Cambridge city council vote recount
Jan. 7th, 2014 09:53 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
magidNow that the new councilors have been sworn in, I should finishing writing and posting about this!
There were 25 candidates on the ballot for nine seats, with three of four candidates that were very close getting a seat. Unsurprisingly, the fourth candidate, an incumbent, called for a recount. It takes longer to do this in Cambridge than other places, because there's proportional voting (more explanation below).
I got involved with the recount when a Facebook friend posted about needing people for the recount. I said I could help, and became an observer for a candidate, going to the youth center where the recount was happening, unsure what to expect.
I was given a bright pink observer sticker, and told to watch as the next precinct's votes were distributed, making sure they made it into the right box. Given that there are 11 wards (each with three precincts (except that one that has an extra: ward 3, precinct 2a; no clue why it just wasn't called precinct 4 when they split it off)), and a finite amount of table space, it meant that some of the 27 boxes in each precinct (one for each candidate, plus write-in and invalid) were stacked, making it just a bit easier for mistakes to happen. It was a bit chaotic: each precinct had at least one person reading the ballot for the number one vote, announcing it aloud, then giving it to a 'runner' who confirmed the number one vote, announcing it aloud again, then holding it up so observers could confirm it had been read correctly, and finally placing it in the appropriate box. All the sound, the tumult of announced names, made me think of what I've heard of party conventions back in the 70s when campaigning was still a real part of it. I'd hear "Carlone!" echoed in one part of the room, "Mazen!" from another, and "van Buezekom!" from yet a third. (Not to mention all the amusing pronunciations of names that people found challenging. Mazen, for instance, was at time Mayzen, Mahzen, Marzen, Mahzeen, and more.)
Once all the votes were separated out for a precinct, the ballots were tallied for each candidate in each precinct, and put in an official envelope with that information on the outside. Then auditors would confirm that the numbers added up and matched previous counts (or knew why they didn't) before another precinct's ballots were pulled from the locked ballot boxes stacked in the middle of the room. (I was amused to see that the non-ballot stuff the city brought was all in fixed-lid boxes, which I associate with travel to Burning Man.)
There is more than one way to have proportional voting. Cambridge uses the Single Transferable Vote system, which basically means that you can vote for as many people as you like, but you have to rank them (which is why all the signs for Cambridge candidates have stuff about getting someone's #1 vote). If the person you vote for as #1 either gets more votes than they need, or gets not many votes at all, the ballot is redistributed to your #2 person, or #3, etc. If you vote for more than one person as #1, but one person for #2, the #2 person gets it, because there's no way to figure out which of the #1s the voter preferred. In some ways this is fairer than a straight up vote for one person, but there are issues with it as well, of course. However, this is the basic idea.
Once the ballots were separated out by candidate, the circle of tables lining the room was changed. Before, it had been by ward, with poll workers on the inside of the tables. Now, it was by candidate, with poll workers sitting outside the tables. Observers could stand behind the poll workers to watch during quality control. This was a recheck that all the ballots had gone to the right place. Poll workers emptied an envelope (observers could ask to see that the envelopes were empty), counted the ballots (there were at least two people doing this, double-checking each other), then looked at and held up for observers to check that the ballot was in the correct place. After that, they were stamped on the back with the candidate's name and a number, using incrementing stamps that might have been made in the 60s given the style of the packaging (and since there were inevitably issues with those, the election commissioners would have to come over to fix and initial any where the numbers were wrong).
As the votes for each candidate were totaled, the stacks of ballots got taller, each precinct separated by a big colored paper.
If someone challenged a ballot, or found an issue, they'd raise a hand, and the election officials would come over. If it was just an obvious error (a ballot ended up in the wrong place), it would be sent on its way to the right place, in another white envelope describing what it was, where it was coming from, where it was going to (I found misfiled ballots a number of times, and it satisfied the editorial part of my brain: if I were awake enough to catch these, then presumably my eyes were open enough not to miss other issues.). If it were a challenge (someone's marks weren't clear, perhaps), there were thee election commissioners who'd vote on what the voter's intent was. (Often there were four commissioners in the room; three were needed at minimum for the recount to continue.)
