Tuesday, August 23, 2011

Hansard: Victorian State Parliament Electoral Matters Committee

Victorian State Parliament - Electoral Matters Committee
Extract from Hansard dated August 23, 2011

The CHAIR — Are you appearing in a private capacity or representing an organisation? If so, which organisation?

Mr van der CRAATS — I am a member of the Proportional Representation Society of Australia and of the Australian Labor Party, but the evidence that I give is of personal opinion.

The CHAIR — I ask you to begin your verbal submission, and we will take it from there.

Mr van der CRAATS — Thank you for the opportunity to speak to this Parliament. I do so with some reluctance, mainly because the last time I gave evidence to this committee I was subjected to what I consider to be harassment, intimidation and vilification by the chief electoral commissioner. It is a matter that I raised before with this committee, or the previous committee, and it failed to respond or to properly act. It is my belief that the actions taken by the chief electoral commissioner constitute contempt of Parliament, and I would like that this committee give further consideration to the complaints that I forwarded to the committee on previous occasions with the view of having the matter properly dealt with by an appropriate authority. I will raise this later in my submission, particularly with respect to item 4 and in relation to the role of the Ombudsman and the electoral commission as it is the Ombudsman that I believe is the appropriate body to review such complaints. However, this committee has a responsibility to ensure the integrity of the parliamentary process, that the committee process is intact and that witnesses are not subjected to some form of harassment or intimidation as a result of evidence given that may be detrimental to or critical of the conduct of the election’s process.

Having said those points of view, I wish to raise a number of issues, in particular in respect of the scrutiny of the ballot and the method of counting optional, preferential and other matters. My main concern at this stage is the scrutiny of the ballot and the way in which the Victorian Legislative Council election is counted. Scrutineers were denied access to the data entry data file as and when the data was being entered into the computer. This meant that the scrutineers had no means of properly validating the integrity of the data file. The information was only provided at the conclusion of the count and, as a result, it had to be taken at face value.

There is no justification or rationale as to why that information could not have been progressively published as the data was being recorded. In fact to have done so would have ensured that the process was open and transparent and that the scrutineers had a chance to monitor the data as it proceeded. It would have also ensured that the data had not been tampered with or altered as the election count progressed because people would have had copies and they could have gone back to see if any changes had occurred.

I think this is an important issue, and it is a fundamental one that the committee must address in respect of the drafting of regulations. It has become quite clear and apparent that the Victorian Electoral Commission is incapable, unwilling or does not desire to ensure that this process is open and transparent. This is a matter about which I have had some disagreement and conflict with the Victorian Electoral Commission going back nearly 20 years to when we first involved ourselves in the data entry computerised counting system. I think there is a trend towards computerised counting of elections, and I think that should be welcomed, but it should not be at the expense of openness and transparency. Obviously those candidates who win an election have no need to complain, but those candidates who feel that the system may not have properly fulfilled their requirements or that it was not as open and transparent as it should have been would be perceived as sore losers if they persisted in putting forward their complaints. I watched with great interest the conduct of the 2006 state election. I think many of the members of this Parliament are aware of the errors and the problems related to the way in which that was conducted and counted. There are still a number of unanswered questions in respect of the 2006 election which have not been properly addressed.

However, in reflecting on the 2010 election, there were some improvements. The data file that was eventually provided provided the information pertaining to the preliminary and secondary count. This was useful information and allowed some form of comparisons to the data quality of the data entry that was recorded. It was noted quite significantly that data quality in the 2010 election was significantly better than the data quality that occurred in the 2006 election. However, it was marred by the fact that the Victorian Electoral Commission failed to provide copies of that data file as it was progressively being built up. It was requested. A number of scrutineers I know from the Labor Party requested the Victorian Electoral Commission provide access to that data. They were refused this information. That, in my view, denied them the right to properly scrutinise the election.

It is probably incumbent on the Parliament to look closer at the roles and obligations of the scrutineer and what their role specifically is in terms of where and when they cannot interact with the counting process to oversee it. At the moment it is fairly well left open to the goodwill of the electoral commission to interpret what they consider to be fair or not fair. Unfortunately I do not believe the Victorian Electoral Commission has acted necessarily in good faith, and I do not believe that it has implemented procedures that ensure that the process is open and transparent. The publication of the data file progressively as the count progresses goes a long way to meeting that requirement and my concerns.

