Ballot in Florida : LUSENET : Joel on Software : One Thread

Do you really think it is a usability bug or just a careless voter? The arrows clearly indicate what hole to punch. If this arguement were true than every candidate on the ballot received votes intended for someone else. The format of the ballot was approved by all parties involved prior to election day. It seems like people are crying foul because they didn't receive the outcome they expected.

-- Anonymous, November 09, 2000


The arrows are just one piece of information on that ballot. If you want to vote for Gore/Lieberman, the SECOND name on the ballot, you must punch the THIRD hole, even though according to the number next to the arrow Gore/Lieberman is number FIVE.

So yes, the arrows clearly indicate that you punch the hole that registers a vote for Gore. But the ordering clearly indicates you punch the Buchanan hole and the numbering clearly indicates you punch the hole for Bush or Browne. (Hell, the only thing that isn't "clearly indicated" by that ballot is how to punch a hole and vote for Ralph Nader.)

-- Anonymous, November 09, 2000

I think it was careless voters.

Are we sinking to the point where we need to have the pictures of candidates on the ballot so that illiterate voters won't be confused?

In my opinion, if you can't figure out that ballot, you shouldn't be voting. In fact, if you can't name your governor, congressman and senators, you aren't qualified to vote.

-- Anonymous, November 09, 2000

I clearly see the unintended benefit Bush recieved from the ballot. Many voters would match the name on the left column with the circles presented for feedback. Bush is "Name #1" and also matches the expected "Circle #1" for input. Gore is "Name #2" and many would select "Circle #2" (a vote for Buchannan in this case)... I believe these circles are punched through so there's no going back if you re-interpret and wanted to push the correct "Circle #3" for Gore... Attempting to re-select invaldates your ballot and they would not give you another, I believe. In an election that's this close this ballot has the capability of giving the entire state of Florida's electoral votes to Bush... even if the majority had "intended" to select Gore (albeit by a small margin). There's ample room here to put the results into arbitration (i.e. the court system). Let's not even consider the Electoral College system making the popular vote irrelevant... This election indicates that our system of selecting based upon "1 vote" being equal and the winner taking all based upon a majority decision. This is not sour grapes but simple analysis of the facts. I personally would love to see a re-working of this process in congress and a "run-off" election for President (in Forida at least) if it must be decided based upon current law (as I'm sure it must). We need to make the process a potent political issue that needs to be addressed by Congress.

-- Anonymous, November 09, 2000

I think the ballot "interface" is obviously poorly designed, but I think that we should expect voters to treat the occasion with great care and respect for the solemn process they are participating in. Regarding the level that voters tend to be informed: I have some raw 1998 polling data from an election studies think-tank. I ran a few analyses of various questions to satisfy my curiousity on a few matters. One result I got was that FEWER THAN HALF of the respondants who said they voted in the prior congressional election could name EVEN ONE of the candidates and his party. And this is a week after they just got done voting.


-- Anonymous, November 09, 2000

To me, the really scary part of this whole thing is that no matter what the outcome of the Florida ballot issue is going to be, just about half of America's population is going to feel cheated. Who's going to be the next president has turned into a political issue instead of the result of the people's voice. I think a lot of people are going to be very upset.

-- Anonymous, November 09, 2000

"Who's going to be the next president has turned into a political issue instead of the result of the people's voice."

No offense, but I find this somewhat humorous. An election is--by definition--a political process, is it not?

Personally, I'm still considering which is more unfortunate:

a) that this poorly designed ballot had an adverse effect on the accuracy of election process (regarding the relationship between who people meant to vote for and who they actually voted for), or

b) that a lot of people, as much as 14% (!) in some areas of Florida, were either so dumb or so careless that they punched more than one hole (and it's hard to say how many others punched a single, incorrect hole).

Right now, I'm leaning towards the opinion that to focus on a) is to not see the forest for the trees.

-- Anonymous, November 09, 2000

Let's remember that some regions in the US are composed of many elderly and uneducated individuals. The correct way to use the ballot is obvious to most of us, but that's really quite irrelevant. Statistically, it certainly does look like there is reason to believe that many mistakes were made. There are no constitutional provisions for requiring a minimum intelligence to vote, so all the "if you can't figure it out, you don't deserve a vote" comments are off target.

