I have seen this subject coming to the fore again with the publication by a group in Dunoon about alleged vote rigging during the referendum for Scottish independence(1). There was talk after the referendum about how the result could have been fixed. One of the things brought up was how the postal ballot could be fixed; there has always been a bad smell around the use of postal voting in the UK(2). This theory was rubbished by a number of people for a variety of reasons, some logistical, some technical. As a purely intellectual exercise, the technical objections got me to think about how I would have rigged the referendum.
So some things need to be put up front, I believe the technical objections to rigging the postal ballot are very weak. 30 minutes thought and I could do this. My background is in computing, with some years spent in a major energy supplier, I have seen over 1 million individual bills printed, folded and inserted with other advertising bumf into envelopes in a week. At no time during this process was human intervention required until more paper needs to be loaded onto the platens.
Thinking as an agent of the state, the idea is how I would ensure that my side won the vote. I set some rules for myself:
- The team had to be small
- The logistics had to be manageable
- The fix had to be repeatable
- The timescale had to fit
- It had to be very hard to detect.
Some of these rules are arbitrary and I know there should be others but I think they keep the thought processes concentrated on the problem in hand. With unlimited resources, time, and money anything is possible so we have to put a limit somewhere on what is allowed.
The numbers game
So how many votes would I need to ensure victory? The voting population of Scotland is 4,283,392, see SIR-turnout-postal-rejected-admin-WEB.xls spreadsheet, cell B36. Of that 4.28m, technically you needed 2,141,697 to guarantee a majority of 1.
Polls taken after the referendum show the “British” vote in Scotland to range from 27% in the Ashcroft poll (3) to 29% in a poll conducted by YouGov for academics from the universities of Edinburgh, Glasgow and Essex(4), this corresponded with the data from the 2011 census which indicated approximately 27% for British only(8.4%) and Scottish/British identity(18.32%)(5). A crude averaging of these figures gives us 28% as the core vote we would rely on. This equates to 1,199,350 if they all turned out to vote. This would mean for a 100% turnout in the election, you would need to secure an extra 942,347 votes to win by 1 vote. In the end, victory was achieved with 802,576 votes above the 28% total (2,001,926-1,199,350).
Towards the end of the campaign it was being predicted that turnout was going to be high, in the region of 97% (~ 4,154,890 votes), see (6) as an example. The actual number of valid votes on the day was lower than this at 84.5% (3,619,915), a difference of some 535,505 (4,154,890-3,619,915) from earlier forecasts. A mid to high eighty percent turnout would be what we would expect and plan for, this would be the highest turnout since the General Election in 1950. (7)
So 802,576 is the magic number. That is a big number to fix; it is roughly 18.7% of the total voter population. So how do you do it? To use a cliché, it is a bit like how do you eat an elephant? And the answer is a little bit at a time. So for a number this large there isn’t one way to do it, the task needs to be split up into smaller portions or it would be very visible and break rule 5. Some of the methods would be visible: the use of propaganda from state media (8); psychological warfare to keep people scared and resistant to change (better the devil you know), this accounts for almost half of all no voters; vague promises of extra powers (9). Some would not be so visible: diplomatic pressures to entice foreign governments to issue critical statements or spin these statements to appear critical (10); The Civil Service acting in a non-partisan manner. (11)
The ones we are most interested in are the covert methods that could be used. These would be low-key methods altering low volume of votes. This is where we get to vote rigging; basically there are 2 types of rigging:
- inflating your own votes or
- reducing your opponent’s votes.
A good strategy here would be a combination of the two methods; every vote you reduce is one less you have to add.
I attended the count at North Lanarkshire and although it was at times a bit shambolic, I don’t think there was any collusion in trying to alter the count in the hall (Yes won in North Lanarkshire) regardless of what the Russians monitors are reported to have said. (12) Rigging votes in the counting hall would break rules 1, 3 and 5.
One of the anomalies of the voting system is the postal vote, according to the SIR-turnout-postal-rejected-admin-WEB.xls spreadsheet there were 801,396 issued postal packs, (this is almost the magic number mentioned earlier). The number of postal votes received before the close of polls was 737,137. Although the main vote was only 85.4% of the registered levels, postal voting returns were 92% with some reports that it was at the 96% level in some areas.
The postal vote is where I would use my technical knowledge to make a difference, it meets all my rules:
- The team has to be small; this keeps the number of people in the loop to a minimum. The more people involved in something like this the less chance it will get out. I would expect that prior to 1979; this was a tiny part of someone’s brief. In 1997 this would have come more into focus and actually become a proper part of someone’s job, to look at the possibility and put together a plan to counter independence. By 2007, this had moved from being a possibility to almost a certainty, at this point a proper team would be put together to handle this. This would be a small team consisting of a couple of programmers, a network specialist, and at most a couple of other specialists.