The next issue I wish to raise, which I have tried raising on previous occasions, is related to the method of counting that is under the legislation of the Electoral Act. There are a number of deficiencies and errors, in my view, in the way in which the election is counted. These deficiencies and errors came about as a result of a manual counting system. Under the manual counting system there were a number of shortcuts that were introduced into the way in which the count was made, and these were done to facilitate the speed, effort and manageability of a manual count. At the time they were considered to be a fair and reasonable compromise to make. However, with the advent and the use of computer technology we need to really review the methodology that we use to count the vote. There is no need to take these shortcuts. In fact the number of shortcuts that we have implemented really distort the outcome of the election process.

I think that is fairly evident when you analysed the Queensland 2007 Senate election. In that particular instance the segmentation of the vote, the way in which excluded candidates’ preferences were redistributed resulted in, in my view, the wrong candidate being elected to the last position on the Queensland Senate. In analysis of the vote that took place in 2007, if you took the last remaining seven candidates in the count and you redistributed the entire ballot paper according to only those seven candidates standing, the Greens candidate, Larissa Waters, should have been elected. This is an error of process due primarily to the shortcuts and segmentation that we implemented to try make it easy to do a manual count. They are no longer required under a data entry process.

I urge the committee to seriously consider the counting rules to make sure that they are a little more accurate and a little more representative of the voter’s intention. I have outlined those under the system which we have referred to as the ‘Wright’ system. Effectively it is a reiterative counting system that every time there is an exclusion you start the count again as though that candidate had never stood. It effectively mirrors the similar process that takes place in a single-member electorate, which I am sure you are all familiar with.

The other issue I wish to raise is optional preferential voting. The first time it was introduced was in 2010. In my view optional preferential voting is misleading in terms of the voter. The way in which the electoral commission sold and promoted optional preferential voting was that it in fact encouraged voters to put 1 to 5 only on the ballot paper. This disadvantaged and disenfranchised many voters. Those people who may have favoured a particular group or segment of the community stopped at the end of their 5 votes, and the result is that those who remained in the count had a greater say as to the outcome of that election.

I think optional preferential voting needs to be seriously looked at. Personally I do not agree with that. I think you are wasting your vote by exhausting your vote, particularly after no. 5. The aim, if your vote is not counted, is that you should have the opportunity to choose an alternative candidate. I think it is incumbent on the electoral commission to encourage voters to maximise their vote, not to cut it short, not to make it simpler for their data entry process. I think it would be more appropriate and more accurate if the Victorian Electoral Commission stated quite clearly that people should vote or indicate a preference for every candidate, and not just stop at 5. That is an issue I again urge the committee to look at.

In respect of the silent enrolment entitlement, I registered to vote on the day before the election day when I submitted my documentation at the polling place, which included a silent enrolment form. Unfortunately my documentation was lost and, as I understand it, my vote was not counted. I think this boils down to some concerns over the administrative processes that are embarked upon by the Victorian Electoral Commission. I believe that everybody has a right to nominate whether they wish to be on a silent enrolment or not. This would be best facilitated if there was just a tick-box on the application form. Anyone who wanted to enrolled to vote could decide whether or not they want their names listed. It is simpler, more straightforward, and it is certainly more cost effective.

Other issues and concerns relate to the overall duplication that is involved in maintaining a state electoral authority. What we are seeing in across-borders Australia is something like $35 million being spent on duplication of resources. We are seeing vast amounts of money being spent on development of software. It is software which is being duplicated in each state, each territory and each jurisdiction. There should be a lot more coordination and liaison going on. I also believe it would be probably more efficient if we had a centralised Australian Electoral Commission where the states had some input through the form of a board of directors or something like that. A single electoral authority can act professionally, with integrity and certainly have the cost benefits that would come with having a single organisation. Vast amounts of money have been spent on the electoral process, and we could save hundreds of millions of dollars by restructuring and reorganising the way in which the states’ electoral processes are delivered.

Point 4 is an issue of great concern to me. Currently under the Ombudsman Act the Victorian Electoral Commission is excluded from the Ombudsman reviewing it. This has implications with respect to freedom of information. It has implications with respect to people who have complaints about the administrative processes that are embarked upon by the Victorian Electoral Commission. The position of Ombudsman is in place to allow various administrative complaints to be considered outside the political arena. It is difficult, I understand, and this is the problem I think the previous committee had, to address issues that may be perceived to be of a political nature, particularly when dealing with the umpire — that is, the Victorian Electoral Commission. Having reviewed and considered the role of the Ombudsman, particularly in view of the number of complaints I have had in respect to the Victorian electoral commissioner, I believe the Ombudsman is the appropriate person that people should have the right to go to to have these issues properly investigated. I do not see any justification for prohibiting the Ombudsman from reviewing the role of the Victorian electoral commissioner on administrative matters. The Ombudsman has the role of reviewing the Victorian police commissioner, and I think he should have the same role of reviewing the Victorian electoral commission.

