A six-step pathway · one of several~6 min · with the lens console

A filing in November becomes a non-vote the following May.

Six lawful administrative steps create a pathway from housing instability to lost registration. The pathway exists. How often it ends in actual disenfranchisement — versus voters who would not have voted regardless — is the harder question, and one this piece marks honestly.

Trace the funnel
ACT 01

The household, registered to vote.

Households tracked
100,000
Registered voters
~180K
Stage
baseline
i.
The baselineAct 01

Start with one hundred thousand renting households.

Pick any large Georgia metro. Roughly 40% of households rent. Most are registered to vote at their current address, the address on the lease, on the utility bill, on the driver's license. Civic standing is anchored to housing.

Housing is the variable. Civic standing is the dependent.

ii.
The filingAct 02

A landlord files in magistrate court.

Georgia's eviction process begins with a dispossessory filing. In Fulton County alone, filings have averaged over 60,000 per year through 2024–25. Many filings result in negotiated outcomes; a portion proceed to writ of possession.

The household receives a summons. The clock starts. The county recorder logs the filing. The civic record is, for now, unchanged.

SourceFulton Magistrate Court2024–25 docket
iii.
The moveAct 03

The household relocates. Often urgently.

Some move to another rental in the same county. Some move across the county line to take advantage of lower rent. Some leave the metro entirely. A meaningful fraction enter unstable housing — couch-surfing, motels, vehicles. The address on the voter file is now stale.

Update-your-registration is a one-form task. It is also one of the most commonly deferred administrative tasks in the modern American renter experience.

iv.
The mailerAct 04

The county registrar sends a confirmation card. Strong

Voter rolls are maintained in part through mailed verification. A confirmation card is sent to addresses flagged for change. If the card comes back undeliverable, the voter is moved to "inactive" status. If the voter doesn't respond and doesn't vote in two consecutive federal elections, they are removed entirely.

The mechanism is lawful, the timeline is statutory, the consequence is silent. Most affected voters do not know it has happened.

v.
The electionAct 05

Election day arrives. The voter shows up. Or doesn't.

If they show up at their old polling place, they are offered a provisional ballot, which may or may not be counted depending on the county and the timing of their move. If they go to the new polling place, they are not on the rolls. If they don't know either polling place is now wrong, they don't go at all.

The pathway from "registered voter who recently moved" to "counted vote" is narrow. The structural failure rate is high.

vi.
The signatureAct 06

By the next purge cycle, the registration is gone. Plausible

List maintenance happens on a statutory cycle. The voter, now invisible in election data, will not appear on the next county turnout report, the next demographic crosstab, the next political-targeting model. They become structurally non-participating — not by choice, but by administration.

The eviction filed in November becomes a registration gap by August. By the following primary, the voter is administratively a non-voter. The chain holds across cycles.

SourceGA SoS purge reportsmulti-cycle

Where the funnel runs fastest.

The mechanism is statewide. The throughput varies by county — driven by eviction-filing volume, rental tenure rates, registrar staffing, and the share of voters in high-mobility ZIP codes.

01 · Fulton

The volume county.

Highest absolute eviction filings in the state. Highest absolute count of voters moved to inactive status post-filing.

~60K filings/yr
02 · DeKalb

The cross-county displacement.

Adjacent to Fulton; significant net inflow of displaced renters who haven't yet re-registered.

net inflow
03 · Cobb / Gwinnett

Suburban absorption.

Receiving counties for outer-metro displacement. Registration lag often runs multiple years.

delayed re-reg
04 · Henry / Clayton

The black-middle-class receiving zone.

Significant inflow from Fulton/DeKalb. Registration update rates lower than the receiving population's voting age.

structural gap
05 · Macon-Bibb

The mid-size pattern.

Same funnel, smaller absolute numbers, similar rate. The mechanism doesn't require metro scale.

replicates
06 · Rural south GA

The under-measured zone.

Lower filing volume but lower data quality, the funnel here is observed less because it's recorded less, not because it doesn't happen.

data gap
A constellation of competing reads · lens console

One fact about the registration gap. Four lenses. Four pictures.

Pick a lens. The room reconfigures. Same fact, different argument, different chart. Press A/B/C/D.

The shared fact
"In high-eviction Fulton precincts, voter registrations dropped at roughly 1.4× the rate of low-eviction precincts in the same county over 2022–2026."
SOURCE · GA SoS list maintenance · Fulton magistrate filings
A · Information design
Income vs cost · the crossing point
household balance · the crossing point IS the mechanism
B · Information design
Attrition through the gates
attrition through gates · who survives and who doesn't
C · Information design
Story salience over 24 months
24-month story salience · the absence is the argument
D · Information design
Persistence by birth cohort
generational decay · each cohort steepens
A · ECONOMIC LENS
Material instability
"The system removes voters when their economic ground gives way."
The registration loss is downstream of the housing market. Rising rents, stagnant wages, and a tight inventory produce eviction filings, which displace households, which break the address-on-file pipeline.
eviction filingdisplacementaddress-stalepurge
householdmortgagedisplacementcost burdenmarket failure
— The distinguishing test —
Do magistrate eviction filings correlate with inactivation rates at the precinct level above the rate predicted by overall mobility?
What this lens makes visible
household balance sheet
income vs cost curve
Key shape in the data
crossing point at month 8
eviction probability inflects
Vocabulary signature
market · cost · burden
economic register
What this lens underplays
statutory mechanism
the procedural fix

What this piece marks rather than concludes.

The chain is documented at the population level. The individual causal link — this specific eviction caused this specific non-vote — requires record-linkage work that is doable, not done.

Documented · solid

Filing volumes

Magistrate court filings are public; Eviction Lab and academic researchers maintain longitudinal datasets for major Georgia counties.

Documented · solid

List maintenance procedure

The NVRA list-maintenance process and its Georgia implementation are statutory and public.

Linkage required

Filing → purge match

Linking specific evictions to specific inactivations requires matched-record work across magistrate and registrar data. Sample studies suggest the linkage rate is meaningful; full enumeration is FOIA-able.

Available · pending

Per-county throughput

Each county's funnel rate (filings → inactivations → purges) can be computed with current public data. We will publish a 159-county table in the companion deep dive.

Permanent caveat

Voter intent

The system removes the registration. Whether the voter would have voted, and for whom, is unknowable. The civic harm is removal, not specific electoral counterfactual.

Out of scope here

Reform options

Same-day registration, address-on-file flexibility, and other reform proposals are documentable but outside this descriptive piece. The reform case starts with the descriptive picture.

The strongest counter-read · the funnel framing click to expand

A reasonable critic — particularly an election administrator — would argue: each of the six steps exists for a defensible administrative reason. Voter-roll maintenance is required by NVRA. Address verification protects against impersonation. The "funnel" framing implies intent where there is procedure.

The critic is partly right. None of the six steps were designed to produce non-voters. But the cumulative effect is measurable, and procedure-without-intent still produces disparate impact under standard civil-rights tests. The episode's argument is structural, not motivational.

If you accept the steps as individually lawful (we do), the question becomes: is the cumulative outcome acceptable? That is a policy question this episode does not pretend to settle.

Press E for evidence mode
Evidence mode · on
Documented Modeled Speculative