It’s so hard for me to frame questions well. I got cute and attached numbers to the two plans, and in doing so obscured what I meant to be the real question. Chronos, you’ve analyzed things how I intended, with those numbers, but others haven’t.
The question I intended to discuss with this hypothetical is this:
Is it better to exclude fewer voters, but in a way that swings more elections? Or is it better to exclude more voters, but in a way that swings fewer elections?
What makes you think it is actually possible to predict the effects of either plan with any degree of accuracy?
Honestly, that question is outside the scope of what I want to talk about.
I am being asked to make a choice between:
A small amount of fraud that has a high likelihood of producing results contrary to the true wishes of the electorate
A larger amount of fraud that has a lower likelihood of producing results contrary to the true wishes of the electorate.
It’s at times like this that I, a mature and seasoned head of Board of Elections, refer to my job description. Does my job description place more weight on ensuring that the results of the election match the wishes of the electorate, or on minimising the incidence of individual fraud regardless of the impact on the relationship between vote results and the electorate’s wishes?
It turns out that I am so mature and seasoned that my boss let me write my own job description. And my job, as I see it, is to ensure the closest possible match between electoral outcomes and what the voters wanted. A scheme in which I knowingly place a thumb on the scales by disproportionately allowing fraudulent votes to count for one side over the other is a scheme in which I am not just failing in my duties but acting against them.
Letting the greater amount of fraud through is not a great option! Every fraudulent voter is trying to skew the results of the election so it doesn’t reflect the wishes of the voters. This is anathema to me. But, through no fault of their own, their efforts are extremely likely to fail. Whereas if I set up a scheme which selectively prevents one set of fraudulent voters from succeeding while giving a boost to the efforts of the other set, then I am effectively aiding and abetting the latter. And it can’t be right for a mature and seasoned head of the board of elections such as myself to knowingly collude with vote-riggers.
Thank you Chronos, that was very kind indeed and did allow my point to stand for what it is worth (which is not much in this example). When we are talking about the difference between fractions of one percent and less the question seems to be posed to arrive at a predetermined answer. Further, as:
PatrickLondon asks, what real world application does the question hold? While I understand there is a principle here to learn, and I appreciate Left_Hand_of_Dorkness for introducing me to the concept. That being said:
It seems to me that without predicting how accurate the model can be (and the numbers used seem absurdly tight and specific – however I defer to those who would know better than me), we can ask a better question. It seems to me that men and women as clever as you folks can determine which Plan to use in an election of x number of voters if the accuracy is y: (and this is beyond me, I am seriously asking)
Five hundred voters when Plan 1 can be .85%, and Plan 2 will be 3% (which is a closer margin than the original - 400+% vs 500%)
Five Thousand voters when plan 1 is 1% and Plan 2 is 5%
And as a last question, I am aware of engineering with remarkably fine tolerances. There are parts in jet engines that operate above their melting points because of very small holes (10,000thds of a millimeter if I recall correctly) and if there are more holes, or the size is even slightly different on one side vs the other – hot spots will develop and the part will explode under normal use. On the other hand, polls are considered tight when they are about 3 - 3-1/2% but in an election there are actual votes being cast so perhaps a more tangible number is obtainable – although it has to be based upon a forecast because you do not have the total number of votes until after they have been cast. So how accurate can these models potentially be in a real life situation? (And sorry it took me so much verbiage to get the question out.) Two One Hundredths of a percent seems rather ideal to me.
I honestly don’t understand why people fight the hypothetical in these threads. He’s not suggesting there’s a real world equivalent where this situation would occur, or that these 2 plans would be enacted with the exact results proposed. It’s an interesting question he proposed without trying to punch a bunch of hypothetical holes into it.
I’m on Team Plan B. A systemic error that affects all sides equally is far better than one that discriminates against one of the subgroups.
I agree, I do like the hypothetical question and it would have never occurred to me on my own. That being said, if there is not a way to apply it to any real world application it becomes little more than a parlor trick.