As for the ballot being approved beforehand, I don't believe it becomes apparent until the ballot is placed in the machine that it is not clear how to vote for whom. So that fact is mitigated to some degree. The ballot was approved out of context.

As for "every candidate received votes intended for someone else", I believe that is clearly false, too. The first candidate on the ballot, GWB, DID correspond to the first hole. And that is apparently how the voters who were making the mistake were viewing the correspondence.

As for what to do now, who knows.

-- Anonymous, November 09, 2000

I bet that when the ballot design was 'approved', they were just looking at whether everyone's names were there and spelled right. I really doubt usability ever entered anyone's mind, and I doubt it was actually tested with anyone who hadn't seen the thing before.

I think there are two main flaws with this hole-punching system. (Not the specific ballot, the hole-punch voting system.)

One is that it's difficult to see what you've done. It's difficult to see which holes have been punched. Seeing paper chaff on the hole doesn't necessarily indicate a punched hole because it could conceivably have come off the prior user's ballot. If you can't easily check your ballot and see that you've made a mistake, you're less likely to request a replacement ballot.

Mostly, I think the system doesn't scale well. For simple ballots, with a few items, it's easy to include lots of whitespace and separate the choices. Putting items on both sides, interleaved, is probably too much for the interface. Then again, it wouldn't do to group candidates such that they aren't all visible at once, so the system has to accomodate many choices.

The solution is probably to just use ballots that are twice as wide. In the contested Florida ballot, they could then use three columns for the holes, with the center column blocked in some fashion, providing two distinct and separate columns of holes for punching.

I don't think it would be good for future voter turnout if people's votes are being thrown out because of bad design, in an election where individual votes count far more than usual.

I also don't recall anything in the Constitution about voting being limited to those with good puzzle-solving skills. I don't think voters should be required to debug the voting system just to vote.

I suspect that those who say that people who mess up should forfeit their vote wouldn't be quite so nonchalant about, say, an ATM machine whose bad UI caused several hundred dollars to irrevocably vanish from their account.

-- Anonymous, November 10, 2000

The 19,000 invalid ballots with two votes for President might very well be due to mistakenly voting for both President and Vice President. The first two holes could easily be mistaken for Bush/Cheney and the second two for Gore/Lieberman. An analysis of the combinations of double punches might easily prove this theory.

Note that the instructions use the plural: "A vote for the candidates will actually be a vote for their electors. Vote for group."

Bob Spence bgspence@ao

-- Anonymous, November 10, 2000

No, Pat's not my uncle. But I'm a bit appalled at the talk about stupid and careless people. That tells me you've not read "The Design of Everyday Things" by Don Norman. Take a closer look at who you call stupid. Every day we all face memory issues, misinterpretations of interfaces (not the computer ones), and mistakes because our attention was focused somewhere else. Be careful with this attitude of calling this "stupidity." Such thinking can make you a slave to the punitive nature of our puritanical roots and belies a disciplinary arrogance. With the first, you'll paint yourself into the corner of your own closed mind. With the second, a lack of willingness to glean from areas outside your expertise could eventually be very limiting - personally and professionally. As Joel's many pages have stressed, it's the whole picture we need to look at. Usability is one of those parts of the picture.

-- Anonymous, November 10, 2000

I don't see the real Florida problem as the bad design of the ballot. Yes, the design was brain-damaged, and should have been changed, but the time for those complaints was earlier; it's too late for that now. For the next election Florida should be working on on-site ballot verification. Here in Wisconsin, for example, you stick your ballot into a machine that tells you if it's valid (you're responsible for marking the candidate you wanted to vote for, but it'll tell you immediately if it can read your ballot) so no one leaves without having cast a valid ballot.

But the troubling question to me is about the tales of people who knew they'd screwed up the ballot, and asked for another ballot to rectify their mistake but the polling supervisors refused to give them another ballot; to be blunt, they were denied their right to vote by some overly officious bureaucrat. Right now those stories amount to nothing but hearsay, and are certainly not enough to base any sort of drastic action on. But they should be investigated; if they're true then there's a serious problem down there.

-- Anonymous, November 10, 2000

Bev: Wow! I had no idea that I was so well on my way to becoming "a slave to the punitive nature of our puritanical roots" (assuming that I'm one of the people you're addressing with those comments). Your points are good, but these comments are over the top (in my case at least).

