Butterfly Effects: Ballot design | Hacker News Reader

Butterfly Effects: Ballot design

99percentinvisible.org

89 pointsmisnamed68 comments

Open articleView on HN

Compare and contrast:

This City of Austin ballot question:

https://cdn1.vox-cdn.com/uploads/chorus_asset/file/6679117/S...

The Brexit ballot:

http://i2.cdn.turner.com/cnnnext/dam/assets/160621144335-uk-...

Can anyone (who hasn't seen it before) easily explain what a vote in the Austin example would do?

From http://www.vox.com/2016/6/23/11979522/brexit-ballot:

    A study of more than 1,200 state-level ballot questions that voters across the
    US were asked to approve between 1997 and 2007 found that, on average, they were
    written at a 17th-grade level — in other words, you’d need more than a
    four-year college degree in order to understand what you were being asked to
    vote on.

    In response, many voters just skipped those questions altogether: A
    more complex ballot question led to a 5 percent decrease in participation. On
    about one-quarter of all ballot questions studied, that was larger than the
    margin between victory and defeat. Confusing language actually could have
    changed the outcome of the race.

In PA we have a ballot question that reads,

"Shall the Pennsylvania Constitution be amended to require that justices of the Supreme Court, judges, and magisterial district judges be retired on the last day of the calendar year in which they attain the age of 75 years?"

If you think that forcing judges to retire at a certain age is a good idea, you will be inclined to vote "yes". The problem is that PA already has mandatory retirement for judges, only at age 70. So the way that the question is phrased is very misleading, and could easily lead to PA raising the retirement age for judges from 70 to 75 despite voter intent to the contrary.

As I understand it, this is the intent of the wording after a couple of previous, clearly worded ballot questions properly went down in flames at the hand of the voters.

Also read there's a particular Supreme Court judge for whom this latest effort was made.

I can't speak to the particular judge being the driving force behind it, but you are spot-on about the wording. In the primary, the question was worded this way:

"Shall the Pennsylvania Constitution be amended to require that justices of the Supreme Court, judges and justices of the peace (known as magisterial district judges) be retired on the last day of the calendar year in which they attain the age of 75 years, instead of the current requirement that they be retired on the last day of the calendar year in which they attain the age of 70?"

That's not misleading, it's a straight up lie. It doesn't say that it would change previous laws. So if there is an existing law that says judges must retire at 70, nothing there says it will be repealed.

But it says "amended".

I have to take issue with this part:

> A study of more than 1,200 state-level ballot questions that voters across the US were asked to approve between 1997 and 2007 found that, on average, they were written at a 17th-grade level — in other words, you’d need more than a four-year college degree in order to understand what you were being asked to vote on.

Flesch-Kincaid is a terrible measure, at least if taken at face value about the "grade level" of a text. See for instance http://languagelog.ldc.upenn.edu/nll/?p=15456 or http://languagelog.ldc.upenn.edu/nll/?p=21847 .

To me, the worst part of the Austin one is not the rather complex ballot question, but the way the answer choices are written. By choosing "For the ordinance", you're actually voting against the ballot measure (which repeals "the ordinance"). Some people might be excited when they hear that there's a ballot measure to bring back Uber and vote "For the ordinance" without understanding that that means keep things as they are now. The options should be something more like -- "YES -- repeal the ordinance" and "NO -- keep the ordinance as is".

Is that actually what it means? The question asks about repealing an ordinance and replacing it with a new one. I thought "for the ordinance" meant for the new one which would also repeal the old ordinance.

Either of us might be right... the fact that we can't be sure without further research is exactly the problem I'm pointing out! My interpretation is based on the only use of the word "Ordinance" in the ballot question referring to the existing Ordinance on the books.

There is a second occurrence, "replace with an ordinance that would".

Indeed, I didn't catch that. And based on some reading [1] it seems my interpretation was indeed exactly wrong. Thereby either proving my original point -- that the answers are quite confusing -- or maybe just that I'm an idiot/outlier. I'd argue it's especially confusing with the capitalized 'O' in the ballot answer and the corresponding capital O when the ballot question refers to the existing 'Ordinance', but not the replacement 'ordinance'. Crazy...

