This is Ian Brightwell's Typepad Profile.
Join Typepad and start following Ian Brightwell's activity
Join Now!
Already a member? Sign In
Ian Brightwell
Recent Activity
Antony I just wanted to comment on the suggestion above to utilise the website lookup. Although this is possible and I have done it with my personal iPhone the interface on the website is made difficult to use for privacy reasons. Basically you have to put in all your name and address data perfectly before it will identify if you are enrolled and identify your electorate. This is not a good interface for a polling place official with a queue of people to deal with and does not help people who may not be sure of some of their enrolment details. The roll lookup devices used by states are very flexible and let you find people through a range of approaches ie. DoB, suburb, given name only, etc. This is particularly useful for some people who have very common surnames or have multiple names for a range of reasons. You would be surprised how many people do not know which name they used to enroll let alone which of the last few addresses they have lived at they may be enrolled. COMMENT: That's my view as well. The AEC's roll check page is very specific on what you are required to enter. For obvious privacy reasons it won't allow general checking of the roll.
1 reply
Antony so glad you have picked up this issue. You are absolutely correct in saying the lack of electronic roll lookup devices is a major cause of ballot issue error. Having worked as an OIC at the last election at both a prepoll and polling place without lookup devices I had the almost comical situation of asking people where they are enrolled. The simple fact is most don't know, they think they know but are often wrong. As you know NSW introduced iRolls for 2007 election and Qld and Victoria jumped on board with the NSWEC soon after. These states have recently refreshed their fleet of shared devices with 6,500 tablets. There is no excuse for the AEC not providing a roll lookup device in every polling place given the obvious high error rate at the 2016 election. I do not believe they can reasonably use security as an excuse, which has been the AEC's past reason for not using these devices.
1 reply
Antony as usual I am going to take as different view on this issue. That is not to say I do not believe your blog and many of the comments above are valid – they are. The point which seems to have been lost in the senate voting reform discussions is that it is impossible under legislation used in 2013 for the AEC to reliably run the senate election count process. The simple fact is the senate system with tickets and full preferencing is an UNSTABLE SYSTEM. What I mean by unstable is simply that a very small input change can cause a large and unpredictable output change i.e. 15 in 1.3M votes changed in WA and causing 2 out of 6 candidates to change!!!!!! So what would have happened if the AEC had to continue running the election under the 2013 legislation. My guess is that every election we would have had a crisis which may not have been able to be resolved without a recount or a rerun election. Intending no criticism of the AEC or any EMB in this situation this situation is not a good look for those responsible for governance of the electoral process. The fact is no EMB can count the number of votes involved in senate election (NSW has 4.5M+ votes) manually without making small counting and handling errors it is just not possible to perform manual operations this accurately. The newly legislated processes will reduce the need for extreme accuracy by creating larger gaps between candidates at exclusion due in part exhausting votes. This situation greatly improves the ability of the AEC to count senate elections without recounts or worse rerun elections. So how likely was it for the WA situation to have happened again. I would simply point to the rerun election to answer that question. In the rerun election at count 197 (out of 256 counts) the difference between the excluded candidates votes and the continuing candidate was ONE vote. Fortunately, for the AEC these two candidates had tickets which aligned at that point in the count thus making the impact of the closeness of the exclusion irrelevant. However, this was only luck, so we really had a system which just relied on luck. You have to ask yourself if it is reasonable for the AEC to confidently state for any future election that one vote could not have been miscounted for any two candidates in question. If they cannot convince the public of this the accuracy of their count and the tickets diverge at the point where the count is close, then a recount may need to be done and we all know what can happen when a recount is done for a senate election.
1 reply
Antony - Good to see senate first preference initial counting has been added into the bill. Still very strange that a minimum of 6 ATL preferences remain. One would assume that if this stays then ALL ballot papers would have to be batched and data entered as only a small percentage would use saving provisions and enter a single ATL preference. I have done some rough numbers and believe the labour cost of batching and data entry of all senate papers Australia wide would be about $30M with proposed voting instructions which could be reduced to $8M if the voting instructions are changed to allow a single preference ATL. Not sure if anyone cares how much it costs but to me this difference is not insignificant given I do not believe it makes any difference electorally.