It seems to me, once you know how to do the math (and I do not – but obviously Chronos and others do), the smart guy (or gal) is the one who knows in which cases to to use method A (they were one and two in the OP), and in which cases to use method B. That is assuming in different races the model might more or less accurate because of outside factors (?? demographic uncertainties, potential envelop problems, voter turnout estimates, whatever??)
For my part I am not trying to shit on the exercise, I am trying to learn how a marginally educated (at best) blue collar mook like me can apply the lesson in my own life, win the girl, get the promotion, impress the right people and retire into luxury. Is that too much to ask from an online community dedicated to eliminating ignorance? Geez, I’m not asking you to make taller, younger, and better looking; just smarter and perhaps how to be a better smart ass.
I got to thinking about this question because of voter ID laws, which propose this tradeoff. They propose to tighten up voting to avoid miscast votes, but the errors that they create tend to consist of preventing people from voting who are eligible to vote, and the people prevented from voting tend to be Democratic voters. They’re very much Plan 1 approaches: reduce the number of errors, but concentrate those errors among Democratic voters. The GOP tends to support Plan 1 approaches.
Plan 2 approaches tend to be supported by Democrats: they loosen the rules around voting, allowing (for example) votes to be cast without the use of IDs, or absentee ballots to count even without correct signatures, or the like. The argument there is that these loosened restrictions may allow votes to count which shouldn’t be counted; but the errors in the system will not be concentrated in one party or another. They’re Plan 2 approaches.
In arguments between GOP and Dems, it often sounds like the two sides are talking past each other. GOP folks tend to talk about how an individual’s vote is meaningless if the vote isn’t secure: they’re more concerned about the process than the outcome (in their arguments). Dems tend to talk about how an election that disenfranchises a group and leads to an outcome contrary to what the majority wants is bad; they prioritize the fair outcome over the fair individual occurrence.
I know these tendencies have exceptions, but broadly speaking, this is what I hear. I wanted to discuss this without bringing partisan specifics into the discussion.
Maybe I’m wrong, but I have a somewhat different reaction to what you’ve said. My sense is that “liberals” prioritize (or say they do) the individual’s right to vote as a significant (even paramount) value in and of itself. So, while I would agree that conservatives emphasize process, I don’t think I would agree that “liberals” priortize outcome over individual occurence – rather, they value individual occurence over process (and perhaps define “fair outcome” differently). If the right to have your individual vote counted is the paramount right, then you’re going to err on the side of permitting illegitimate votes to ensure the counting of all legitimate votes. I think that attitude would augur in favor of your Plan 1 – it’s not about the .2% of voters that are disenfranchsed under Plan 1, it’s about the 1.8% that only get to vote under Plan 1.
A lot of the contemporary debate gets caught up in accusations of ulterior motives or schemes. A lot of it depends on factual claims that are dubious (we’ll disagree on which ones are which, to some degree). But to me (and this isn’t limited to voting, I’ve had a similar debate with statutes of limitations on this board), a clear set of rules creates a workable process that yields an acceptable outcome, even if it “harms” some number of individuals (i.e., all rules are both overinclusive and underinclusive, but it works out in the end, and the efficient administration of the system has its own value).
I expect that to be seen as a “conservative” view (while the “liberal” view would focus on the individual who was prevented access to some right because of a blind application of the rules). But maybe it’s not, because so much of the debate ends up reverting to claims that the rules aren’t facially neutral (and/or aren’t intended to be).
The attempt to quantify your plans (and the knowledge in advance of the partisan effect of the rules) is what makes your hyopthetical difficult. I’d almost be inclined to pick Plan 3.
I appreciate that response, Falchion. Yeah, I’m definitely oversimplifying things. There is a liberal tendency to say that letting some illegal votes slip through is better than discouraging legal votes, whereas a lot of conservatives say the exact opposite. My intention was to tease out a different tendency, though, which I think is accurate.
This also shows up in census counts. Years ago (2000?) there was the possibility of gaining a more accurate census count by using statistical extrapolation. Everyone seemed to agree that it was more accurate; but conservatives disliked it, because it added something besides individual counts to the census. The outcome was, again, less important to conservatives than the process, and a less accurate count was preferable to changing the process.
Plan 2, definitely, because both parties suffer equally, which makes it as “fair” as possible.