To amend my original post, the answers should have been phrased more like "YES -- repeal and replace the existing ordinance" and "NO -- keep the existing ordinance".

[1] http://kxan.com/2016/04/25/early-voting-begins-monday/

A great example of why I hate "Ok" and "Cancel" for most dialog boxes even if the dialog explains what each of them will do.

Crucially, Britain has a well-funded independent body tasked with overseeing every aspect of an election. The EU referendum question was decided after lengthy research and testing; this testing included subjects with limited English literacy and learning difficulties. Similar efforts went into the design of the Scottish independence referendum ballot.

We do not use any form of voting machine, preferring the simplicity, reliability and integrity of pencil and paper. If multiple elections occur simultaneously, each election will be conducted on a separate ballot paper. This naturally results in a slower and more expensive counting process, but we avoid the threat of hanging chads or broken touchscreens. That greater procedural cost is more than offset by strict campaign finance restrictions.

http://www.electoralcommission.org.uk/__data/assets/pdf_file...

I've never seen it before. Read it for ~30 seconds.

Vote yes to repeal the existing ordinance that requires background checks and to install an ordinance that smooths the way for corporate ride finding companies.

I can see an argument for splitting the various issues out more though.

At least one other commenter so far thought "for the ordinance" meant for the capital O ordinance, the one they're considering repealing. I agree with your interpretation, but I'm not totally convinced we're right.

Never seen this before. A vote "for the ordinance" would "allow" Uber, Lyft, et al back into Austin, no?

That sums up the Brexit, ask them and none of them new what the hell they were really voting for....for or against.

The Brexit ballot paper illustrates the exact opposite of that: unlike the confusing Austin example, the Brexit ballot paper makes it absolutely clear to the voter what they're voting for.

I think what the parent poster refer to isn't so much whether the voter understood the wording of the ballot and how to cast their vote, but whether they understood the implication of it. Indeed, the fact that no one is able to define what leave means even now should be evidence enough that no one could possibly have understood it at the time of the referendum.

Putting politics aside and focusing on design alone I agree with you though, the ballot is very much on point.

I'm surprised how on discussions about the 'butterfly ballot', the impact of the viewing angle is rarely mentioned. See these images [1][2][3] that show that as the viewing angle deviates further from top-down, the recessed spine that holds the circular marks to be punched gets misaligned from the arrows that are supposed to be pointing to the proper spot.

As the ballot was installed on a stand at approximately chest-height [4] (it's really hard to find a picture of a person standing at it -- does anyone have one?), shorter people had a more oblique view making the arrows more likely to get misaligned.

[1] http://www.washington.edu/news/files/2016/03/3479146036_dda3... [2] https://c2.staticflickr.com/6/5291/5481639542_1e342a5f69_b.j... [3] https://upload.wikimedia.org/wikipedia/commons/0/06/Butterfl... [4] http://i.ebayimg.com/images/g/~jcAAOSw~bFWOQO0/s-l1600.jpg

ATMs have the same problem with viewing angles. It's funny with those too because that interface will always be used at extreme angles: From above (potentially) when a person is standing at an ATM, from straight on if they happen to match the ATM's screen height, and from below if someone is sitting in their car.

A good case for touch screens, where the label is the interaction point.

Never seen an ATM to be used from a car, but as for touch screens, many of those have severe alignment problems that would get even worse due to parallax when used from wildly different viewing angles. Requires you to have even larger buttons to counteract that or frustration sets in where people constantly get the impression their touches don't register (just because they manage to hit slightly below the button).

ATMs I tend to use have the following design: https://www.sparkasse-einbeck.de/module/aktion_if/wunsch-pin... with a row of four large physical buttons to the left and right of the screen. It's impossible to get a viewing angle extreme enough that this gets ambiguous.

Around here (Washington, DC area) drive-up ATMs are fairly common at banks which have drive-through service.