So, the thing I'm not a huge fan of in this system is that order matters for how ballots spill over, or cascade, or whatever, to another candidate once one candidate has reached the quota for winning (I think in this case it was one-sixth of the ballots cast, plus one, but I'm not 100% on that). Some precincts are very much one candidate's territory, so just ordering by ward and precinct (ward 1 precinct 1 to ward 11 precinct 3) means that the cascading votes aren't necessarily representative. Instead, before the election, there's a random ordering of precincts, and then every nth ballot is taken out to cascade to the next candidate (with n being determined by the number of spare ballots, though if a ballot had no further choices, I believe a different one was taken instead). When doing a recount, this means that the ballots have to be put back into order. Since one candidate won in the first round with extra ballots to cascade, his ballots had to be put back into the order they had been in for the original vote. There were facsimiles of each ballot that had numbers on them, for order, but the originals didn't have that. However, there were 25 versions of the ballot, each with a different candidate listed first (and otherwise alphabetical), since it's known that being towards the top is an advantage, so it turned out to be less of a problem to match the facsimiles with their originals than I'd expected.
Once those had been ordered, then redistributed, no one else had a quota, so the ballots for the candidate who'd gotten the least votes were redistributed to the #2s.
Side note: it's much more difficult to check a ballot quickly past the #1 votes, since it could be #2 or far higher, depending on who was voted for who's already been either knocked out or voted in. At one point I saw a ballot that went to the fifteenth choice.
As the votes were redistributed, there was another single station set up with one poll worker reading out who the vote was going to, and others putting them in a slowly-decreasing row of boxes. (I was impressed by that poll worker; she read thousands of ballots aloud.) Once one candidate's ballots were put into the boxes, they went to the stations around the room for quality control check as before. Still, no one won.
And in fact, the next winner came in round 15, partway through the redistribution of someone's ballots. The process stopped, quality control was done, and the quota being met was confirmed. Then the rest of the ballots were redistributed, with one fewer box.
The winners were all determined by round 17, and were the same as the original count. It had taken nine days, with over 60 poll workers in addition to dozens of observers.
Some other bits and pieces about this:
* It was interesting to see some of the ballots. Some people voted on the diagonal, or the diagonal minus any incumbents, while others voted only for the ones they knew about and wanted.
* Some protests are more useful than others: skipping numbers so you vote for #5, #12, and #20, for instance, only means that they move up to #1, #2, and #3. Writing in "none" seems pretty useless. On the other hand, real write-ins work. One I was amused by: someone who voted by ranking all the candidates.... plus a write-in at #9. I couldn't figure out if the voter thought that would be useful.
* I'd decided that I wasn't really a partisan observer, more someone interested in accuracy. This meant that when someone who was a partisan observer told me that they'd let two ballots be misfiled, hoping they wouldn't be put right in the quality control phase, I lost a lot of respect for them.
* It was interesting how social a process it was, partly because there were a lot of times when nothing much was happening (the commissioners might be meeting or something). I met some lovely people, both poll workers and observers. I think I ended up in a lot more conversations than I would have otherwise because I was crocheting pretty much the whole time I wasn't looking at ballots, and people were intrigued.
Poll workers were provided with snacks and lunch, but the observers went out to lunch or brought their own. This lead to observers bringing in baked goods to share with the crowd of observers. I brought some spicy chocolate bread one day, which was liked.
I ended up chatting a bunch with the Nadeem Mazen folks. I am impressed with him; he's an engaging person, runs Danger!Awesome, and is also a religious enough Muslim to not eat meat out (ie, cares about halal). I was glad he was one of the winners of the election.
* In talking to my math professor cousin about my frustration about how order matters in the cascade, he suggested something that I don't think would actually work in practice (certainly not making a recount happen in a short time), but would be more satisfying to me, anyway: run every possible selection for the excess votes cascading onward, and see how the percentages fall. I'd really like some way of having preferential voting, but not having to touch any ballot more than once. Not really sure how that could work.