This changes the entire matter for me. The question as I first read it was a math problem pure and simple. I will always view it that way probably because of first impressions, or perhaps due to the lengthy thought process invested in it as a math problem.
What you are proposing now is a moral question and uses a whole different part of my brain (and heart). I am myself without a party at the moment so it is quite easily to be nonpartisan in my case.
As a long time Republican who HATES the current GOP, my initial reaction is often conservative. As a thinking human being, my reaction upon reflection is almost always liberal. In addition, I admire Biden and have no respect for Trump, but neither party appeals to me at this point in time. I know exactly what you are saying about talking past each other. Let me give you an example.
I am pro life;I am also pro choice – there is only one place they overlap and that is abortion. Abortion isn’t the issue to anyone highly invested in the topic; it is the result or the symbol of their views. Personally I do not see a conflict between allowing a woman to decide for herself if she will use birth control or receive an abortion – and respecting human life. Oddly, most opposed to abortion also favor capital punishment. That is inconsistent at least and possibly hypocritical.
Here is how I would frame the entire debate: Using abortion as birth control is less than ideal; try to anticipate that eventuality and find other solutions. Full stop. In addition, I will defend any persons right to consider THEIR OWN pregnancy as a fully valid human life from the moment of conception. I will not support the right for that person to make that decision for all others. Now, on the other side of that debate I will say this. I know a woman who was the youngest preemie to ever survive and live a full life (not sure if she still is - and she does have minor medical challenges). I would say she demonstrates a viable, separate life form independent of her mother. I do not recall the detail of which week she was born but it was about the third trimester point. I would suggest that right around that many weeks, abortion should be a much more extreme consideration. The mother could give up the child and let the hospital take it the rest of the distance. I am not going to call an abortion after that point murder – but neither do I consider it good or right or moral or an automatic right. You did have six or seven months to address it and you didn’t; now the child has rights as far as I am concerned. This very personal (and impossible) view is only to demonstrate that compromise is possible even in a very controversial matter and that the reasons for the stance is more important than the stance itself.
At any rate, now that I see the question as philosophical and a moral matter I will consider it on that basis and reply. I would still like to have answers to the entirely hypothetical questions I asked above because the use of advanced math to solve a problem rather than dead reckoning very much appeals to me.
See, I think they disliked it because it was more likely to count people who would dilute Republican political power. Same as with Voter ID, they aren’t taking a position because it is more accurate, they take it because it solidifies their political power.
There is no evidence that Plan 1 (Voter ID) reduces the number of errors. In fact, there is a legitimate claim that Plan 1 increases the number of errors by multiple orders of magnitude, as long as you count “preventing a US Citizen from voting” as an error. The reason there is no evidence that errors are reduced is that there has been no study of (Voter ID preventable) voter fraud, despite assertations that it is a problem.
Your hypothetical assumes a good understanding of voter fraud, who is impacted by it, and how each plan influences it.
I’m not sure I understand what you’re trying to tease out. (And I don’t know enough about the census to have a strong opinion on it, other than that I think the constitutional text requires an “actual enumeration” and that stands somewhat in tension with a statistical model. So, one potential problem is that a statistical extrapolation approach may suffer from the disadvantage of being unlawful).
You propose that a system which accurately counts the votes of 99.8% of voters is less accurate than one that counts only 98% of voters’ votes. And that’s because the hypothetical assumes that we know the number of votes that won’t be counted; the partisan breakdown of those votes; and (I gather) that there is nothing that a voter can reasonably do (or refrain from doing) to make his vote more or less likely to count (the fickle finger of fate disenfranchises at will). (And I’m not sure we ever know those things (although we often claim to)).
And, I gather the argument is that the priority is to conduct an election that accurately reports the “will of the people” (assuming we can ever know that) instead of ensuring any individual’s right to vote. And I don’t necessarily disagree. I guess I’m just not sure I understand why this is the “liberal” view (and, if it is, I’d be inclined to propose that it is because we assume that liberal voters are most likely to be disfavored in Plan 1 – although that’s, obviously, not part of the hypothetical).
Edit: I guess my point is that seems pro-“process” to me.