All posts by Justin Levitt

DOJ increases pressure to vacuum up voter files

Department of Justice, election administration, voter registration

A month ago, I noted that the DOJ’s Civil Rights Division had sent demands to a bunch of states for copies of the voter files.  These letters were pretty unusual.  I didn’t think DOJ’s ostensible rationale for demanding the files made much sense.  I didn’t think the statutes DOJ cited entitled them to the information they were demanding.  Indeed, I thought federal law precluded DOJ from amassing this information (at least until they do some more homework).

Now the Civil Rights Division is out with a new letter (as before, it’s not clear how many states were recipients, but it seems to be widespread, and perhaps national).  The new letter is from the head of the Civil Rights Division, Assistant AG Harmeet Dhillon (previous iterations were from the acting chief of the Voting Section).  It cites some different statutes.  But it’s not ultimately more persuasive on any of the points above.

It sure looks like the Civil Rights Division is trying to do what the Pence-Kobach Commission failed to do eight years ago: assemble a national voter file.  (Why?  I’m not sure.  And that’s part of the problem.  I’ve got guesses, but federal law says I don’t have to guess.  Before the federal government does something like this, it has to explain itself.)  I thought that was likely illegal then, and I think it’s still likely illegal now.

What follows takes a closer look at these new letters. (It’s a longer analysis, so I’ve tucked the good stuff away under the headers below.)

The ostensible rationale for the files (the NVRA/HAVA).

The NVRA and HAVA each require states to conduct general programs of list maintenance that make reasonable efforts to clean the rolls.  (Here’s the NVRA version, here’s the HAVA version.)  There are more specific provisions in each statute protecting against inaccurate overcleaning: list maintenance comes with a thumb on the scales for not tossing potentially eligible voters if you’re not sure.  But the main point is that both statutes put list maintenance firmly in the hands of the states, and in that endeavor, both require reasonable maintenance procedures, not record-by-record assurance that every ineligible voter is removed.

The new letters offer a very different vision.  “In charging the Attorney General with enforcement of the voter registration list requirements in HAVA and in the NVRA, Congress plainly intended that DOJ be able to conduct an independent review of each state’s list.”  Not just a review of a state’s general program of list maintenance – a review of each state’s list.  Each state’s list.  That staggering assertion of federal intervention comes with no citation to any specific statutory provision, and is not supported by any that I’m aware of (nor any historical DOJ practice, nor any judicial precedent).  But the Civil Rights Division thinks it so important that it copied the sentence twice, verbatim, in a 2½-page letter.  Maybe they think repeating it makes it seem truthier?

If the Civil Rights Division were focused instead on the actual language of the statute, they’d recognize that the responsibility they actually have is reviewing a state’s general program of list maintenance.  That means that it’s entirely legit to ask questions about state procedures for cleaning the rolls, and even to ask for statistics that might touch on those procedures’ efficacy.  (While federal law puts some constraints on collecting personal voter information – see below – it’s entirely agnostic about DOJ questions on procedures and stats.)  But data on particular individuals isn’t particularly helpful in enforcing the NVRA/HAVA provisions the letters cite.  Knowing whether “Justin Levitt” is on the voter rolls in a particular state, and knowing Levitt’s Social Security digits, isn’t especially responsive to whether that state has or doesn’t have the general program required by statute.

The new statutory citation (the Civil Rights Act).

Earlier versions of the gimme-all-your-data letter focused on the NVRA’s provision for public access to the voter rolls, which has been repeatedly construed to allow states to keep sensitive info (like Social Security digits) private even while requiring public disclosure of a lot of other material.  The new letter makes clear that the Civil Rights Division wants all fields – which, again, the letter says twice, because then you know it’s serious.  And the new letter cites a different source of authority: the Civil Rights Act of 1960. But that statute doesn’t do what DOJ wants it to do either.

The Civil Rights Act of 1960 has a provision that requires election officials to keep election-related records for 22 months.  And it has a provision that authorizes the Attorney General to ask to inspect, reproduce, or copy any such record (which is cited in the letter), with a written note explaining the basis and the purpose for the demand (which is not cited in the letter).