There were 25 candidates on the ballot for nine seats, with three of four candidates that were very close getting a seat. Unsurprisingly, the fourth candidate, an incumbent, called for a recount. It takes longer to do this in Cambridge than other places, because there's proportional voting (more explanation below).
I got involved with the recount when a Facebook friend posted about needing people for the recount. I said I could help, and became an observer for a candidate, going to the youth center where the recount was happening, unsure what to expect.
I was given a bright pink observer sticker, and told to watch as the next precinct's votes were distributed, making sure they made it into the right box. Given that there are 11 wards (each with three precincts (except that one that has an extra: ward 3, precinct 2a; no clue why it just wasn't called precinct 4 when they split it off)), and a finite amount of table space, it meant that some of the 27 boxes in each precinct (one for each candidate, plus write-in and invalid) were stacked, making it just a bit easier for mistakes to happen. It was a bit chaotic: each precinct had at least one person reading the ballot for the number one vote, announcing it aloud, then giving it to a 'runner' who confirmed the number one vote, announcing it aloud again, then holding it up so observers could confirm it had been read correctly, and finally placing it in the appropriate box. All the sound, the tumult of announced names, made me think of what I've heard of party conventions back in the 70s when campaigning was still a real part of it. I'd hear "Carlone!" echoed in one part of the room, "Mazen!" from another, and "van Buezekom!" from yet a third. (Not to mention all the amusing pronunciations of names that people found challenging. Mazen, for instance, was at time Mayzen, Mahzen, Marzen, Mahzeen, and more.)
Once all the votes were separated out for a precinct, the ballots were tallied for each candidate in each precinct, and put in an official envelope with that information on the outside. Then auditors would confirm that the numbers added up and matched previous counts (or knew why they didn't) before another precinct's ballots were pulled from the locked ballot boxes stacked in the middle of the room. (I was amused to see that the non-ballot stuff the city brought was all in fixed-lid boxes, which I associate with travel to Burning Man.)
There is more than one way to have proportional voting. Cambridge uses the Single Transferable Vote system, which basically means that you can vote for as many people as you like, but you have to rank them (which is why all the signs for Cambridge candidates have stuff about getting someone's #1 vote). If the person you vote for as #1 either gets more votes than they need, or gets not many votes at all, the ballot is redistributed to your #2 person, or #3, etc. If you vote for more than one person as #1, but one person for #2, the #2 person gets it, because there's no way to figure out which of the #1s the voter preferred. In some ways this is fairer than a straight up vote for one person, but there are issues with it as well, of course. However, this is the basic idea.
Once the ballots were separated out by candidate, the circle of tables lining the room was changed. Before, it had been by ward, with poll workers on the inside of the tables. Now, it was by candidate, with poll workers sitting outside the tables. Observers could stand behind the poll workers to watch during quality control. This was a recheck that all the ballots had gone to the right place. Poll workers emptied an envelope (observers could ask to see that the envelopes were empty), counted the ballots (there were at least two people doing this, double-checking each other), then looked at and held up for observers to check that the ballot was in the correct place. After that, they were stamped on the back with the candidate's name and a number, using incrementing stamps that might have been made in the 60s given the style of the packaging (and since there were inevitably issues with those, the election commissioners would have to come over to fix and initial any where the numbers were wrong).
As the votes for each candidate were totaled, the stacks of ballots got taller, each precinct separated by a big colored paper.
If someone challenged a ballot, or found an issue, they'd raise a hand, and the election officials would come over. If it was just an obvious error (a ballot ended up in the wrong place), it would be sent on its way to the right place, in another white envelope describing what it was, where it was coming from, where it was going to (I found misfiled ballots a number of times, and it satisfied the editorial part of my brain: if I were awake enough to catch these, then presumably my eyes were open enough not to miss other issues.). If it were a challenge (someone's marks weren't clear, perhaps), there were thee election commissioners who'd vote on what the voter's intent was. (Often there were four commissioners in the room; three were needed at minimum for the recount to continue.)
