Selfish Reasons to Let Felons Vote
As a dovetail to the post on the felons arrested for “voter fraud” in Florida, it’s worth diving into some important minutiae that’s otherwise glossed over.
First and foremost, it’s helpful to establish some precision over the vocabulary used when discussing voter fraud, which can be split into different types [I pulled most of these categories from the Souter dissent in Crawford v. Marion]. For example, “impersonation fraud” can refer to voting under a fake name or someone else’s name. “Eligibility fraud” can refer to casting a ballot when it is not legally permitted to do so (e.g., voting twice, a disenfranchised felon voting, etc.). “Harvesting fraud” has a lot of overlap with the first two categories, but it more specifically refers to instances where the ballot harvesting process (generally legal in most states but a felony in Alabama) is somehow manipulated. An altogether different type is “tabulation fraud” where the ballot-counting process is somehow compromised. Any others?
Of course the common denominator here is fraud, which implies some element of deception, which in turn implies some element of intent. This reflects the statutory language that criminalizes voter fraud (e.g., Fla. Stat. § 104.16), which requires the government to prove the accused acted “knowingly”.
In case it needs to be said again, voter fraud absolutely happens. Just because Trump’s 2020 election fraud claims flamed up in a pyre when asked to provide evidence does not mean that voter fraud never happens. We know that voter fraud happens, because people get prosecuted for it, like this Nevada guy who submitted his dead wife’s ballot and was sentenced to probation. We also know that not every criminal gets caught, which means there undoubtedly is some voter fraud that went by undetected. How much happens in total and whether it’s enough to matter is a separate discussion.
For this post, the relevant category is eligibility fraud, and specifically that which stems from felony disenfranchisement. This may come across as specious, but it is perfectly possible to eliminate this kind of eligibility fraud 100%: Make all felons eligible to vote.
It’s not that crazy, because the norm across the world is letting people vote from prison. Literally ballot boxes installed in prisons. To the extent there are any limitations imposed, they’re doled out selectively, with apparently fewer than a handful of countries even considering restricting the vote of criminals post-release. In contrast, the United States is rather unique in its disenfranchisement zeal. Only Vermont, Maine, and DC allow voting from prison, but otherwise, the norm in most other states is automatic voting restoration upon release. In total, about 4.6 million Americans can’t vote today because of a felony conviction, which is about triple the percentage it was in 1976, but down from a peak in 2016.
Despite all the words here, I’m actually not someone who particularly cares about democracy. While I can acknowledge the strong correlation between democratic governments and overall quality of life, I’m in the consequentialist camp on this issue. Give me Hong Kong under British colonial rule over democratic India any day of the week. Beyond that, voting is a waste of time on an individual level and not something I ever engage in (to answer the tiresome what if everyone thought that? retort: “Then I would vote”), and my anarchist foibles generally leave me politically stranded.
But my egalitarian foibles are why felony disenfranchisement bothers me. A steelman could be either consequential or an appeal to fairness. If you take a “wisdom of crowds” defense of democracy — that it is a mechanism to arrive at better policies — then perhaps giving former criminals a say would lead the ship astray. But most of the world seems to function OK despite letting criminals vote, and neither Vermont or Maine seem notably dysfunctional in any way (maybe DC does, but not sure how much you can pin that on the voting prison population). But even if consequences be damned, perhaps violating the social contract is cause enough to muzzle you. I concede it’s a slightly stronger argument, but I’m not convinced the justification isn’t used as a pretextual excuse to tip the scales in some political party’s favor. This wouldn’t be a novel effort, as Mississippi implemented literacy tests and poll taxes in 1890 with the express purpose of indirectly suppressing the black vote without explicitly violating the 15th Amendment. The state’s governor, James Vardaman, said outright in 1903 the restrictions were imposed “for no other purpose than to eliminate the nigger from politics”. Nowadays, nefarious motivations require a little more finesse. Good data on felon voting trends is hard to come by, but the obvious demographic skew (blacks are significantly more likely both to vote for Democrats and to have a felony record), combined with the energy in sustaining felony disenfranchisement coming almost exclusively from Republicans, is enough to sustain my suspicions that this is a pretextual exercise.
Regardless of who you think should be allowed to vote, the real focus in this post is the administrative implementation of your preferred policy. I argue that voting eligibility is progressively more difficult to determine (and thus less defensible) the more restrictions on voting there are. If we examine the gradient of policies within the US, we can broadly categorize degrees of felony disenfranchisement as follows:
No restrictions
Restoration upon release from incarceration
Restoration upon completion of supervision (including parole or probation)
All of the above plus more
Jurisdictions that allow jailhouse ballots have it easiest: Their election authorities don’t have to do anything about checking for a criminal record.