There’s admittedly not a ton of caselaw on this provision, and what little there is was developed around Southern jurisdictions at a time when records involved less personal data and more blatant discrimination.  But I’ve always thought of the retention provision like a litigation hold: election officials, don’t throw anything out for 22 months, in case there’s a problem.  If DOJ has reason to believe there was a problem, it may come asking for specific records later.  The litigation hold is for everything, but any particular DOJ request from that universe isn’t automatically coextensive with “everything”: each request has to have not merely a “purpose,” but also – and separately under the statute, lest it be mere surplusage – a “basis.”  That’s a little like probable cause — it’s not terribly demanding (the articulation in an internal justification memo for opening an investigation in the first instance would normally more than suffice), and properly gives a lot of latitude to the government, but “I’m just curious” isn’t enough. 

And this all makes sense, given the recordkeeping environment of 1960.  If the Civil Rights Division had reason to believe that particular eligible voters had been denied the right to vote, it’s clear that this provision authorized the Division to go look up their registration cards and go look at the pollbooks to see if they checked in.  But it’s just as clear to me that this provision didn’t authorize DOJ to collect copies of all of the registration cards from all of the states — “each state’s list” — to go fishing to see what they could find.  Such a demand wouldn’t have the “basis” the statute requires.

The affirmative restriction on the federal government (the Privacy Act).

Everything above is about why the statutes in the letter don’t authorize the Civil Rights Division to get what it wants.  But there’s a different federal statute that I think affirmatively prevents the DOJ from getting what it wants, and the statutes above have to be read in light of that statute too.

That statute is the Privacy Act of 1974.  It says that before the federal government collects records on individuals, the government has to facilitate a public conversation – a Federal Register notice and notification to congressional committees — about what information it plans to collect, why it needs the information, who has access, and the like.  That’s 5 U.S.C. § 552a(e)(4) and 552a(r).  Federal officials who collect info on Americans without this public notice are committing a federal crime.  State officials who intend to help the Civil Rights Division blow past this notice requirement may be abetting that crime.

The new letters mention a totally different part of the Privacy Act, and otherwise make some noises about privacy, but those are mostly non sequiturs.  I still haven’t seen any indication that the Civil Rights Division has done the homework the Privacy Act requires to collect the voter files.  They’ve provided notices on some information the Division maintains in the course of regular enforcement work: enforcing civil rights means you’ll collect some info about victims and targets and witnesses.  But none of those notices fairly flag that the Division plans to accumulate a national voter file, with the personal information (and First Amendment activity) of Americans who aren’t any of the above.

The Privacy Act isn’t just a process barrier of its own.  It also provides important context for understanding the litigation-hold provisions of the Civil Rights Act of 1960.  Given increasing congressional skepticism of federal government acquisition of Americans’ personal data in 1974, it would be deeply weird (not to mention ahistorical) to read the 1960 statute to override the careful constraints in the Privacy Act, giving the Civil Rights Division the authority to vacuum up information on more than 155 million voters, without  any individualized “basis” and in service of an invented federal power to double-check every state’s list.  Instead, reading the two statutes together helps confirm that the Civil Rights Act authority is as we thought it was: authorizing the AG to get specific information where there’s reason to believe there was a particularized problem in an election within the last 22 months.

Share this:

Texas calls second special session for redistricting

election law and constitutional law, political parties, redistricting

Gov. Abbott called a second special session Friday morning, calling the legislature into action for a second shot at redistricting maps (and, at least in theory, other issues).

The call for the first special session asked the legislature to consider “[l]egislation that provides a revised congressional redistricting plan in light of constitutional concerns raised by the U.S. Department of Justice.”

The call for the second special session just asks the legislature to consider “[l]egislation that provides a congressional redistricting plan.”

Guess they finally realized the DOJ’s pretext didn’t pass the laugh test.

Share this:

The TBD details on California’s redistricting special election

election law and constitutional law, political parties, redistricting, The Voting Wars

Justin here. Today, California Gov. Newsom announced what he’d been foreshadowing for a while now — an effort to seek voters’ approval for new congressional district lines in a Nov. 4 special election. (Rick noted the border patrol “escort” for the press conference. I can’t help note that the Japanese American National Museum – for those who haven’t been, it’s got a tremendously powerful series of exhibits on the WWII internment — is a little on the nose as a backdrop for a militarized show of force at a political event.)