My final point of concern is in respect to the 2008 municipal elections. Again this was a failure of the previous committee. It failed to review or provide an opportunity for people to have some input into a review of the municipal elections. There are a number of commonalities that exist between the state and municipal elections but many of them have not been properly addressed. The disclosure of the data entry file as an election progresses is one of them. Municipal elections also have another aspect to them that state elections do not have, and that is the countback system. In reviewing the countback system for the City of Melbourne election which recently occurred as a result of Peter Clarke’s resignation, I became aware of a number of deficiencies in the technicalities of the way in which the countback process has been undertaken. Rather than take that up in detail — because I have not canvassed that particular issue in my submission — I would urge this committee as a matter of urgency prior to the 2012 municipal elections to undertake a review of the 2008 elections and see what comes out of that review.

The CHAIR — Thank you very much. Any questions?

Ms RYALL — Anthony, you mentioned the difference between data quality between the two — —

Mr van der CRAATS — Count A and count B.

Ms RYALL — Yes. What indicated to you that there was a difference?

Mr van der CRAATS — You can do a comparison between the first data and the second data, a direct comparison, and you can get an idea of where the preferences may have changed as a result of data entry error.

In the 2006 election the Victorian Electoral Commission refused to provide copies of the preliminary data entry, therefore it was impossible to make a comparison with the final data entry, so I had no idea where the changes occurred.

Ms RYALL — So when you suggested that there were data quality differences, that was an assumption. Is that correct?

Mr van der CRAATS — No, the records differ. You can see a very clear difference in the record data set itself, the preferences that are written there. For example, in some cases a preference might have been recorded as a ‘050’, or it may have been a ‘50’, or it may have been a ‘5’. I do not know how the systems interpret it; we did not have access to that information. The number of areas that we did identify in the 2010 election were not significant, certainly not enough to warrant a change or closer scrutiny of the vote itself. The time that would be required, obviously, to do that would not be justified by any benefit it would produce.

In the 2006 election that was not necessarily the case, particularly in Western Metropolitan Region. Not having access to that preliminary data and the secondary data, I think, was a major deficiency. The Victorian Electoral Commission at the time claimed that the preliminary data had been destroyed or overwritten. I find that extraordinary and very difficult to believe. As a systems IT person I would certainly have backup processes in place. If I have got data files that are worth hundreds of thousands or millions of dollars to collate, I would certainly have backup copies of that information. Had we been able to do a cross-examination of the preliminary data and the secondary data, we would have realised where the changes occurred. All I can go off is what was published by the Victorian Electoral Commission, and in 2006 there was a discrepancy of 250 of the total number of votes recorded between count A and count B went missing. I can understand that mistakes occur, but I would like to know where those mistakes did occur, and the information published by the electoral commission did not make that clear. It concerns me greatly because in 2006 the change of the result between the first count and the second count was something like — you would remember this very clearly — 150 votes. I do not know where they went, and I am a very fastidious person when it comes to looking at data.

Ms RYALL — Thank you. Just one more question. You spoke of shortcuts. Can you give examples of those shortcuts?

Mr van der CRAATS — When you have got a group of ballot papers and you are excluding a candidate and some candidates are already elected, you skip past those candidates and you deliver the ballot paper to the next available candidate at a higher value. In a proper analysis of the way in which the count should be taken, those votes actually should be contributed to the candidate who was originally elected, or previously elected, to form part of their quota and therefore part of their surplus. That is a shortcut that we took. We said, ‘They’re already elected; we’ll skip them and we’ll dump the vote directly at a lower candidate on the preference pile’. This has the effect of upping the value of that vote — it is no longer equal to the other votes which have contributed to the previous person’s election. That is one shortcut.

Another shortcut occurs in the way in which we distribute surpluses. We do similar things, we skip candidates.

We end up delivering votes in that process. It gets down to the segmentation. If we are distributing the votes, we stop at the point when we have distributed one particular pile of votes. For example, if a candidate has been excluded, they may have six different segmented vote piles: primary votes, secondary primary votes and subsidiary surplus. We distribute them each separately, and that can have a dramatic impact in terms of calculating the transfer value that takes place. This is all related of course to the upper house. These are shortcuts that facilitate a manual count. In a computerised count we can take the time quite readily because we are talking about 3 hours as opposed to 20 minutes to count an election using a computerised algorithm. Do you understand?