Having used both types of ATMs, I find the touch-screen ones less error prone. The touch targets are always gigantic. They pretty much have to be anyway to ensure that everybody can read the labels. I just touch what I want and it works. For ATMs with physical buttons on the side, I always have to crouch down to get the right angle so I can figure out which button does what. With the ones I've used, it's quite easy to get a viewing angle where it's ambiguous.

In my area, there are several banks that had drive-up ATMs at some point, but most are unused nowadays.

One feature they had in common: each function button had Braille (raised dots for the blind) instructions also.

Yes, Braille. On the driver's side.

This could explain much of what I encounter on the roads.

Some filling stations also use the arrangement of four buttons on either side of a large dot matrix screen, for a total of eight buttons. But even this has alignment problems -- ones I've experienced personally despite images being difficult to find. But these [1][2] show the problem: the options listed on the in-set screen come out of alignment with the buttons at extreme angles. That, coupled with other design decisions [3] makes some filling stations frustrating to use.

[1] http://wlbpa.org/wp-content/uploads/2014/02/gas-pump.jpg [2] http://image.mlive.com/home/mlive-media/width620/img/kzgazet... [3] http://dbhurley.com/media/uploads/2016/03/bad_ui_gas_station...

Here's examples https://www.ncr.com/sites/default/files/WrightPatt_Interacti...

http://image.shutterstock.com/z/stock-photo-woman-making-a-d...

That is abysmal design, so many things wrong with it!

This article has a clearer image of the Florida 2000 ballot: http://www.asktog.com/columns/042ButterflyBallot.html

Also, good podcast interview about the post-vote legal wrangling https://soundcloud.com/candidate-confessional/the-man-who-tr...

And another podcast specifically on the issue of ballot design: https://www.relay.fm/presentable/11

That ballot could definitely be designed better even without changing the machine design (e.g. funnels pointing toward the valid hole), but if the arrows were clearly visible on the actual ballots, doesn't that say something about the level of awareness of some of the voters?

Look at this: https://c2.staticflickr.com/6/5291/5481639542_1e342a5f69_b.j...

Now, we might get that correct 99% of the time. But, in close elections 1% is meaningful.

That does make it look much worse.

Of course. But so what? Do we want people's votes to be counted differently from what they wanted just because they were less than entirely aware of little details of the ballot design?

Naturally in a democratic republic we want every vote to be counted correctly. I was trying to say that we also need to work on our collective awareness, especially around critical things like voting.

That's fair. Any ideas as to how? I can think of ways to make ballots more understandable, but I have no idea how to get people to pay more attention to them.

I had been aware of the hanging chads in the Bush-Gore election, but I had not seen the butterfly ballot design itself. In such a close race, many voters likely voted for Buchanan when intending to vote for Gore... crazy.

What really blows my mind is that Buchanan himself conceded the point - he said very clearly he thinks the ballot design accounted for him getting a bigger percentage in districts that used it.

Although only tangentially related to this article, the big thing I see missing from discussions of voting systems in the US (and UK) is the fact that you are only allowed to express a single top preference for one candidate.

What is the Buchanan supporter to do in 2000? Vote for their top preference in the knowledge that they have no chance of getting elected (given that he is well behind in third or worse place in the polls), or vote for their least worst candidate from the two front runners.

If instead, the voter was allowed rank their candidates in order of preference, with their vote going to their #2 preference after Buchanan is eliminated, their voting intent could be clearly recorded.

Imagine if there was no need for republican or democrat primaries, as each party could field multiple candidates, as e.g. all republican votes would still end up going to the top republican candidate after less popular (to the public) republican candidates are eliminated.

The system I'm most familiar with which implements this is Single Transferrable Vote [1] and isn't a pipe dream but is used effectively in multiple countries (albeit not in a presidential vote that I know of).

If such a system were in place in the US I believe it would reduce the extreme polarisation that is obvious to me in U.S politics and also I think could enable a third major party to grow slowly over the course of multiple election cycles, or indeed make it conceivable that e.g. (current election) a major pro-immigration faction within the Republican Party could split to form a new party without the prospect of electoral annihilation, which I believe is why that is inconceivable.