I have no desire to see any voter punished; that would serve no good purpose. Making an educated, well-informed decision as to who should lead this country and where is an extremely difficult thing to do well--for anyone. The issues are complex, and polticians are experts at manipulating public perceptions. But, _by comparison_, understanding a voting ballot (even one as badly designed as the one in FL) is just plain trivial. Yes, some people (including intelligent people) will still make mistakes sometimes--that's just life. But 14%*? That seems awfully high, and I do not buy the "bad ballot/people make mistakes" explanation for _all_ of those people.

My intent here is not to put people down. My intent is to express concern about the ability/willingness of voters to understand complex issues. Having a well designed interface for voting is very important, but is not the _most_ important thing. Understanding complex issues is.

* -- CNN reported on 11/8 or 11/9 that in some precincts in FL, as many as 14% of the votes had to be discarded because more than one presidential candidate had been voted for.

-- Anonymous, November 10, 2000

Consider countries with large uneducated populations and many "official languages" - take a look at the ballot used in the 1994 South African elections:

Party name, party logo, party abbreviation, candidate photo, large box. Voters need only "make your mark next to the party you choose" Instructions are in 11 languages.


-- Anonymous, November 10, 2000

Bev, voting is such an important and fundamental right we have in the United States that going into a voting both with your attention elsewhere is absurd. When making important decisions such as this your mind should be clearly focused on what you are doing.

-- Anonymous, November 10, 2000

Florida is not the only place in the country that uses the punch card system. Many counties use it and many counties are successful any using it. Why is it only in this one county and with an election this close that this is happening? I don't think it has anything to do with intelligence. Whether the name was second and the hole was third really doesn't matter because there was an arrow indicating which hole to punch.

-- Anonymous, November 10, 2000

Taylor: It may be that a similar thing did happen in other places, but all we're hearing about is Florida because the vote was so close there. I don't know if this is the case, but I think it's a possibility.

-- Anonymous, November 10, 2000

Taylor: I don't think it's the punch-card system; the problem is where there are choices on both pages, interleaved into one column of holes.

I experienced exactly this problem here in Chicago. Most of the votes were easy, because the choices were only on *one* side. The section for approving judges was doubled and interleaved, and I found it much more difficult.

Yes, there were arrows, but the booklet *moves* relative to the holes. If the arrows move, they aren't necessarily pointing at a hole, or at the correct hole.

And, it's difficult to tell where you've punched a hole, so you can't tell when you've made a mistake.

Apparently, in Chicago's Cook County (my county), 120,000 ballots were discarded due to problems.

-- Anonymous, November 11, 2000

-- Scott McCaskill (, November 09, wrote: << Personally, I'm still considering which is more unfortunate:

a) that this poorly designed ballot had an adverse effect on the accuracy of election process (regarding the relationship between who people meant to vote for and who they actually voted for), or

b) that a lot of people, as much as 14% (!) in some areas of Florida, were either so dumb or so careless that they punched more than one hole (and it's hard to say how many others punched a single, incorrect hole).

Right now, I'm leaning towards the opinion that to focus on a) is to not see the forest for the trees.


Dear Scott:

I strongly disagree with your "leaning". You seem willing to acknowledge that design errors in the ballot could have led to inadvertent votes for Buchanan from Gore supporters. I agree with this; but also, further consideration of the ballot design leads me to suspect that related design errors combined with misinformation could also explain many of the "double hole punch" problems. In fact, I think this explanation more likely than that in "some areas" of Florida 14% of people were inexplicably dumber or more careless than in other areas in EXACTLY the same way. Doesn't that very statistic make you stop and ask whether there might not have been a structural problem involved?

Consider this: if you did not know that you were to vote only once for BOTH president and vice president, you might very well expect to punch two holes instead of one. And, because of the butterfly design, the hole for a candidate from the other "leaf" almost lines up with the name for a Vice Presidential candidate. This could have affected voters for EITHER Bush/Cheney or Gore/Leiberman. Look at the design and think about the reality of seeing it in the booth for the first time. Since I suspect these invalid ballots were tossed we probably have no way to know.

Alternatively, if you punched a hole and then realized you had made a mistake and tried to fix it with a second punch - admittedly irrational, but an understandable human reaction in a situation where, in cognitive dimensions terms, you have "premature commitment" design problems with a vengeance -- ... then some of the dual votes might have been frustrated reactions to the other confusing aspects of the ballot.