Ms RYALL — Thank you.

The CHAIR — There being no further questions, thank you very much, Mr van der Craats, for your time today and your contribution. A copy of the transcript will be coming your way in about a fortnight. Any typos that you discover in that transcript may be corrected but not matters of substance.

Mr van der CRAATS — Thank you.

Wednesday, August 10, 2011

Election: NSW Under Meek and Wright

Analysis of the NSW March 2011 State Legislative Council Ballot based on the detailed preference data published blatantly by the NSW Electoral Commission on August 3, 2011 indicated the following candidates are elected. (OpenSTV - Source Data)

| Count after transferring surplus votes. Keep factors of candidates who have exceeded the
| threshold: GALLACHER Mike, 0.081266; GAY Duncan, 0.088605; PEARCE Greg, 0.097127; CLARKE
| David, 0.107637; COLLESS Rick, 0.120694; MacDONALD Scot, 0.137060; CUSACK Catherine, 0.158742;
| MACLAREN-JONES Natasha, 0.188697; PHELPS Peter, 0.232255; BLAIR Niall, 0.302617; JOHNSTON
| Sarah, 0.432234; ROOZENDAAL Eric, 0.163670; DONNELLY Greg, 0.195524; SHARPE Penny, 0.242614;
| PRIMROSE Peter, 0.320513; KELLY Tony, 0.469467; FERGUSON Andrew, 0.870859; SHOEBRIDGE David,
| 0.335056; BARHAM Jan, 0.503748; and BROWN Robert, 0.919975. Candidate GREEN Paul has reached
| the threshold and is elected.

Winners are GALLACHER Mike, GAY Duncan, PEARCE Greg, CLARKE David, COLLESS Rick, MacDONALD Scot,

Likewise we under took analysis of the published data by excluding all but the last 22 candidates (Including Pauline Hanson) simulating the count  using the Wright System.  The results are the same - Detailed OpenSTV sheet sheet results.

To help you understand more how the system and method of counting the vote can and does distort the outcome of the election take a closer look at the 2007 Queensland Senate count.  

In 2007 Green's Candidate, Larissa Waters, was wrongfully denied representation.  The reason she was not elected was due solely to the distortion in the method of segmentation that is used in counting the Senate vote. A fact that Antony Green, ABC Electoral Analyst failed to understand or review. 

More information including count sheets using Meek and Wright methods of counting the vote.

Friday, August 5, 2011

Analysis: How easy is it to rig the outcome of an Electonic Election?

Question need to be asked "Just how easy is it to rig the NSW Legislative Council election?

The reality is its quite easy if you have access to the data file and no one else has copies of the data so a comparison cannot be made.

The NSW "Below-the-line" preference data fiels that habve just been released exclude preferences recorded as being informal. Votes where a preference has been omitted or duplicated. This could be as a result of a data-entry or voter error.

Without access to the missing data it is impossible to verify the quality of the data recorded.

What’s even more scary is that if a person had access to the original data file they could easily run a simple query against the data set, removing preferences for a given candidate where that candidate has a higher preference than another candidate. The number of primary votes would still be the same but the ballot paper would exhaust during the count if the preference order had been altered in any way.

Unless you are able to undertake a comprehensive audit or a random check it would be impossible to verify the correctness of the data recorded. The only means if validating the integrity of the election is to rely on the overall data network security and any log files that might provide a forensic trail to any wrong doing. Problem is this that access oog files are not available to scrutineers.

The risk of data being altered is increased if copies of the data files are denied or only made available after the conclusion of the count. All that would be distributed after the count is the altered data file with no means of checking that the data is infact correct

The preferred option and best means of protecting the integrity of the data file is to make copies of the file available, showing all preferences (Including informal and mismatched data-entry preference,) progressively throughout the data-entry process.

Publication of progressive compilation of the data limits the time and opportunity available to alter the data records, Scrutineers would have access to a copies and can monitor and perform random or structured data quality checks as the count progresses. Any close calls would signal warning and alert scruineers and analysts to pay closer attention in the review.

The fact that the data files have not been readily available during the count leaves the system wide open to potential misuse and abuse.

The Electoral Commissionm of course, prefers that no one has access to this information in which case there would be no means of challenge the accuracy and validity of the count. What they do not know can not be questioned.

One thing is clear, If we are to maintain confidence in the electronic counting of our votes the system needs to be designed as such that it is much more open and transparent. The current procedures and practices are far from open and transparent.

In the absence of Electoral Commission's implementing procedures that ensure that data is more readily available and subject to proper and fill scrutiny, Parliaments will have to review the detailed procedures and legislate to protect the system from possible corruption and abuse.