The next step up on the restriction gradient is still fairly easy to implement. There is the occasional fuck-up, but governments generally can tell you who exactly is currently in their carceral resorts. Election authorities having to consult one of the widely available searchable databases is a hurdle, but a relatively trivial one.
The next step further up — restoration upon completion of supervision — is where the difficulty really starts to ramp up. Unlike inmate rosters updated on the daily, when exactly someone’s supervision ends is information that will be buried within reams of figurative dossiers in filing cabinets scattered across the state. There’s nominally a system in place, such as the National Voting Rights Act, which allows different parts of the country to keep everyone up to date about voting registration. But I’ve written about how judicial record systems have to straddle an unenviable position: simultaneously maintaining an iron grip on legacy compatibility (imagine the nightmare of a computer upgrade wiping out entire convictions) while cracking the door just widely enough to allow cross-pollination with other systems.
Consider the situation in detail. Let’s say that I, your favorite public defender, am able to track down a judgment & sentence order from the 1990s and find that my client was sentenced to X months in prison and Y months of supervision after release. I can’t just plug that into a date calculator. First, I would need to know if this was the only charge they served time under, including, potentially, an extradition hold for a warrant from another jurisdiction. Then I’d need to track down whether any early release for good behavior applied to their charge, including noting any legislative changes that may have occurred and been retroactively applied. Even if I have a definitive release date, the length of supervised release is far less static. Maybe there was a court order that ended it early, or maybe there was a change in the law for that specific offense, or maybe their supervision time was tolled or extended for whatever reason by the probation authority. And so on. Despite what I do for a living, I have absolutely no confidence that I am able to accurately calculate the precise end of someone’s supervision, and this is why I always leave that task to the math wizards at the Department of Corrections. I hope and pray to Allah they get it right, because there’s no fucking way I’ll know otherwise.
Take a place like Texas, which has four hundred and seventy two district court jurisdictions, each potentially with a different computer system that Billie the COBOL guy cobbled together back in 1997 and that nobody wants to touch again. It seems inevitable that mistakes will be made.
Though rare, a mistake is likely what happened to Hervis Rogers, a convicted felon who nevertheless managed to register to vote and participate in two elections while still on supervised release. There’s no dispute that the election authorities who approved Rogers’ registration fucked up. But for Rogers to be guilty of a crime, he had to have known he wasn’t eligible to register (proving he knew he was on supervised release is not enough; Texas must prove he knew he was ineligible to vote) and also have known or anticipated that the authorities would fuck up. Maybe the prosecutors have evidence to establish Rogers’ Machiavellian sophistry conclusively, but until then, this mens rea theory is patently absurd.
The intent element must also be evaluated in context of how confusing parsing eligibility can be. Pamela Moses’ case in Tennessee illustrates how much of a bog this is even for experienced legal professionals. Moses was previously convicted of an evidence tampering felony, and in 2019, she tried to run for mayor. Election officials told her she was not eligible because she had not yet finished her probation. A court echoed what those officials said, but her probation officer later signed off on a certificate of restoration that Moses submitted when she registered. Moses was convicted of voter fraud and sentenced to six years in prison before her conviction was overturned on appeal. The probation officer was wrong about her probation term being over, but that wouldn’t have mattered anyway because her predicate conviction — evidence tampering — was one of the few Tennessee offenses that led to permanent disenfranchisement. This was a fact that neither the probation officer, his supervisor, nor the trial judge knew about, as seen from page 24 of the trial transcript (cleaned up):
PROSECUTOR: The tampering with evidence we’re addressing today, which is permanent. I don’t remember all the ones. I know murder, probably rape —
THE COURT: That’s something I didn’t know. Are you telling me if you get convicted of tampering with evidence, you can never vote? Where is that in the law?
DEFENSE: It’s titled– I think it’s 39-15 or 39-17 where it talks about the interference with government operations. Those are —
PROSECUTOR: It’s 40-29-204.
THE COURT: “Those convicted after July 1, 1996, but before July 1, 2006 — those convicted after July 1, 2006, any of the offenses set forth in one and two above, voter fraud, treason, murder in the first degree, aggravated rape.” And then it goes on to say, “Any other violation of title 39 chapter 16 part one, four, and five, designated as a felony” — so are you telling me I’ve got to go back and look at 39-16?