There are degrees of democracy and for me the STV system is just vastly more democratic, more akin to how people initially express their ideal position at the start of a debate, but then still get a say as the debate progresses and they realise they need to compromise and support the next best outcome of their first preference is untenable to a majority of others.

[edit] To get back to the point of the article, I think that entering '1', '2', '3' beside candidates names is a lot clearer than the traditional tick or 'X' (which could be reasonably interpreted to mean NOT this candidate!). Filling up every space with a number is a great way to ensure someone can't easily modify/spoil your vote by e.g. adding an extra 'x' in one of the remaining empty spaces.

[1] https://en.m.wikipedia.org/wiki/Single_transferable_vote

Unfortunately STV has the same problems as first-pass-the-post, and some actually worse ones. Which is unfortunate, because people don't understand that and think it's much better.

http://www.rangevoting.org/Irvtalk.html

But the site you mention does highly recommend approval voting, which is much simpler and doesn't have those problems.

Thanks, I wasn't aware of range voting and it indeed looks superior from a reading of that page.

It has some of the same kinds of problems that approval has, given that approval voting is a special case of range voting with only 2 options, 0 or 1. But I think it's better than all of the others, and much simpler to implement and understand I think.

The root cause of a lot of this is how complex ballots are in American elections. For comparison, this is what a ballot looks like in a Canadian federal election:

http://www.capyouthvote.org/wp-content/uploads/2015/07/ballo...

The reason ours are simpler is because the elections are simpler: we just elect our local member of parliament. At the federal and provincial levels, that's the only decision to make in an election. We don't vote for senators or judges or any other offices at these levels. Instead, these are appointed by parliament.

The municipal elections have the most options because there we elect mayors, city councillors and school board trustees. But they're still just paper "put an X next to the name" ballots that are manually counted.

The federal portion is pretty small in an American election too, three races at most. President every four years, one congressman every two years, one senator at a time (six year terms, but they are staggered so a state's senators get elected in different cycles). At the federal level, we don't vote for judges or referenda or anything like that.

On the other hand, my state is vote by mail, so no tricky machines. Just drawing a line to complete an arrow, then mailing it back in. Pretty hard to screw up.

Even if you have 10 candidates, how hard is it to just list them on a sheet of paper with a nice big box to draw an X in next to each. The computer can then just scan the boxes and see which one has a big X in it.

Seems to work fine here in the Netherlands, and we have a bit more than ten candidates.

http://i.imgur.com/pdW1ZTe.jpg

The most effectively designed ballot for its purpose was probably the one used to decide the Anschluss referendum

https://en.m.wikipedia.org/wiki/Austrian_Anschluss_referendu...

Why aren't there more modern systems that use a weight of some sort? Easy to measure in multiple ways (weigh the container, clicker that counts entered balls), anonymous, hard to misunderstand.

That way we can have the result immediately.

For postal votes it appears you're not anonymous anyway, so no harm in the returning officer dropping a ball in for you.

The Florida 2000 thing is a recipe for disaster. Who's ever heard of a weird fold-in-the-middle ballot that you have to punch a hole in?

> For postal votes it appears you're not anonymous anyway

The vote-by-mail ballots we have in Oregon come with two envelopes: An envelope printed with your name and address; and a slightly smaller envelope with no identifying markings. You put your completed ballot in the "secret" envelope and you put that envelope in the larger envelope.

Assuming everything goes correctly, the name and signature on the outer envelope is verified against the voter role. If it's a valid vote, the envelope is opened and the ballot envelope is placed in a secure place. Later, on election day, the ballot envelopes are opened and the ballots are read electronically.

There's a risk that the vote-counter may open both envelopes but it doesn't seem like there's a bigger risk with that than there is with other voting methods.

There is a bigger potential risk of intimidation- e.g. a company saying to its employees "bring us your ballot so we can verify that you voted correctly". I don't know how you solve that.

> There is a bigger potential risk of intimidation- e.g. a company saying to its employees "bring us your ballot so we can verify that you voted correctly". I don't know how you solve that.