What’s needed is a comprehensive independent Parliamentary review of the processes and procedures in the conduct of electronic elections.

We certainly should not be complacent or lulled into a false sense of security thinking that the system is protected by itself, the fact is it is not.

They said the Titanic was unsinkable. History has proven otherwise. We only need to look at the mistakes made in the 2006 Victorian State Election to realise the extent of errors that can and do occur in an electonic counted election .

Thursday, August 4, 2011

NSW Electoral Commission recants original false and misleading advice.

In an extraordinary, yet not surprising move, the NSW Electoral Commission, having previously claimed that copies of NSW Legislative Council election preference data files were not available, has now recanted its previous advice and published the detailed results of the election.

This about face decision came some three months after the State election

At first the Commission claimed the data files did not exist, Then when we made a submission under the NSW Information Act the Commission refused to consider the application as the prescribed $30.00 application fee had not been paid.

A second application, which included the payment of the prescribed fee, was made along with a request for details on the software certification documentation. This second request was submitted on July 5, 2011 (Received on July 7, 2011).

The Commission under the provision of the NSW information Act had until July 26 to respond. Come July 27 no response was received. So we immediately lodged an application for review by the NSW Office of Information Commission expressing our ongoing concern at the Electoral Commissions misuse and abuse of process.

On August 1, 2011 (some five days late) the NSW Electoral Commission finally replied to our FOI application, stating that the information requested has been published by the Commission and was now available on the NSW Electoral Commission's Web site.

The location of this data is not easy to find as it is hidden away within the context of the Commission's summery  A direct link to the data can be found here.

Review of the data published indicates that this information was available late June but was not notified until August 2011.

In a further example of inefficientcy, and what we consider may be a further attempt to again avoid scrutiny and accountability, the information published was over 3Giga Bytes in size and excluded details of ballot papers that recorded duplicate preference numbers or had preferences missing. Why this information was excluded is anyone guess?

The Commission went to extraordinary steps to filter out and expand the data file so that the data file provided was excessively large, much larger then was required or necessary. Instead of one record per ballot paper, as requested, the commission demonstrated just how inefficient its data management is that it decided to produce a data file that included a single record per preference. Causing the data file to be some 100 times bigger than would otherwise be required.

It also turns out the Commission’s software had been modified after the2011 State election. Certification documentation provided, dated June 2011. indictates that amendments had been made to correct a number of errors in the original software.

The software, costing taxpayers 10's of millions of dollars, was developed in India. (Why Australian Developers were not used uis another issue?) Software that is, effectively, a duplication of software that already exists and is used by the Australian Electoral Commission.

OK, It is recognised that the NSW electoral provisions do differ from the Senate electoral provision but the cost of modifying the AEC software would not have been anywhere near the amount of money spent by the NSW Electoral Commission in having India . There are still unanswered questions as to who owns the intellectual property rights and copy right of the software developed?

After a three month delay we still do not have a full set of data. Some of the issuing or duplicate preference data could have been a result of data-entry errors. Without access to the full data set we are prevented from undertaking a full and more comprehensive analysis or review.
On a more positive side, more like an admission of guilt then a jester of good will, the Commission refunded the $30 application fee.

The question still remains why was this data not made available during the data entry process? Why has it taken them so long to publish the data? . And why did they opt to publish only a subset of the data and it in such an inefficient format?

We hope the mistakes of the Commission will not be repeated in future public elections and that Parliament takes a long and serious hard look at the role of the NSW Electoral Commission and legislation pertaining to the method of counting our votes.

Name Size Last Modified
File:2011 LC FP First Preference Results by District-Grp-Candidate v1.xls 1013 KB 3/08/2011 19:26:00
File:Functional Requirements for Vote Count v3 2 - changes accepted.pdf 310 KB 19/07/2011 17:03:00
File:PRCC Fn Spec v3.1 Certificate of Legislative Compliance.pdf 23 KB 19/07/2011 17:37:00
File:PRCC Fn Spec v3.2 Certificate of Legislative Compliance - Final.pdf 164 KB 29/07/2011 13:09:00
File:PRCC LC Birlasoft Test Certificate v3.2.pdf 181 KB 20/07/2011 13:38:00
File:PRCC LG Birlasoft Test Certificate v3.2.pdf 177 KB 20/07/2011 13:38:00
File:Readme.doc 34 KB 13/07/2011 19:32:00
File:SGE 2011 LC Preferences.zip 636988 KB 29/06/2011 16:44:00