Apropos of all this, consider that a memo was apparently circulated to all election officials prior to the election, warning them that the layout was unclear, and telling them to advise people that they had to vote just ONCE for president and VP, and how to punch the holes. Apparently that extra verbal instruction was rarely offered if at all (understandable under the circumstances). Further, the text on the ballot itself only vaguely and indirectly suggests these instructions (vote for group). Also, some people realized they had made a mistake and requested another replacement ballot while still in the polling station and were refused one by election staffers.

The dual votes should be viewed as additional evidence that there were structural problems with this ballot: not thrown out as an argument to blithely dismiss the rights of voters for their votes to be counted due to imputed stupidity or inattention. Bad enough that the general public reacts this way -- from those with professional background in these usability issues I'm dismayed to see such signs of -- shall I say, inattention?

Mark Simos

-- Anonymous, November 11, 2000

let us also not forget the -two- sources of error, constant (or bias) and random (or noise). all ballots will be subject to some level of random error (mis punches and mis scores due to random causes such as eye-motor-coordination, dust and dirt in the scoring machine) and some ballots will be subject to bias. If the scoring machine is mis-alligned so that some batch of scored votes are 'off by one' or some ballots are misprinted then we have bias, a relatively consistent shift in scoring. On the other hand, random error should exsist at about the same levels in -all- ballots, this is the 'error' of statistical fame. Bias should not.

Personally, I see the ballot design as contributing to bias, not to the random error. Similarly, deciding an election on the basis of all of these recounts seems OK -but only if the difference of the scores exceeds the expected random error. I'm guessing here but I'd be willing to bet that running 1,000,000 ballots through a scoring machine would result in at least some variability in the scores (vis reliability). Does anyone have any data on what the scoring random error is? I heard at least one antecdotal report on NPR stating that the various scoring machines are not reliable.

-- Anonymous, November 11, 2000

It is definitely a usability bug. American and the western world are trained to read from left to right, top to bottom. The Florida ballot required people to observe the ballot carefully before they punched the hole. If you are in a rush to get back to work, most likely there will be an error in punching the hole. Who read the manual, any manual? Do you read the manual on your VCR? People just use their common sense such as from left to right, top to bottom. Florida's ballot definitely has a user interface problem. I wrote automation software for well known company, and the radio buttons were always going down from top to bottom, never zig-zag like the Florida's Ballot. I always tested on real people before putting it out to production. Why didn't they tested the Florida's Ballot on some small scale before putting it out to the Presidential Election?

-- Anonymous, November 11, 2000

Tonight, 11/11/00, the Palm Beach county election officials announced unofficial totals for the manual recount for four precincts representing approximately one percent of the total votes in the county. Bush picked up 14 votes and Gore picked up 33 votes for a net increase of 19 votes for Gore. This extrapolates to a net gain of approximately 1900 votes for Gore in the entire county.

They also announced counts for over punched and under punched ballots. These represent ballets, which were either not voted for president, or were punched more than once for president. The results of those counts reflect additional usability problems with the Palm Beach county ballot. The three most common over punches were for the combinations 3 + 4, 3 + 5, and 5 + 6.

Eighty ballots were punched with the combination 3 + 5. This represents a vote for Bush and Gore. A reasonable explanation for this selection would be a voter choosing one of the major candidates, and then changing their mind to chose the other.

Eleven ballots were punched with the combination 3 + 4 and twenty-one ballots were punched with the combination 5 + 6. These reflect a vote for Bush and Buchanan or Gore and McReynolds (the Socialist candidate). The other punch combinations found occurred five or fewer times.

Why did the Palm Beach county voters find these combinations of candidates attractive? The ballot itself provides the answer. In many elections the voter is expected to chose more than one candidate from a list of candidates. In the case of voting for President one must chose both President and Vice President. If a voter chose to vote for Bush and Cheney, they might easily chose to punch the holes for these candidates which are immediately to the right of their names. A Bush/Cheney vote would be holes 3 + 4. A vote for Gore/Lieberman would be holes 5 + 6.

The ballot instructions say "ELECTORS FOR PRESIDENT AND VICE PRESIDENT (A vote for the candidates will actually be a vote for their electors) (Vote for Group)." These instructions use the plural. They instruct the voter to "Vote for Group" not "Vote for a group." This is ambiguous. Most voters got it right, but some got it wrong. They punched once for President and once for Vice President.