Ask anyone who receives such intimidation to report it to their state/federal election commission for prosecution. Voter intimidation is a federal felony, and typically against state law as well.

absolutely, I've never actually heard of it happening, I'm just saying it's a possibility. I would hope the fact that it's really illegal would prevent it.

Intimidation could be avoided by allowing a voter to send in a ballot multiple times, with only the last one counted, or allow them to void their mail-in vote by visiting in person on election day.

I wouldn't worry about employers but I could see this being a major issue for spouses.

To cast a free and fair election, we should run a good election. Perhaps a standard ballot design could be required in all states?

The problem is that voting machines are a competitive system sold to various levels of government. At the state level, there are different requirements, but not unified at the federal level [0]. Due to the differences, there cannot be a single mandated design, as the voting machine companies are not required to be consistent from state to state. There is lots of history here, which I haven't gotten very far into, but I think most would agree that reform and consistency would solve these problems. That wouldn't benefit the voting machine company profits though, so there is huge push-back.

[0] https://www.eac.gov/assets/1/Page/State%20Requirements%20and...

How do you deal with states that have propositions vs. states that don't? What about states without bicameral legislatures [0]? Or states where a single candidate can run for multiple parties [1]? The reason you can't have a standard ballot design is because states don't have standardized races or systems. And so long as the US is a federation, you really wouldn't want to have all states with the same electoral system.

[0] https://en.wikipedia.org/wiki/Nebraska_Legislature

[1] https://en.wikipedia.org/wiki/Electoral_fusion

Just because there's one standard design doesn't mean that each state won't have slightly different requirements; it just means that the design of the standard ballot would have to incorporate those requirements.

Multiple ballot papers.

In Switzerland, people write themselves Yes or No in the appropriate field, or they write the names of the people they vote for (misspellings will still be counted, unless two candidates have similar names and it's hard to make out which one was meant). How hard can ballot design be? Not hard, if you're not trying too hard.

"misspellings count unless it's hard to tell"

This sounds like you're just asking for controversy. You'll always be able to find someone with bad handwriting whose vote was missed, or someone who wrote in "Democrat" instead of the name, or "Si" instead of "Yes", or something.

Also, this makes automated counting out of the question. It seems almost obvious to me that having people fill in a circle is much better than having them write out whole words. Florida's design was bad, but this would be worse.

No, there was an article about it in the newspapers. It's actually really simple. If there is any doubt, it doesn't count. Otherwise, they'll count it as what you meant. It works beautifully.

Ok, I just don't see how you can argue that's better than just filling in a circle next to your choice. Maybe you can say it's better than Florida's stupid hole punch thing, but I don't see a single advantage over filling in a single circle, and I see many disadvantages.

It's way too easy to imagine someone losing an election by a tiny margin and finding hundreds of ballots that were thrown out but should arguably have counted for them. What is the benefit of this extra room for error? Who gets to decide when there is doubt?

I am sure it works great, but I am also sure that the system most US states have at the moment is better.

Statistics have actually shown that only a very, very small part of votes was both misspelled and ambiguous in such a way that it became invalid.

We're talking about 'murica here...

America also doesn't have a national language, and ballots are going out in Spanish, Vietnamese, Chinese, etc.

"Your vote will only be counted if you write 'YES' or 'NO' in the space provided."

"Su voto sólo se contará si escribe 'YES' o 'NO' en el espacio proporcionado."

"phiếu bầu của bạn sẽ chỉ được tính nếu bạn viết 'YES' hoặc 'NO' trong không gian được cung cấp."

"如果你写您的投票将只能算作“YES”或“NO”的空间提供。"

Google Translate used, so these translations are probably wrong. But the concept is the same; you only accept a particular response and provide instructions in the native language on how to respond correctly. All other responses are discarded.

I suppose. It sounds like there'd be a lot of havoc over having people write "yes" or "no" vs "si" or "no," that's kind of what I was thinking would cause the problems.

With bad ballot design/wording, voting on a weekday, reportedly long lines (read something about a 4 hour wait in places yesterday!), the lack of valid ID among potential voters, first past the post and out-of-control gerrymandering, why does the US have such trouble carrying out elections?

Reply on news.ycombinator.com