1 reply
I note Daniel's comment about NSW data entry being centralised for legislative council. This is done more for scrutiny purposes for a state wide election than any other reason. Note we use the same PRCC system for local government Council elections and operate from many RO offices around the state, again to allow convenient scrutiny. It always fascinated me that the AEC were able to leave most ballots back in RO offices for a state wide senate election. Certainly NSWEC experience has been that you need to have every ballot paper (including unused) from a voting centre in one location to reconcile and batch for data entry. Any thing less leads to staff making invalid reconciliation assumptions at the time of batching. COMMENT: The WA Senate re-count revealed a level of errors in the original count that should be seen as unacceptable.
1 reply
The thought of manual counting or data entering a 100% of senate papers would be terrifying for the AEC. I agree with Michael that manual counting the senate in NSW with the newly proposed minimum 6 above the line preferences is not viable. NSWEC gave up manual counting local government councillor elections because of a general lack of skill in the community and evidence of mistakes being made. So it beggers belief anyone would seriously consider manually counting 5M ballot papers in NSW. Even data entry of 100% of all ballots would be very challenging and no doubt require the AEC to develop a new system or adopt and modify the NSW PRCC system. These are not trivial issues and the AEC needs time to get the solution in place and procedures and training done regardless of which proposal is picked. So it is good to see that this issue may be settled soon in legislative terms.
1 reply
Antony you are correct to say in 2012 data entry was done in country councils in the RO office close to the council by the Commission, which seemed to satisfy all the councils involved. The same will happen in 2016. There were at least three councils in 2012 who ran their own election and did a manual PR count (Sutherland, Gunnedah and Lane Cove) at least one of these had to redo it. Interestingly, this council appears to be doing a manual count again at the coming election. One thing we noticed with councils that did manual counts was that they did not provide very detailed results which made it hard to determine the efficacy of the count. Certainly at the 2008 election where we did a lot of the country counts manually we experienced challenges of the nature Michael described. That is why we moved to full computer counting. COMMENT: If computers are going to be used, you can go the whole hog and use whatever transfer value is thought best. It might need an alternative be specified if manual counts are done.
1 reply
On the question of scanning. NSW has an obvious problem with upper house ballots. Put simply they are just too big. The only scanners which can do the job are plan scanners and they are far too slow and expensive. At the NSW local government election in 2012 the alternate election supplier to the NSWEC scanned the Councillor ballots and I believe had a lot of problems. They certainly took much longer to do their 12 Councils than NSWEC did to process 136 Councils by keying the below the line and multiple preference above the line. Note the Councillor ballots are smaller than NSW senate for the most part so a bigger problem for the Senate. Final point, NSW keyed all their Legislative Assembly ballots in 2015, that's how we got the data file of preferences. It was a success in that it took no longer and gave a more reliable result which allowed much more flexible reporting. We did again look to scan these ballots but the cost of settling up 186 scanners (you need at least two for each office to allow for breakdowns) in 93 locations and supporting them. Plus dealing with a more complex workflow than keying was not attractive.