Using this interpretation Bush would have received 11 votes and Gore would have received 21 votes. This is a net gain of another 10 votes for Gore. This extrapolates to another 1000 votes for Gore in the county. This reading of the meaning of the punches reflects very closely the preference for the Gore ticket in Palm Beach County.

If the 80 votes for both major candidates are ignored, the will of the voters of Palm Beach County reflected in a hand recount would be approximately 2900 votes.

If the 80 vote switchers were included and we assume the countywide ratio of 152954 Bush to 269696 Gore votes, we might apportion them 30 for Bush and 50 for Gore. This would be another net 20 for Gore, or 2000 votes countywide. This would award Gore a 4900 vote gain in Palm Beach County. It would probably be unfair to award these votes to Gore without applying similar methods to all the counties in Florida. Those results would probably be a wash.

The gains awarded Gore due to the badly designed ballot are unique within Florida. It would be reasonable to award those votes to Gore if the purpose of a recount is to determine th

-- Anonymous, November 12, 2000

First, it is unclear whether the 19,000 rejected ballots were rejected by the machine or by the voter. Florida law allows a voter three chances to get it right. If you screw up a ballot, you can ask for another one and try again. The failed ballots were marked as invalid and placed in the pile of ballots. It is illegal in Florida to destroy any ballot.

One should note that the machines that are used in Florida are old and have a high rejection rate. If that same rejection rate is used throughout the state then the overall result will balance out. If we hand count only democratic districts then there is no doubt that we will find more voes for Gore. If we do the same in Republican disricts (where the rejection rates were sometimes higher than in West Palm Beach) we would expect to see more Bush votes.

The manual counting also leads to problems. These ballots were designed to be read by a machine not a person. Does a "hanging chad" or "swing chad" really mean that the person meant to vote for that candidate? Or does it mean that the ballot has been handled so many times by so many people (all of who are pro-Gore) that perhaps the Gore chad just happened to come lose? Then we say a 3+4 means this and a 4+5 means that and 5+6 means something else. How can we know what was on the mind of the voter when they were in there casting their votes? Are we going to determine the president of the USA by hiring Uri Geller to wave his hands over each ballot?

As far as fraud, it seems that Wisconsin is the state to look into, not Florida. If there was no voter fraud, Bush would probably have won the state easily. Many college students voted 25 times or more for Gore in an election decided by just a few thousand votes.

-- Anonymous, November 15, 2000

If 4% of the voters had problems, that tells me 96% didn't have problems. This is ridiculous; 2nd and 3rd graders have been given similar ballots with Disney characters on them, and haven't had problems.

I'd love to see all states go to ballots like this one. Weeding out the folks at the bottom of the clue-distribution curve would be fine with me.

-- Anonymous, November 15, 2000

Ok. The ballot was poorly designed. There is already a margin of error. The margin of error was increased by the design. Hopefully something will be learned from this.

Ok. The voters should have been more careful. Maybe in the next election, more people will pay attention to what they are doing.

It doesn't matter for the outcome of THIS election. The election is over. Any "objective" counting by hand will not give the counters the ability to read the mind of the voters, and the very act of touching and re-touching the ballots (which are not "read only" devices) invalidates them.

-- Anonymous, November 16, 2000

Let me attempt to set aside the political component for a bit by talking not about a person designing a ballot but about a team of engineers designing the user interface for, say, a microwave oven. After the initial design is complete, all of the engineers agree that they are happy with the interface, and that it's time to invite some past customers into the usability lab to test it out.

To the engineers' horror, things don't go well. Out of a large sample of customers, 4% are unable to make the test recipe--pineapple chicken surprise--without starting a fire. The very same recipe is tested on other microwave ovens, and no more than 1/2% of the experiments end in flames.

At this point, the engineers have several choices:

a) Point out that not a single engineer had a problem with the interface, and that the most likely cause of the test problems was the outside testers, who, to all appearances, are a bunch of idiots.

b) Note that 96% of the customers had no problem, conclude that that's good enough, and conveniently forget to mention the decision to the company's liability lawyers.

c) Admit that there's a usability problem, and try to fix it.

