have to be stored for any reason.

Talking of Sunderland don't take their theatrics in getting thier result out as any more significant as a baromiter of a UK election that Hart's Location or Dixville Notch are in the US.
There was a report by the Electoral Commission that included a section on the central counting of votes vs. precinct counting
Can you give me a link to this report I can't find it on http://www.electoralcommission.org.uk/
One feature of the central county is that it makes it a very public process
Yes, the whole count is watched by many people this is part of the openness, admittedly it does also include the theatrics of the declaration at some point over night, but that isn't the point of all those people being there.

broken ladder said...

I am only interested in methodologies that maintain or improve a free and fair election and count.

Then your priorities are gravely mistaken.

Using FPTP instead of Score Voting is as harmful to your democracy as if you were to switch from FPTP to completely non-democratic random selection, say using a name randomly drawn out of a hat.

Score Voting is generally much better than STV/IRV in terms of transparency and fraud issue,s but even if that were not the case due to atypical aspects of the UK counting process (as you have described), it is untenable any realistic increase in fraud could negate that improvement.

we don't do use tally sheets for counting, if you re-read section 4 counting above, piles are counted into bundles (of 25, 50, 100 whatever), bundles are counted.

Okay, for the sake of simplicity, say I have 32 ballots for a candidate, going into bundles of 10. I can either:

A) Count 1,2,3..10 *bundle*, 1,2,3..10 *bundle*, 1,2,3..10 *bundle*. 10, 20, 30, 31 32 = 32 ballots, or

B) Count 1,2,3..32.

The latter is actually less counting. How are these piles helping?

Now, with Score Voting, say I had:
Ballot: Fourier=10, Bernoulli=3, Bohr=0
Ballot: Fourier=4, Bernoulli=0, Bohr=10
Ballot: Fourier=10, Bernoulli=0, Bohr=7
...

So I count
Fourier 10, Bernoulli 3, Bohr 0.
Fourier 14, Bernoulli 3, Bohr 10.
Fourier 24, Bernoulli 3, Bohr 17.
...

One pass and we're done, whereas with STV you often have to recount lots of ballots due to reallocation.

each individual ballot still has to be inspected in order to tabulate them
Nope as you already know who all the votes in that bundle are for.


Each individual ballot obviously has to be inspected, to know what pile it goes into.

Tally sheets are on the face of it are far more open to fraud, over or under tallying for example and they don't have the transparency for candidates and their agents at counts that I keep saying is important.

How is tallying in one's head less prone to fraud than doing so on paper?

say a Florida Republican decides to add a vote for Douglas if he sees that a FPTP ballot was cast by a black man
Is that a feasible in your elections?
I am horrified that that is possible.
How can someone know who cast a ballot and have access to it?


I'm exaggerating, but it seems our fraud situation is vastly worse than yours. Say the polling place is in a black neighborhood - you just do this to lots of random ballots, knowing they are much more likely to be Democrat voters. Or if its just a precinct that you know is heavily tilted in favor of a party you dislike.

How do you think George Bush got elected?

I don't care how much easier more difficult is compared with two other voting systems, I want to know about Score Voting's openness and transparency at the count.
Both STV/IRV can be counted using relatively simple extensions of the piles of paper ballots system I have outlined for FPTP


I'm still waiting for you to explain how these piles add any transparency or other notable benefit.

broken ladder said...

The ballot boxes are unsealed and the count proceeds in full view of the candidates, their agents, their counting agents, the police, the press and independent observers that is a lot of people to have to get onside to commit fraud.

What if they were tampered with en route, or replaced with lots of fake ballots and placed into new boxes?

Is that really inconceivable given that Labor was widely alleged to have disproportionately disenfranchised voters in Conservative-leaning areas?

While there is nothing wrong in letting the scrutiny of the major parties secure your share of the vote, if you want to isn't entirely abhorrent I don't want a system where you have to.

Well, the alternative is that you HAVE TO have the ballots be transported to a different location. Why do you have more confidence in the transporters than in a group of election inspectors from the major parties, who at least have an incentive to protect your vote to make protect their own interests?

local constabularies take electoral security very seriously and are heavily involved in ensuring the integrity of the ballot box.

Then why not just have those local constabularies observe the ballots as they are counted at the polling location, rather than transporting them elsewhere for counting?

And why do you trust them so much to take electoral security "very seriously"? Do you think they are apolitical angels?

You seem to have this preconceived notion that the current system is excellent, even though some of these details seem arbitrary to an outsider such as myself, and don't inherently strike me as being the best way to ensure integrity.

You apparently take it for granted that these constabularies are more trustworthy than who ever might get their mitts on the ballots if they were counted at the precincts.

There was a report by the Electoral Commission that included a section on the central counting of votes vs. precinct counting
Can you give me a link to this report I can't find it on http://www.electoralcommission.org.uk/


I ask him to link you to it here. Check back in a bit later and he may have an answer for you, or may even post it here.

His name is Jim Riley and on certain electoral issues, he has a brain like an encyclopedia. He made these videos:
http://www.youtube.com/watch?v=fOwDyGCaOFM

One feature of the central county is that it makes it a very public process
Yes, the whole count is watched by many people this is part of the openness...


I think you get election integrity by having that public process, with bipartisan election observers and transparent ballot boxes and such. I don't think these piles are doing much for you.

Tony Kennick said...

I am only interested in methodologies that maintain or improve a free and fair election and count.
Then your priorities are gravely mistaken.
This isn't a green field situation, setting up a new society so we can choose the best of everything. We have a long standing system that makes our elections for the most part free, fair, transparent and subject to scrutiny.
There is currently a movement for change to our voting system to one that represents the people better, but we cannot allow for this to take us down a retrograde path in terms of fraud. You said it seems our fraud situation is vastly worse than yours so you can see why I don't want to change to a system where fraud can go up.

Okay, for the sake of simplicity, say I have 32 ballots for a candidate
All well and good but not if you have 35,000 ballots for a candidate out of a total turnout of 50,000 it doesn't really scale. which brings me on to How is tallying in one's head less prone to fraud than doing so on paper? and 'm still waiting for you to explain how these piles add any transparency or other notable benefit. no one tallies in their heads they use bundles, yes they count up to the number that has been decided in each pile but that is it and that counting is checked and scrutinised..
If I point you at a table and you see twenty bundles of papers and you know because you watched them being counted and checked that each bundle has 100 votes for Kate Bobson the independent candidate, how many votes (to the nearest hundred as there will of course be the "and the rest" pile) does Kate Bobson have?
Can you see the transparency there, you have watched the ballots being verified, separated, counted and bundled up, you can see the bundles the process is beautifully simple.
My fundamental point is that there are many other voting systems that can be counted in this way but range voting isn't one of them.

The End

As the newspapers say "This correspondence is now closed"