1 reply
Antony I read with interest you analysis of the informality of iVotes verse other votes and it is true that iVotes were 4th lowest compared to postal votes which is different from 2011 when they were only a bit more than postal votes. I am not sure either what the reason for this is. It is interesting to note though that if you look at the LC the iVote informality is 3.6% and second lowest to Postal votes informality at 1.7% and closer to half the average informality for LC. Your readers may wish to try iVote themselves and see what they think. Use the link ivote.nsw.gov.au and the button for the practice system at the bottom left of the page. As for why more preferences are marked for iVote I suspect it is just because iVoters may have a better understanding of how preferential voting works. Interesting to see if you find the same pattern in the LC data for above the line preferencing. COMMENT: Here's the practice site https://practise.ivote.nsw.gov.au/#/remote-login
1 reply
Antony a small comment on the management of postal voting applications by parties. It is my understanding that the letter sent by the party contains both the application and the HTV material and the returned envelope is sent back to the party typically unbeknown to the applicant. the party then forwards it to the Commissions. I do not believe a separate letter is sent with HTV material when an application is received by the party. My view on the Xenophon proposal is that the impact of making data entry a 100% of the ballot papers would cause enormous problems for the Commissions. In NSW at 20% we take at least 15 days of keying with 16 hour a day data entry using 120 operators. You can multiply that by 5 for the Xenophon proposal. Which would cost several million more just in NSW. I personally do not see a problem with candidates getting elected without quota. To my mind the main issue is that the voters is voting for those candidates they want to vote for not ones they know nothing about, just to make the distribution process work. The experience in NSW with Groups is that the number of candidates has not really reduced with the removal of GVT in early 2000s. This in part because the need for minimum of 15 candidates in a group but even with increased nominations costs they still keep nominating? We had more groups in 2015 in NSW than 2011. Not sure what will happen in 2019 if something is not done as we can not fit any more groups unless we wrap the groups on the paper as done in 1999. COMMENT: The flood of postal vote applications is sent out at the start of the campaign before the close of nominations. There may be a difference between what the parties do at state and Federal level. The point of how-to-votes at Federal elections is to ensure a formal vote and I understand the parties post details out. It may be different under OPV. I see three critical criteria in Senate voting reform. These are that the system should be simpler for the voter to understand what happens to their vote, that the AEC should be able to count, data enter and distribute preferences in a reasonable time, and that no vote that is currently formal should become informal through the changes. I think the Xenophon changes start to run into problems with the second criteria concerning the count.
1 reply
Antony once again great blog. One little comment about electronic voting for out of Sydney pre polls. Yes we wanted to do it at Town Hall and yes it would have greatly reduced distribution and counting issues which were horrible. We had 50 people counting on Sunday for 18 hours finishing at mid night just for prepoll. The proposal was to use basically the same iVote internet voting system as remote voters but with a verifiable paper trail. This is considered the most acceptable approach even by the most cynical security expert.
1 reply
Antony - I can see you are having a lot of fun with these data files. We are certainly keen for people to review the data and if you have the technical skills write your own program to verify our LA and LC counts. The spec and test data for LC are on the Commissions website. I thought your point at the end of the above blog regarding forcing electors to fully preference was excellent. Electors do not have a view of candidate presences beyond 2 typically so why force them to make it up? The other interesting problem full preferencing creates is a lot more work for electoral authorities. It is doubtful we would have been able to create these files had we had full preferencing in NSW, as it was the data entry of our OP system proved to be easier, more reliable and gave much more information about voter intent than manual counting under optional preferential could ever do.
1 reply
For all those that are interested the NSWEC created a TCP Tool which allows all candidate combinations to be published for each LA contest using final count data. Just go to VTR.elections.nsw.gov.au and select Legilatve Assembly and TCP for given district. Note full preferences will be provided for LA and LC at the end of May.
1 reply
Wrt question "NSWEC ..... will post Two-Candidate Preferred results for the LA in booth-by-booth format for districts". NSWEC will be putting up next week a tool which will provide TCP results for any candidate combination by polling place or declaration vote type for each district. This will use final count data not the initial count data previously posted. NSWEC plans to put up a file with preference data in about a months time.
1 reply
Antony this is an interesting analysis which needs more discussion. I note the effect you identified seems to be only for minor parties as group E actually went down for iVote? Also I note the same group had high postal take up, which shows how campaign techniques can change electoral outcome not just ballot layout. One solution which we considered was to have the screen window land randomly at a point on the ballot not just on the top left. This would not need any legislation. Interested in the comments of others. I will publish spreadsheet with all the LC initial count results once the initial count is fully entered so others can do this analysis for themselves.
1 reply
Ian Brightwell is now following The Typepad Team
Apr 2, 2015