Many of us own microwave ovens and VCRs designed by engineers who would go with (a) or (b). But the fact of the matter is that whether we as engineers have problems with an interface is irrelevant. If we design a new interface that produces a far higher interface error rate than other designs, we have a usability problem, by definition. And we have a problem regardless of whether we can come up with an explanation for why the new interface doesn't work. The problem is there, period.

Back to the ballots. When I first saw the picture of the now-infamous butterfly ballot, I could not for the life of me explain why anyone would have a problem with it. I've read explanations of the factors that may have caused difficulty for voters, and while some or all of them may be right, I know only one thing for sure: the ballots are badly designed. The mispunch statistics are tough to argue with.

Recounting manually seems reasonable to me, but I can't think of a remedy to the double-punch problem (e.g., re-voting, reinterpreting double punches, etc.) that's fair. We may just have to resign ourselves to the fact that bad industrial design decided the outcome of an election. But when an election is this close, you can point to just about anything as the determining factor, and have a good chance of being right.

-- Anonymous, November 18, 2000

I am going to go out on a limb and assume most of the participants here are programmers. Programmers shouldn't design interfaces. Period. Users should. So although all this reasoning around interfaces is fairly useful and crucial to improving our sensitivity to how interfaces affect the world--we could be doing better things. I think the one thing that deserves a good look by all of us is computerized voting system over the internet. Can any of you suggest ways to make it cryptographically uncheatable? Or cryptographically secret? Or cryptographically anonymous? Or just plain fair? Or most responsive to amazing large democratic societies where people often feel like their vote means little--no matter how much you say "every vote count". Etc.

I think a discussion on this will show what sharp cookies subscribe to Joel's mailing list.

I'd like to be the first to show my ignorance of how computer voting works:

My understanding is that privacy is dead. You could easily rig web scripts and desktop operating systems to snoop with keystroke recorders and mouse motion loggers. For example, a javascript-based poll could have a helper script which could tell you that someone held their mouse for so and so moment over Gore but all of a sudden picked Bush instead. Do people view source any more? Will they object to this? And how do you answer to such an objection? I would just refer them to "Core Complete" or some interface book saying we are doing interface testing.

Well what if you eliminate the desktop somehow.. like with one of those Java or ROM/RAM iButtons we heard about back a few years. So you wrote a IE5 plug-in.. it's got RSA encryption library.. and something proprietary and secretive chip is doing the talking between this library of stubs and the actual serially connected ibutton? So great, your vote actually makes it all the way to the central machine without plain text snooping! So this is a solution. But will the USA be able to ensure every PC have this chip in the future? Who will review this chip? How can we be sure that the company making the chip won't build in a back-door?

So this is back to the issue of trust. Can I trust someone to create a hardware tunnel between a SSL/PGP connection to some central voting database and a hand-held voting device? If I could, how? I guess we need a way to have a chip foundry and chip plant so open that people can walk in on a museum-like tour and learn from a everyday guide (MIT freshman?) and verify with common everyday algebra that the NSA don't have a say in what blueprint of the chip actually gets printed. It doesn't seem feasible or possible. So openness will be difficult.

If we can have a system that allows everyone to vote online. Can we trust it to help us make an inform choice too? I don't know now... The graphics designers and Cambridge stamped P.R. specialists and Oracle DBAs all work for special interest groups. They don't work for joe million who wrote their opinion on what matters in their neighborhood. We like the glitz. The sharp and lucid writing. We tend to trust the professional voice. Even if it's wrong. This bothers me some. I don't really want everyone to become a Adobe illustrator expert or know how to make web community websites like Philip can in order for them to share their ideas on which candidate will best serve my neighborhood. Perhaps we need to inform some 250 million people that they ought to check LUSENET out and chat here. Perhaps we should aim to make LUSENET scale to that level.

So very few of us will make an informed decision. And since we all know by now that the specialists who work for our most trusted news source also face the same uphill informational challenges--we have to worry even more about taking in news coverage as the foundation we use to make our final decision. It's a signal to noise problem. I think this will have to start with helping ourselves. If average citizens have access to a smarter online forum to learn things from. Professionals will too have an idea how to learn smarter.

Totally random thought: I have just read the Persuasion book as recommended by Joel. And now I wonder if all we do at a voting booth is a psychological thing. If it is the physical action of publically waiting a long time in line.. and physically pressing a button.. that makes us consistent in supporting positively (through inner justification) what our politions do. I know the world isn't that simple, but will politicians insist on having people make this physical trip just for that psychological effect? If this is essential to their success it's another blow to the adaptation of computer voting.