PROSECUTOR: Yes. Now you have to, and that’s where the tampering with evidence, along with — it falls under, like, bribery, contraband, false pretense, the ones that are felonies.
Apparently, it’s impossible to wade through the cobwebs of cross-referencing statutory codes without tripping up somehow, even if wading boots are part of your job uniform. And absent malicious intent, these examples illustrate how easy it is for mistakes to happen. What purpose does punishing these types of mistakes accomplish? Focusing one’s ire toward the people ensnared by the cobwebs doesn’t do anything to get rid of the cobwebs. Getting rid of the cobwebs gets rid of the cobwebs.
The next and final notch on the disenfranchisement dial requires everything mentioned on the list before and more, and the webs just keep getting worse. The exact process varies, with some requiring an additional waiting period and others requiring a court petition (see Table 2). Although it’s the most visible example nowadays, Florida is neither the only nor the first state to require that felons pay off all their legal financial obligations (LFOs) before their right to vote is restored. A handful of other states (it’s unclear exactly how many) have exactly this same system — Arizona, for example. But the recent change in Florida law and the litigation that followed gives us an intimate look into what happens behind the scenes.
Consider the same issues raised above about how much effort it takes to figure out when supervised release ends, but now shatter the mirror and multiply it by the number of shards.
I’ve previously written about the numerous issues inherent in tabulating an individual’s LFOs accurately. The best catalog of the issues is found in this painstakingly detailed 125-page federal court opinion from 2020. Consider this example from page 53 of the opinion:
The case of one named plaintiff, Clifford Tyson, is illustrative. An extraordinarily competent and diligent financial manager in the office of the Hillsborough County Clerk of Court, with the assistance of several long-serving assistants, bulldogged Mr. Tyson’s case for perhaps 12 to 15 hours. The group had combined experience of over 100 years. They came up with what they believed to be the amount owed. But even with all that work, they were unable to explain discrepancies in the records.
So a handful of experienced bureaucrats, working for twelve to fifteen hours on a single case, still cannot arrive at an accurate answer regarding how much this felon owes. This is a breathtaking amount of wasted state capacity. Florida’s estimates on when they expect to complete the avalanche of felon registrations are dire. From page 65 (note that the estimates are likely out of date):
Even without screening for unpaid LFOs, all the Divison’s caseworkers combined can process an average of just 57 registrations per day. The LFO work, standing alone, is likely to take at least as long as—probably much longer than—the review for murder and sexual offenses and for custody or supervision status. Even at 57 registrations per day, screening the 85,000 pending registrations will take 1,491 days. At 261 workdays per year, this is a little over 5 years and 8 months. The projected completion date, even if the Division starts turning out work today, and even if screening for LFOs doesn’t take longer than screening for murders, sexual offenses, custody, and supervision, is early in 2026. With a flood of additional registrations expected in this presidential election year, the anticipated completion date might well be pushed into the 2030s.
It’s reasonable to assign blame to growing pains. It takes time for a state apparatus to get into gear and find its groove, and so it’s possible that Florida eventually will start to put a sizable dent in this backlog. After all, other states have been checking LFOs for years already.
But no matter what, the amount of resources devoted to checking whether Individual X paid off Fine Y will never be zero. It will always take time and resources to run the apparatus. Time and resources that could be devoted to other tasks. Having a bunch of Poindexters focused on meticulously tabulating every dollar a convicted felon might owe means they’re not available for other things, like maybe cracking down on actual fraud. It’s an opportunity cost no matter how you look at it.
I’ve already made my position on felony disenfranchisement clear: I don’t think there should be any. If you believe otherwise, that’s fine, but the argument in favor needs to take into account the additional resources such a regime necessarily eats up. You need higher state capacity to check people’s convictions, calculate the terms of their sentence, and tabulate their LFOs, and an entire additional apparatus to investigate and prosecute scofflaws. Any argument in favor of disenfranchising felons has to explain why these additional costs are worthwhile.
The cost here doesn’t just come in the form of bureaucrat man-hours. Recall that Pamela Moses was sentenced to six years in prison for voting under circumstances that confused even her probation officers. Setting aside the personal cost of that sentence, the average cost of housing an inmate in Tennessee is about $28,000 a year, so Moses’ cost to the state would have been $168,000 total had her appeal not been granted. Is that worth the cost? You tell me.
I'm still waiting for the selfish reasons... were your selfish reasons just that it takes more administrative effort?