So, the thing I'm not a huge fan of in this system is that order matters for how ballots spill over, or cascade, or whatever, to another candidate once one candidate has reached the quota for winning (I think in this case it was one-sixth of the ballots cast, plus one, but I'm not 100% on that). Some precincts are very much one candidate's territory, so just ordering by ward and precinct (ward 1 precinct 1 to ward 11 precinct 3) means that the cascading votes aren't necessarily representative. Instead, before the election, there's a random ordering of precincts, and then every nth ballot is taken out to cascade to the next candidate (with n being determined by the number of spare ballots, though if a ballot had no further choices, I believe a different one was taken instead). When doing a recount, this means that the ballots have to be put back into order. Since one candidate won in the first round with extra ballots to cascade, his ballots had to be put back into the order they had been in for the original vote. There were facsimiles of each ballot that had numbers on them, for order, but the originals didn't have that. However, there were 25 versions of the ballot, each with a different candidate listed first (and otherwise alphabetical), since it's known that being towards the top is an advantage, so it turned out to be less of a problem to match the facsimiles with their originals than I'd expected.
Once those had been ordered, then redistributed, no one else had a quota, so the ballots for the candidate who'd gotten the least votes were redistributed to the #2s.
Side note: it's much more difficult to check a ballot quickly past the #1 votes, since it could be #2 or far higher, depending on who was voted for who's already been either knocked out or voted in. At one point I saw a ballot that went to the fifteenth choice.
As the votes were redistributed, there was another single station set up with one poll worker reading out who the vote was going to, and others putting them in a slowly-decreasing row of boxes. (I was impressed by that poll worker; she read thousands of ballots aloud.) Once one candidate's ballots were put into the boxes, they went to the stations around the room for quality control check as before. Still, no one won.
And in fact, the next winner came in round 15, partway through the redistribution of someone's ballots. The process stopped, quality control was done, and the quota being met was confirmed. Then the rest of the ballots were redistributed, with one fewer box.
The winners were all determined by round 17, and were the same as the original count. It had taken nine days, with over 60 poll workers in addition to dozens of observers.
Some other bits and pieces about this:
* It was interesting to see some of the ballots. Some people voted on the diagonal, or the diagonal minus any incumbents, while others voted only for the ones they knew about and wanted.
* Some protests are more useful than others: skipping numbers so you vote for #5, #12, and #20, for instance, only means that they move up to #1, #2, and #3. Writing in "none" seems pretty useless. On the other hand, real write-ins work. One I was amused by: someone who voted by ranking all the candidates.... plus a write-in at #9. I couldn't figure out if the voter thought that would be useful.
* I'd decided that I wasn't really a partisan observer, more someone interested in accuracy. This meant that when someone who was a partisan observer told me that they'd let two ballots be misfiled, hoping they wouldn't be put right in the quality control phase, I lost a lot of respect for them.
* It was interesting how social a process it was, partly because there were a lot of times when nothing much was happening (the commissioners might be meeting or something). I met some lovely people, both poll workers and observers. I think I ended up in a lot more conversations than I would have otherwise because I was crocheting pretty much the whole time I wasn't looking at ballots, and people were intrigued.
Poll workers were provided with snacks and lunch, but the observers went out to lunch or brought their own. This lead to observers bringing in baked goods to share with the crowd of observers. I brought some spicy chocolate bread one day, which was liked.
I ended up chatting a bunch with the Nadeem Mazen folks. I am impressed with him; he's an engaging person, runs Danger!Awesome, and is also a religious enough Muslim to not eat meat out (ie, cares about halal). I was glad he was one of the winners of the election.
* In talking to my math professor cousin about my frustration about how order matters in the cascade, he suggested something that I don't think would actually work in practice (certainly not making a recount happen in a short time), but would be more satisfying to me, anyway: run every possible selection for the excess votes cascading onward, and see how the percentages fall. I'd really like some way of having preferential voting, but not having to touch any ballot more than once. Not really sure how that could work.
no subject
Date: 2014-01-09 01:23 pm (UTC)