The problem with online voting system is mainly 1. how well we utilize cryptography to ensure no cheating.. and 2. how well we inform the voters.. and 3. how to make regional decisions better than trying to encapsulate all our regional wishes into a single candidate who promises too much to too many special interest groups... A. It is clear to me with the Internet we can make use of the network effect to better inform the everyday real world voter with regional problems. B. It is also clear that with just a few strategic hardware additions.. the unnecessary secrecy-violations can be reduced. C. It is not clear to me how we can make voting work well on small regional issues.

-- Anonymous, November 20, 2000


You went out on a limb and missed the point of the discussion. Also, don't assume the occupation of your audience.

We are not talking about online voting we are discussing the usability of the ballot.

-- Anonymous, November 20, 2000

I was sort of hoping I didn't miss the point of the dicussion Mark Smith :) 1. First I make my stand that usability comes from full contact observation of how a user thinks. Labs. Lab coats. The whole shabang. It's not something you can reason up thoroughly as a programmer. 2. You're right, I shouldn't assume that people who subscribe to Joel are programmers. My bad. 3. I am not American, so I don't have the right to share my opinion on regional or national issues to close to you guys heart. You guys duke it out among yourselves! :) But I will talk about things that affect all countries looking to computerize their decision making process--and that's what I have decided to talk about. You can't talk about usability of punch cards without talking about the feasibility of computerized ballot systems I'd think.

-- Anonymous, November 20, 2000

The photograph of the ballot, being 2 dimensional, can't tell us how deep the metal guide is nor how much play exists underneith the punch card for the punch to do it's job. I heard on the radio today that the hole is somewhat deeper then it seems and that the last 1/16" of play is the range needed to definitly push the chad through. This would mean that the typical 4th grader, who has a lot of energy, would have no problem ramming that punch down, where as a feeble Jewish pensioner might be unsure just how hard they need to push, not wanting to break anything..

And don't forget how alien these new-fangled contraptions are to the elderly. I would expect some of them to be a little intimidated by it all.

-- Anonymous, November 22, 2000

Bob Perlman wrote: "But the fact of the matter is that whether we as engineers have problems with an interface is irrelevant."

Specifically, it's irrelevant because the designer of an interface and the user of the interface face completely opposite cognitive tasks. The designer's task is to go from knowing what the interface is supposed to represent to creating that interface. The user's task is to figure out what the interface is supposed to represent. The designer already knows the latter, and no matter how hard he tries, he's not going to be able to mentally block that knowledge while evaluating the interface. The designer's task is *synthetic* while the user's task is *analytic*.

In fact, very smart people often forget that the correct completion of a task requires not just intelligence, but knowledge as well. And when we forget that, and discover that someone else has trouble with a task that we find easy, we forget that that someone else isn't bringing the same knowledge to the task that we are, and that the difference in task performance could be due to the difference in knowledge rather than a difference in intelligence. Intelligence can't substitute for knowledge, any more than a faster processor can retrieve more records from an incomplete database.

The designers of the ballot knew exactly how it was supposed to work. They knew that well before they even designed it. The voters, OTOH, *didn't* know that; they had to figure it out when they got into the voting booth. They had a very short time to figure it out (there were plenty of people in line behind them, creating a psychological pressure to hurry up). There were plenty of visual cues (the arrows) showing how to use the ballot correctly, but there were also plenty of secondary visual cues (such as the lines separating the candidates) that were pointing in the wrong direction. Everything we know about perceptual psychology says that conflicting cues are going to lead to mistakes. Those of us who are highly practiced in the art of abstract symbol manipulation (most of us could correctly solve a logic problem in which we were told that the word "true" was to be understood as representing logical falsity and that the word "false" was to be understood as representing logical truth) will probably make fewer mistakes than those (i.e. the vast majority) who aren't. But the US Constitution does not provide for intelligence-based, experience-based, achievement-based or aptitude-based qualifications for voters, for the simple reason that everyone is equally subject to the laws passed by elected officials.

-- Anonymous, November 27, 2000

But what it comes down to is what proof is there that the users actually did have a problem. The answer... none. There is an after- the-fact claim by a few people that were prompted by questions ("You did have a problem, right?") but no statistical evidence that anyone actually had a problem. Let's look at he facts:

Pat Buchanan got more votes in that county... but in every election in which Buchanan has participated he has gotten more votes in that county than any other Florida county. In fact, every independent on the ballot for every office got more votes in that county than they did in other counties.