The exact contours of the legislative package to make the redistricting initiative happen probably won’t be crystal clear until the legislature gets back next week. But among the pieces I’ll be watching:

  • Timing. Current law seems to say that the governor can call a special election 148 days out. The legislature, of course, can change that law — and it’ll have to in order to hit a November 4 special election target (82 days away from today). (That timing provision can be changed by statute, I believe – the date change doesn’t have to itself go on the ballot.) I’ll be looking for whether this is a one-time-only change or whether there are more general conditions for the exception.
    (Update: Derek Mueller reminds me that there’s another statute that doesn’t regulate when an election can be scheduled, but does say that the legislature can only put an initiative on the ballot for a scheduled election if that election is at least 131 days away. That statute would also need amending to run this Nov. 4 – and I’ve got the same questions as above.)
  • Funding. Part of the reason for leaving time before a special election is to give election officials the runway they need to run the thing. As Doug Chapin used to hammer home on the regular — fast, accurate, cheap: pick any two. This schedule will be fast, and the results have to be accurate. I’ll be looking to see who’s picking up the extra tab for the pre-election prep overtime.
  • Substance. The best reading of the state constitution is that maps are drawn by an independent commission, once per decade. That authority can be changed with a ballot initiative. But the reporting says that Newsom also plans to “put a new map” in front of voters on Nov. 4. The change to authority has to be in the constitution, but I imagine the map would be an initiated statute (and not itself constitutionalized). There’s nothing inherently weird about having both a constitutional change and a statutory change in the same measure — the proposition to put the commission in place in the first instance combined the two. But a single initiative to both change the process and pass a new specific map has some risks under the state’s single-subject rule. Two separate initiatives create questions about what happens if the electorate passes one but not the other. I’ll be looking to see how the package resolves those questions.
  • Additional constraints. Article XXI is the part of the state constitution that gives authority to the commission. It also has a bunch of other constraints, procedural and substantive. If the new initiative is effectively a temporary contingent carveout, how temporary? What’s the threshold of the contingency? How complete a carveout? Texas has essentially no state rules for drawing congressional districts – the only rules are the few rules in federal law. For its response, does California do the same?
  • Additional triggers. Newsom has said that California will respond to Texas in the maps he puts forward. I’m told that California and Texas aren’t the only two states in the Union. If another state says that it’s re-redrawing its maps to respond to California, does the initiative include provisions for re-re-redrawing the maps to respond to the response? Does the initiative include a provision allowing for later legislative amendment of the (presumably statutory) map?

Lots still TBD here.

Share this:

Election case curriculum videos

election law and constitutional law, Supreme Court, theory

Justin here. With the semester either beginning or around the corner, I wanted to lift up a series of great videos (illustrated, annotated, featuring case text, etc.) that the incomparable Guy-Uriel Charles and Sasha Natapoff have pulled together. Leading scholars (and also me!) each explain and analyze the text of a different seminal Supreme Court case. The videos are all shorter than 15 minutes long, and as the series editors suggest, “can be used to supplement the traditional written casebook and/or to deepen students’ engagement with key cases and doctrines.”

The direct link is https://CRBW.law.harvard.edu, and it’s all free for educators and nonprofits. Just register on the website if you’d like to take a look. (For crim pro faculty, there’s a crim pro version too.)

The roster (so far):

Get ’em while there’s still law to discuss!

Share this:

As redistricting eyes turn again to California

election law and constitutional law, redistricting,

Justin here. I find much of the work of the UCLA Voting Rights Project to be really valuable — but I have to say, I agree with Rick’s take on their read of the California Constitution‘s provisions on redistricting. The California Constitution can be amended (as Rick says, it takes approval by the voters), but I don’t think current law permits the legislature to just draw maps on their own. And I think following the Project’s memo would end up putting the legislature in a worst-case bind.

Rick says that the memo uses a sort of “wooden textualism” — the same sort of methodology, giving overly short shrift to what drafters clearly intended, that has produced the 8th Circuit’s dead-wrong decisions on private rights to enforce the VRA. But I think it’s even less persuasive than that. The memo describes portions of the California Constitution that “retain for [the] Legislature the power to adopt ‘a statute establishing or changing boundaries of any legislative, congressional, or other election district.’” That’s not what I read those sections to be doing.