- The logistics have to be manageable; a major plan that can’t be quickly and efficiently carried out is no use to anyone. This would include physical logistics. We are talking about some very specialised hardware and somewhere to store it. This wouldn’t be done using a couple of laptops and a multi-function printer from PC World. The hardware although being more specialised would just be bog-standard business computing hardware. The only specialist kit required would be the printing equipment. This is where the most capital outlay would be. There are more specialised companies providing bespoke hardware and as expected the costs rise dramatically. I am looking at cost not being an issue with a capital budget of at most £1.5m, dependent on whether we buy or lease.
- The fix has to be repeatable, using computer systems makes this easier, and we can prototype and dummy run this until all the bugs are ironed out. A possible dry-run could have been any by-election. The by-election in Glenrothes in 2008, could be an example of a dry-run. Even Labour activists were surprised to win that one, this prototype being used to iron out the bugs in the process.
- The timescale has to fit; this is the one I had most trouble with but once I stopped thinking short-term then the solution presented itself. This isn’t something thrown together just before the Edinburgh agreement was signed. This would have been in place for a long time; the state will have been war gaming attacks on its sovereignty since WW2. How to fix elections so that no external threat can destabilise a state would be high on any secret service’s agenda and the British state would be no different. Independence would be treated like any other threat to sovereignty. Someone, somewhere in the Security services has to think the unthinkable (13)
- It has to be hard to detect; this means that when the votes are injected into or removed from the system they don’t stand out nor do they leave a trace. The smaller the numbers to be manipulated either way, the better it will be. The receiving system can’t know it has been fed a rigged vote. My preference would be to use the PV to change 5% of the vote, roughly 200k votes. This would be hard to detect but in a close race it could be the difference between failure and success, we are gambling here by playing the percentages game.
The Voting dead
So where would we get our list of voters to be added to the postal ballot. This is amazingly simple, the dead get to vote again as they did in 1979. From 2008 till 2012 there have been just over 272K deaths in Scotland(14) and a further 54700 in 2013. (15) This is a very big pool to draw numbers from; a cross reference between voter rolls and death certificates would be incredibly easy to carry out. No special equipment or access required for this, just computing power. We could either keep entries valid or insert some to top up a particular area’s numbers. This isn’t the only scope for increasing numbers: emigrants could also cast their vote although no longer living at their old address. However, emigration levels are probably too low to make a difference but once again we are back to playing the percentages.
As I said at the beginning of this piece, this was done only as an intellectual exercise to see if it was possible and to see if I thought it was feasible. I can still punch holes in my theories but it might be a start for someone else. In the end it doesn’t matter as we can never prove it and the biggest weapons in their armoury were right out there in plain sight for everyone to see.
The last weeks panic wouldn’t have been caused by a rogue poll just showing the Yes campaign slightly ahead but a combination of them being slightly ahead and an apparently record turnout could cause a panic. All calculations so far have been based on a mid to high 80% turnout. To suddenly have newspaper and polling companies showing voting intention running at a very much higher rate would mess up the calculations we had made so far.
With the vote looking like it is getting to close for comfort this is when we get panic mode, and possibly suppressing the number of opposition votes. Depressing votes could be done in one of 2 ways.
- Physical removal of ballot boxes. This only works if you know all the votes in a box favour the other side or you could end up removing more of your own votes.
- Replacing ballot boxes with boxes you have stuffed with your own votes.
Neither of these options is good, they break almost every one of the rules we set but if there isn’t much time left and the outcome is looking tight, we might have no other option. One of the most surprising vote totals for the day was from Glasgow. A city that was reportedly having record numbers of registrations with long queues of people waiting to register to vote. In fact this seems to be the place where the 97% turnout story starts. Yet, on the day, Glasgow had one of the smallest percentage turnouts. In a short period of time we have people anxious to register to vote and then after going through that process they didn’t bother to vote.
When all is said and done one of the main problems is, why were there no proper exit polls on the day? According to some expert witnesses in an American court case about poll rigging, one of the most important things if you don’t want to get caught is, you don’t allow exit polls(16). Did they learn from Glenrothes about the level of distrust that exit polls raise with anyone examining the results?(17)
Should we stop PV?
The longer you think about the Postal Voting process the more the holes in the system become apparent(18). The polling process is to a large extent based on trust and when this trust is broken the whole thing begins to look suspicious. In North Lanarkshire we were denied the opportunity to follow the PV from the council headquarters to the count apparently on the recommendation of a senior police officer. So who is to say that the votes that left the council offices were the same ones that arrived at the counting hall? If we are to continue with PV, then some serious work needs to be done to ensure the process is transparent and fully auditable.
Failure to address the weaknesses in the PV system will just bring the system into more disrepute and look more like someone is trying to protect vested interests.