A reasonable assumption would be that if there were a large number of confused voters we would see other Democratic candidates with more votes than Gore. We would expect to see, for example, that a large number of Buchanan voters also voted for the Democratic Senatorial candidate. We simply do not see this. The difference between Gore and the Democrat running for Senate was less than 500 votes. But, and this is very significant, the difference between Bush and the Republican Senatorial candidate was more than 2,000 votes. So the Buchanan voters most likely voted for the Republican senatorial candidate which would lead one to believe that Buchanan voters were not confused at all. They were Republicans who didn't like Bush. This is exactly where you would expect Buchanan's strength to be found.

So the result is that there was no confusion with the ballot. The confusion is an urban myth created by the pro-Gore forces and pushed on us by the media.

-- Anonymous, November 29, 2000

Just to clarify: I specified that the difference between Bush and the Republican senatorial candidate was more than 2,000 votes. I didn't say it (although it should have been clear from the context), the Republican senatorial candidate got over 2,000 votes more than Bush.

-- Anonymous, November 29, 2000

There was no proof of a problem? An article in the San Jose Mercury News a couple of weeks ago said that the double-punch rate in Palm Beach County was about 10 times that of any other county they surveyed. Was this article incorrect?

-- Anonymous, December 02, 2000

And how many other counties did they survey? Does the equipment they used for the comparison even permit double-punching (it is not possible to double-punch a ballot in my district). And if double- punching was such a widespread problem, why isn't it reported more widely than in just the San Jose Mercury News?

-- Anonymous, December 04, 2000

As I recall, the Mercury News surveyed perhaps 12-15 other counties that used punched ballots, and looked for the highest rates of double punching they could find. The point of the article was not that double punching is a major national problem, but that the rate of double punching was about 10 times higher for the butterfly ballot in Palm Beach County than for non-butterfly punched ballots used elsewhere.

I don't read other papers regularly, so I don't know that other papers *haven't* reported the very same problem. The question is not whether other news sources have reported it, but whether it's true.

And remember, this is a discussion about usability, not about who won the election or should have.

-- Anonymous, December 09, 2000

Everybody's been talking punch-card ballots' usability from the voter's point of view. But the latest round of legal battles reveals usability problems from a vote counter's point of view.

If a county that used optical-scan ballots was doing a manual recount, I have a hard time imagining disputes over ballot interpretation that would affect the outcome of the count. If someone wrote an X where they should have filled in a circle or made a horizontal line, it's easy to tell that the voter intended that as a vote. Likewise, if there's a stray mark where the rest of the ballot has clearly marked spaces, it's easy to tell that the voter did not intend that as a vote.

But the link between a partially dislodged chad and the voter's intent is harder to determine, because punching a ballot with a stylus is more complicated and more unfamiliar to everyone involved than marking a space with pencil or ink.

-- Anonymous, December 11, 2000

I personally think that lever voting machines are the best. There are no ballot cards to count so there is no need to go through this nonsense.

It's difficult to determine by looking at a handful of counties as to whether there truly is a problem with the butterfly ballot. It is possible that the problem was not caused by the ballot but rather by the age of the voter. IOW, if the voter had some other type of ballot they would have managed to screw that one up as well. T

he interesting claim is that the voters got confused and voted for both president and vice president (thus the double punching). But since the claim is that it was old voters who were confused by the ballot, wouldn't these old voters know from having voted countless times before, that you don't vote for these two positions separately? If there was a group likely to screw it up, wouldn't that group be more likely to be first time voters?

-- Anonymous, December 14, 2000

When should you recount ballots? Absent clear reasons to suspect the results are based on fraud or other misconduct by officials, the closer the outcome is, the less you should be willing to recount. Why? Because there is a certain amount of error you simply cannot eliminate no matter how you count or recount ballots. Call it counting noise. Thus, the "true result" cannot be known. Ever.

Since the closer the election is, the more likely any "true result" that could change the result is in the noise, a recount is not likely to get a more accurate outcome. You're better off using the original result, which is already as good as any you'll get by recounting and has the obvious benefit of being final sooner.

-- Anonymous, June 21, 2001

Moderation questions? read the FAQ