The memo cites two sections of Article IV of the California Constitution – section 8(c) and section 10(b) – as giving the legislature power to draw redistricting statutes (and override the state’s independent commission) whenever it wants. Article IV, section 8, subsection (c) is about effective dates for legislation. It has a default, a special exception (I think) for bills passed at the end of the first year of a two-year session, and an exception to the exception for redistricting statutes. Article IV, section 10, subsection (b) is about bills becoming laws without the governor’s signature if he sits on them long enough. It too has a default, and an exception for redistricting statutes.

Neither of those sections purports to assign the legislature a role in drafting redistricting statutes. It just says what the effective dates and law-without-signature timing might be for legislative redistricting statutes … if such statutes existed. And the best read of the constitution, I think, is that Article XXI just means there are none of those legislative redistricting statutes now (because the redistricting commission has the power to do that work instead). That doesn’t make these procedural bits superfluous: if the commission provisions are withdrawn or modified, they kick back in.

It’s very hard to read those small procedural exceptions in 8(c) and 10(b) to imply a giant substantive power when the much much much clearer provision on substantive power gives that power instead to the independent commission. Also, it sure seems weird for Article XXI of the California Constitution to go to all the trouble to specifically take the pen away from the legislature and prescribe a bunch of criteria for the commission to use, if a different part of the constitution just lets the legislature undo that work however they want whenever they want.

I am, of course, not a court. And maybe a court would disagree with me. But I think it’s far more likely that a court would read these provisions of Article IV as vestigial procedural caveats rather than affirmative authorization. And if that’s true, then a legislature acting on this theory (and not, say, putting a measure before the public if they really want to effectuate a retaliatory gerrymander) is putting itself in a worst-case scenario.

If the legislature draws a radically gerrymandered map on this theory, it ticks off all of the reform voters who put the commission in place in the first instance, and anyone who doesn’t love the idea of a Democratic gerrymander. And if a court (as I think most likely) then strikes the map down for lack of legislative authority, it ticks off all of the Democrats gunning for pure partisanship – because now the legislature has accomplished nothing, and it’s too late for a special election that would actually change the rules before 2026. I think this is a recipe to claiming action while actually affecting nothing, and ticking _everyone_ off in the process.

Share this:

The recent rash of DOJ voter file “requests”

Department of Justice, voter registration

Recent reports have disclosed that DOJ has now “asked” at least nine states for copies of their voter rolls, and I know that there are at least a handful of counties that have separately also received requests/demands for the county’s voter file.  I’m not at all sure the requests are lawful.  And what’s more, depending on why the requests are coming (the explanation the DOJ is giving doesn’t make much sense), federal law may not only prohibit DOJ from asking, but may also create liability for states and counties that respond.

Some voter registration data (including parts of the voter file) are public (and required to be public by federal law).  Some information on the voter file is more sensitive data (like Social Security digits, driver’s license numbers, and signatures) that states have protected from public release, and in every case I’m aware of (this decision collects several of the others), courts have ratified states’ ability to keep that information protected.  (In a truly odd case, DOJ has sued over this sensitive PII in Orange County, and if I were betting, I’d bet on them losing.)

But here’s the thing: even data that’s available to the public may not be legally available to the federal government.  In the Privacy Act of 1974, based on some then-recent disclosure of unsavory federal efforts to amass personal data, Congress regulated federal access to data records of identifiable individual Americans (statistical compilations are treated differently).  Congress was particularly sensitive to the treatment of data on individuals revealing activity protected by the First Amendment (like, say, voting).

For the most part, Congress didn’t prohibit federal agencies from collecting this data outright.  But it did set up some procedural protections.  The law says that if a federal agency is going to collect data — even data otherwise available to the public — we first have to  be able to have a public conversation about what’s being collected for what purposes (and how it’s stored and secured and transmitted and accessed, etc.)   The agency has to provide a notice in the Federal Register with an opportunity for comment, and has to notify Senate HSGAC and House Oversight committees, for any new set of data records, or any new use of existing records.  The rules are serious: criminal penalties are attached.

And what’s more: if the federal agency plans to use the data records in a matching program that affects an individual’s benefits, there are legal repercussions not only for the federal agency, but for any other governmental entity sending them data.  Those sorts of programs have to be set up with careful written agreements, available to Congress and to the public … and state and local providers of data are on the hook if they send information they know is going into a program like this, without dotting “i”s and crossing “t”s first.

I flagged these potential Privacy Act issues in DHS’s apparent expansion of a system designed for tracking immigrants, to expand to contain searchable data on U.S.-born citizens.  And there are similar (and similarly disturbing) issues if DOJ’s just accumulating voter files.  DOJ keeps its Privacy Act disclosures online, here.  (At least, I’m not aware of another repository.)  It’s the Civil Rights Division that’s asking. And so I’ve looked through the Privacy Act notices submitted by the Civil Rights Division, and the only one that’s even plausibly anywhere near close is the authorization to keep general case files.  That notice says it’s for “case files, matters, memoranda, correspondence, studies, and reports relating to enforcement of civil rights laws and other various duties of the Civil Rights Division” – if you’re investigating a civil rights violation, you’re going to have personal information about the victims and targets and witnesses.  Fair enough.  But I don’t think anyone would look at that notice and expect that DOJ plans to collect and keep a bunch of voter files, with identifiable information about individuals’ registration and/or voting history who aren’t themselves plausibly part of any investigation into state practices.

So I don’t think DOJ has jumped through the necessary hoops to collect what it’s requesting, even if it were telling the truth about why.  But I’ve also got questions about the why.

Collecting the voter files is also of extremely limited utility in investigating what DOJ says it’s investigating in these requests.  The requests cite provisions of the National Voter Registration Act and the Help America Vote Act that require officials to have general programs of voter list maintenance.  To assess whether officials are satisfying these statutory mandates, you’d want to know what procedures they have in place, and maybe some statistics about overall list maintenance activity.  But the individual records on the list aren’t really relevant to that inquiry: whether “Justin Levitt” happens to be on any state or county’s voter list has very, very little to do with whether the state or county has the general program it’s required to have under federal law. 

I’m not claiming that temporary access to particular voter registration data will never be necessary for any federal investigation.  But if the voter rolls aren’t really necessary for the asserted purpose here, that raises questions about why the DOJ actually wants the lists – not only whether it’s not telling the truth (or the whole truth) in its demand letters, but also whether it has other purposes in mind, like matching databases to each other in a way that might impact benefits (with the mistakes that follow from bad matching).  That’d have distinct protections under the Privacy Act, with distinct consequences for the state and local officials asked to supply data.  There have been reports about the degree to which the DOGE approach is to move fast and break things — which may be fine for the private sector, but not consistent with federal statutory protections for public activity.  I guess I’m wondering whether the demands for voter rolls are more of the same.

Share this:

Mamdani’s electoral coalition in the primary (a revision)

primaries, voter registration,

Justin popping back in. Today, the NYT issued a correction to its story on Mamdani’s impact on the NYC primary electorate, and what I described yesterday as a “staggering” departure from the norm:

A correction was made on June 30, 2025: An earlier version of the chart in this article showing voters by age incorrectly identified the age group with the largest turnout. It was voters aged 30 to 34, not those aged 18 to 24.

Here’s the original chart:

And here’s the update:

To be clear, that turnout by younger voters (both 18-24 and 25-29 year-olds) is still eye-popping, both as a primary-over-primary increase and as an absolute. The 18-24 turnout is just a little less “staggering” than it looked yesterday.

Share this:

“The Trump administration is building a national citizenship data system”

voter registration, ,

NPR’s latest report is truly a blockbuster.  The SAVE system run by DHS has always been a system keeping track of the immigration status of individuals at some point in the immigration system (including after people naturalize).  As the National Conference of State Legislatures said as recently as May 12, “USCIS is clear that SAVE does not provide information on citizens born in the US [or on individuals outside of the immigration system].”  But DHS and Social Security have now apparently expanded SAVE to effectively serve as a database of natural-born US citizens as well – without the notice to the public or the notice to Congress required by the Privacy Act of 1974.

We’ve never had a federal government database for identifying all of the citizens in the country.  (And even if DHS and SSA have added natural-born citizens, we still don’t: not every citizen is in the Social Security database.)  Part of that is based on concerns about logistics and data quality and data security.  Part of that is based on vigorous public resistance.  Part of that is based on the law: because of the other two categories of issues, Congress has never authorized the executive to create one, and has put legal limits in place to prevent the government just up and creating one without telling us.

And, it appears, the Trump Administration has just been building one on its own, with zero transparency.  According to the article, the primary public indication appears to be a June 13 fact sheet posted deep in the bread crumbs on the USCIS SAVE website, that says “SAVE can verify U.S.-born citizens for voter verification agencies.”  Zero details there.  DHS has also apparently briefed Cleta Mitchell’s Election Integrity Network on the system.  For the rest of us, bupkis.  I don’t understand how any of that is in line with federal statute, which requires advance notice to Congressional committees and to the public of any significant change in the scope or use of federal databases on individuals, with criminal penalties for noncompliance.

To be clear, a reliable database could be really useful for some purposes, IF DONE IN THE RIGHT WAY.  (Just as one example, it would facilitate cutting many of the bureaucratic hurdles to accessing federal benefits reserved for citizens.  Including removing any claimed need for specific documentation of citizenship that citizens may not have readily available.)  But the fact that there might be some beneficial use cases doesn’t resolve any of the questions about logistics and data quality and data security and public support and legal authorization.  NPR reports that DOGE was involved in this Administration’s decision to just plow ahead and given DOGE’s other extremely well-publicized challenges with both accuracy and security, those questions loom quite a bit larger in my mind.

Share this:

“Meet the Faces of Democracy”

election administration, election law biz, election subversion risk

Issue One has a new interview with former Georgia State Election Board member Edward Lindsey.   I raise it up mostly to highlight the exceptional set of interviews they’re amassing.  As they put it: “More than 10,000 officials across the country run U.S. elections. This interview is part of a series highlighting the election heroes who are the faces of democracy.”

Share this:

“The first rule in Trump’s Washington: Don’t write anything down”

Uncategorized,

WaPo’s subhead: “A new culture of secrecy in government is taking root – among career staffers and new political appointees alike.”

Across President Donald Trump’s administration, a creeping culture of secrecy is overtaking personnel and budget decisions, casual social interactions, and everything in between, according to interviews with more than 40 employees across two dozen agencies, most of whom spoke on the condition of anonymity to avoid reprisals. No one wants to put anything in writing anymore, federal workers said: Meetings are conducted in-person behind closed doors, even on anodyne topics. Workers prefer to talk outdoors, as long as the weather cooperates. And communication among colleagues — whether work-related or personal — has increasingly shifted to the encrypted messaging app Signal, with messages set to auto-delete.

BTW, while there’s nothing unlawful about holding a conversation, holding that conversation on a messaging app without preserving the conversation will violate federal public records laws in many circumstances.

Share this:

“Voting Rights Litigation: A Blueprint for Strategic Investment in America’s Polarized Political Landscape”

The Voting Wars

I’m can’t-look-away-from-the-train-wreck fascinated by the guidance on how investors can profit off of civil rights struggles:

The Supreme Court’s 2013 Shelby County v. Holder decision, which gutted the Voting Rights Act’s (VRA) preclearance protections, has unleashed a wave of litigation across the U.S. While the ruling aimed to reduce federal oversight, it instead ignited a firestorm of legal battles over voting access and redistricting. For investors, this isn’t just a civil rights story—it’s a roadmap to sectors primed for growth in racially polarized states. Let’s dissect the opportunities and risks.

Share this:

“How Successful Were We in Running the 2024 Elections?”

election administration, Election Assistance Commission

John Fortier is moderating an AEI panel next Tuesday, July 1, keyed to Monday’s release of the EAC’s latest Election Administration and Voting Survey data, which includes a huge amount of valuable information about how we did what we did in 2024.  Panelists include the EAC’s Chair Don Palmer and Vice Chair Tom Hicks, with MIT’s Charles Stewart rounding out an all-star cast.

Share this: