Georgia Prisoners’ Speak

Supporting paper, September revision  |  Prepared for the House Blue-Ribbon Study Committee on Criminal Post-Conviction Litigation  |  September 2026

← All committee materials  •  Download the package (PDF)

For every right there shall be a remedy

Supporting paper: Four years, thirty minutes: Georgia’s habeas deadline, and the means the State provides to meet it

Prepared for the House Blue-Ribbon Study Committee on Criminal Post-Conviction Litigation
Georgia Prisoners’ Speak · The GDC Accountability Project, Inc. · September 2026 revision of the paper submitted in August 2026

What changed since August: the counsel figures and the deadline figures now come from the completed fifty-state survey (counsel coded under a written rule; the deadline pass finished); the count of decisions citing § 9-14-42(c)(4) is stated at eight, the reproducible subsection-citation figure; one passage on counsel for persons under sentence of death is added at Part III; and the prose has been tidied. No authority or quotation has changed.

This paper carries the authorities behind the consolidated summary of the same title. Every
authority cited here has been retrieved and read in full. Emphasis within quotations is added
unless otherwise noted.

Our position

We ask this Committee to recommend repeal of the four-year limitation in O.C.G.A. § 9-14-42(c).

That subsection provides that a habeas action “shall be filed… within four years in the case of a
felony… from” one of four alternative dates, the first being “[t]he judgment of conviction
becoming final by the conclusion of direct review or the expiration of the time for seeking such
review.” It was enacted in 2004. Ga. L. 2004, p. 917. Georgia’s habeas statute carried no
filing deadline for the 141 years from the Code of 1863 until then.

The statute names three other starting dates. The period may instead
run from the removal of a state-created impediment; from the recognition of a new retroactive right;
or from “[t]he date on which the facts supporting the claims presented could have been discovered
through the exercise of due diligence.” § 9-14-42(c)(2)–(4). The statute does not lack a discovery rule. What it lacks is any record of that rule doing
work. Part V sets out that record: eight decisions citing it in twenty-two years.

The calendar and the opportunity are different things, and this paper is about the distance
between them.

Every limitation period rests on a premise: that the person subject to it had a fair chance to act
before it closed. Georgia has removed the only device by which that premise can be tested, and it
has done so for a population it holds in prison, denies counsel, and permits thirty minutes a week
at a law-library terminal.

I. What a limitation period assumes

We take the case for the deadline at its strongest. The four-year period is not arbitrary. It
serves finality, the principle that at some point a criminal judgment should be settled, so that
victims are not indefinitely unsettled, prosecutors are not indefinitely on call, and the State’s
resources go to present cases rather than old ones. That is a real interest and we do not dispute
it.

But finality rests on something. A limitation period does not merely close a door at a fixed
time. It presumes that the person subject to it had a fair opportunity to act before it closed.
That premise is what distinguishes a statute of limitations from an arbitrary cutoff, and it is why
nearly every limitation period in American law carries some device for testing whether the premise
held: tolling for disability, for concealment, for extraordinary circumstance.

Georgia has removed the device.

In Stubbs v. Hall, 308 Ga. 354 (2020), the Supreme Court of Georgia held the four-year period
not subject to statutory or equitable tolling, because granting relief would require the
Court to “create — for the first time in Georgia law — an equitable remedy that allows habeas
petitioners to circumvent the statute of limitations provision enacted by the General Assembly.” It
found a “complete dearth of authority” for doing so. That holding, in Division 5, was unanimous.

So the premise is never examined. The period runs whether or not the person had any opportunity
at all. Not for a lockdown, not for a closed law library, not for counsel’s error, not for a person
who could not have known. The question a limitation period exists to ask, whether this person had a
fair chance, is one no Georgia court is permitted to reach.

That is the defect. The rest of this paper is about whether the premise actually holds.

I-A. The conditions under which the premise is tested

No lawyer. There is no right to counsel in a Georgia habeas proceeding. That is settled law
and needs no witness: the Sixth Amendment right does not extend to state collateral review
(Pennsylvania v. Finley, 481 U.S. 551 (1987)), and Georgia has created no statutory right in its
place. It is the reason the Georgia Appellate Practice and Educational Resource Center was
established in 1988.

Thirty minutes a week. Under GDC Standard Operating Procedure 227.03, “Access to Courts,”
effective 30 June 2020:

“Each requesting offender shall receive at least 30 minutes of access to the electronic law
library.”

“The Librarian will determine the amount of time that each offender receives based on the number
of computers and the number of offenders requesting access.”

“No offender will be allowed to utilize the printed legal collection while using the electronic
law library.“

The actual allotment is expressly a function of scarcity the Department itself sets: computers
installed, divided by people asking. Access requires a written request to the Librarian and
placement on a call-out.

Note what the rule does not say. It fixes a duration for the electronic law library and
states no frequency. The Department therefore sets both variables, the length of the session
expressly, and the interval between sessions by saying nothing at all. In reported practice the
allotment is thirty minutes per week.

What else the same SOP provides, and why it does not answer the point

The same SOP says more than that, and the rest of it reads more favorably. It also provides that “[o]ffenders may request two (2) hours of
reference library time per week“; that a person “subject to a court-determined or statutory
deadline may request four (4) hours of additional library time per week up to thirty days prior
to the deadline”; and that prison reference libraries shall be open “a minimum of 20 hours per
week.” Read together and at face value, those figures describe someone with several hours a week
in which to work.

Three things separate the figures from the opportunity.

They are not the same library. The thirty minutes is the electronic law library, the
searchable collection in which case law is actually found. The two hours is the reference
library, a print collection. The SOP forbids using them in one visit: “[n]o offender will be
allowed to utilize the printed legal collection while using the electronic law library.” A person
trying to identify a constitutional defect in his own trial needs the terminal, and the terminal is
the thirty minutes.

The deadline supplement arrives too late to reach this problem. The four additional hours require
both a “court-determined or statutory deadline” and a request falling “up to thirty days prior to”
it. Whatever else that provision reaches, it does nothing for a person spending four years trying to
discover whether he has a claim at all: the extra time becomes available, if ever, in the final month
and only to someone who already knows the deadline is there.

The twenty hours belongs to the library, not to the person. It is an opening-hours requirement on
the facility. It says nothing about what any individual receives.

And “may request” is not “receives.” Access depends on a movement call-out
requiring staff to escort. Chronic understaffing and facility lockdowns routinely cancel it. A
person may go weeks without reaching a terminal, through no act or omission of his own, while the
clock runs without interruption.

The Department of Justice’s findings report of 1 October 2024 does not address law libraries, and
we do not cite it for that. We cite it for the condition that makes the policy unperformable. DOJ
found correctional-officer vacancy rates of roughly 50% systemwide, exceeding 70% at ten of the
State’s largest facilities. A library visit in a Georgia prison requires an officer to run the
call-out and escort the movement. A policy that entitles a man to library time does not deliver it
when there is no one to take him there, and GDC’s own rule subordinates library access to the
facility’s schedule: access “shall not conflict with mandatory facility schedules such as work,
meals, and sick calls.”

I-B. What the State is asking a person to accomplish under those conditions

To bring a habeas petition a person must recognize that something in his own trial was
constitutionally defective. That requires understanding criminal procedure, the rules of evidence,
the standards governing effective assistance of counsel, and the procedural rules determining where
and how such a claim may be brought at all.

Georgia licenses no one to do that work with less than seven years of full-time study: four
years of undergraduate education and three of law school, followed by a bar examination, and then
by the years of practice that make an issue recognizable on sight. That is what the State requires of anyone it licenses to do this
work.

And the four years is not for learning the law. It runs concurrently with everything else.
Inside the same period he must obtain his own trial transcript, identify what went wrong in it,
discover facts that by definition were not apparent at trial, locate and correspond with witnesses
he cannot visit, draft a petition that satisfies the pleading requirements, and file it in the right
court.

What the same task takes elsewhere. Persons exonerated in the United States in 2024 lost an
average of 13.5 years to wrongful imprisonment. (National Registry of Exonerations, 2024 Annual
Report, 2 April 2025.) Those cases typically involved lawyers, investigators, innocence
organizations, journalists, and laboratory testing unavailable to anyone acting alone.

With every professional advantage, correcting a wrongful conviction in this country has taken an
average of more than thirteen years. Georgia allows four, to a person doing it alone, thirty minutes
at a time, and permits no court to ask whether that was enough.

II. Neither of the obvious substitutes reaches this problem

Two cheaper answers than repeal deserve consideration first, and neither reaches the problem. We set them out
before making our case, because if either did work we would be asking for the wrong thing.

“Add a tolling provision and keep the deadline”

A tolling provision excuses delay for a reason a court can name: the petitioner was prevented,
misled, disabled, or the State concealed something. It is built for the person who knew he had a
claim and could not act on it.

It does nothing for the person who did not know. He has no impediment to plead and no
concealment to prove. He was simply a layman in a cell who had not yet learned that what happened at
his trial had a name. There is nothing to toll. Tolling reaches the man who was stopped at the
door; the problem here is the man who never learned the door was there.

“Move ineffectiveness to habeas and provide counsel”

This is the Chief Justice’s proposal and we support it. But it answers a smaller question than the
one before this Committee, for two reasons.

Habeas corpus is larger than ineffective assistance. It is the only proceeding that carries any
constitutional claim after appeal: Mitchum v. State, 306 Ga. 878 (2019), holds that post-appeal
constitutional claims “could be pursued only through habeas corpus.” Suppressed evidence, a coerced
confession, a juror who concealed a relationship, counsel laboring under an undisclosed conflict:
a right to counsel for ineffectiveness claims reaches none of them. Those claims keep the
four-year deadline and keep no lawyer.

And a person cannot ask for a lawyer until he knows what to ask about. Any right to counsel this
Committee creates will attach to a claim already identified: someone must first recognize that
something in his own trial was unconstitutional, and name it well enough to seek help. That step
happens alone, in the conditions §I-A describes, and it is the step the deadline consumes.

The claims that matter most surface last. Suppressed evidence emerges when a file is finally
produced, a laboratory scandal when an audit is published, a juror’s concealed relationship when
someone talks. None of these arrive on a four-year schedule, and none of them can be hurried by
appointing counsel to a claim nobody has identified yet.

What that leaves

Repeal is the only one of the three that costs the State nothing. A right to counsel requires an
appropriation and an administering body. A tolling provision requires courts to litigate whether each
petitioner’s excuse qualifies, a new question in every case. Removing a limitation period requires
neither. It creates no proceeding, appoints no one, and funds nothing.

A deadline that admits no exception is not a deadline about diligence. It is a deadline about the
calendar. And it is worth being precise about what Stubbs decided: not that the petitioner
before it had been dilatory, but that no Georgia court has authority to consider the question at
all.

II-A. Georgia’s courts route these claims to habeas because habeas was untimed

There is a second population the deadline closes out, and the reason it closes them out is a
premise Georgia’s case law still relies on.

O.C.G.A. § 17-9-4, from the Code of 1863 and never amended, provides that a judgment “void for any
other cause[] is a mere nullity and may be so held in any court when it becomes material to the
interest of the parties to consider it.” The right is not in doubt. What has been litigated for
more than a century is where to assert it.

The answer Georgia’s courts give traces to a single 1906 decision, and it is a routing
instruction. McDonald v. State, 126 Ga. 536 (1906), in its entirety on this point:

“A motion to set aside the judgment is not the appropriate remedy in a criminal case if the
indictment is void. The judgment may be arrested upon motion made during the term at which the
verdict is rendered, or the prisoner may be discharged upon a writ of habeas corpus at any time
thereafter, if no question as to the validity of the indictment was adjudicated at the trial.”

The reason matters as much as the rule. McDonald does not hold that a void conviction goes
unremedied. It holds that a set-aside motion is not the appropriate remedy, because two other doors are open,
and one of them, habeas corpus, is open “at any time thereafter.”

In 2004 the General Assembly closed that door after four years. Ga. L. 2004, p. 917.

In 2009 the Supreme Court of Georgia reinstated the 1906 rule. Harper v. State, 286 Ga. 216,
described the contrary approach as “an improvident departure from more than a century of precedent”
and confirmed that the remedies for a void conviction are the three statutory procedures: an
extraordinary motion for new trial, a motion in arrest of judgment, or habeas corpus.

The rule survived. The premise that justified it did not. The other two doors close faster than
habeas: a motion in arrest of judgment “must be made during the term at which the judgment was
obtained” (§ 17-9-61(b)), months, sometimes days, and an extraordinary motion for new trial cannot
carry a constitutional claim, because the 1967 Act narrowed it “to exclude such claims.” Habeas was
the door that stayed open without limit. It is the reason the other two could be closed.

Stated fairly: Harper resolved a real conflict, and we do not suggest it was wrongly decided.
Nor do we suggest the Court overlooked the limitation. The point is narrower and it is one only
this body can act on: the precedent Georgia relies on to route void-conviction claims to habeas was
built on an assumption the General Assembly had already withdrawn five years earlier, and a court
cannot restore it.

This is why repeal reaches further than ineffective assistance. A void conviction is not a trial
error. It is a judgment the Code calls a nullity. Georgia’s courts send that claim to habeas on the
strength of a case that assumed habeas had no deadline.

II-B. What the writ reached before 1967, and what the 1967 Act traded

McDonald‘s routing assumed an untimed writ. What the writ could hear was another matter, and
the modern shape of Georgia habeas comes from a single trade the General Assembly made in 1967.

Before 1967, the writ was narrow. In Mitchum v. State, 306 Ga. 878 (2019), the Supreme Court
of Georgia adopted Professor Donald E. Wilkes, Jr.’s account of that era: post-conviction habeas
relief, first granted in Georgia in 1893, “had been available in Georgia only on grounds of lack
of jurisdiction. Unless the judgment of conviction or the sentence was void for want of subject
matter or personal jurisdiction, the writ would be denied. While denial of counsel could render a
conviction void, a strict doctrine of waiver of constitutional rights prevented the deprivation of
any other right from constituting grounds for relief.”

The Habeas Corpus Act of 1967 did two things in one motion. It widened the writ to
constitutional claims, § 9-14-42(a) reaches “a substantial denial of his rights under the
Constitution of the United States or of this state.” And it closed the alternatives: the
extraordinary motion for new trial, which courts had construed to allow constitutional claims,
“became narrower, by enactment of the Habeas Corpus Act of 1967, to exclude such claims”
(Mitchum), and § 9-14-41 made Article 2 “the exclusive procedure for seeking a writ of habeas
corpus for persons whose liberty is being restrained by virtue of a sentence imposed against them by
a state court of record.”

The General Assembly wrote down why. Section 9-14-40 still carries the 1967 findings: Georgia
convictions were being attacked in federal court on issues Georgia courts had never considered; this
“tends to weaken state courts”; and the answer was that “the scope of state habeas corpus be
expanded” and the waiver doctrine modified.

Note what the trade produced. After 1967, every post-appeal constitutional claim has one door.
After 2004, that door has a clock. And the class of claim the writ had always reached, the void
judgment, the want of jurisdiction, is the class the clock now extinguishes. The oldest and
narrowest core of the writ is the part with nowhere left to go.

III. Who the deadline falls on

From the State’s own caseload reporting: Judicial Council of Georgia / Administrative Office of the
Courts, superior court case type Habeas Corpus, calendar years 2021–2025. (The figures below were
extracted from the Council’s public superior-court dashboard and were confirmed in writing by the
Council’s Office of Research and Data Analysis on 7 August 2026. The Office advised that updates to
this data are expected shortly; if they change these statistics appreciably, we will provide the
Committee an updated table on receipt.)

• Georgia superior courts received 4,425 habeas petitions and disposed of 3,955, roughly
  nine hundred a year.
• In 2,309 of those 3,955 cases (58.4%) the petitioner had no lawyer. The statewide
  self-represented rate across all case types is 27.7%. Habeas runs at 2.11 times that rate.
• The respondent is the warden; under O.C.G.A. § 9-14-45 every petition from a person in Department
  of Corrections custody is served on the Attorney General.

The ordinary Georgia habeas case is a person in prison, without a lawyer, without an
investigator, without the ability to interview a witness or obtain a transcript on his own, holding
a claim that requires proving what his trial lawyer failed to do, against the State of Georgia,
represented by the Attorney General, with the resources of the State behind it, inside a period
that never yields.

That is not an adversarial proceeding in any ordinary sense. One party has the machinery of
state government. The other has thirty minutes at a terminal, if he asks in writing and his name
is placed on a call-out.

And no lawyer stands opposite. §I-A sets out the absence of any right to counsel. The gap is old
enough to have produced an institution: the Georgia Appellate Practice and Educational Resource
Center was established in 1988 to provide free representation to people under a sentence of death,
because capital habeas could not be left to the unrepresented. The principle, counsel where a
claim cannot honestly be litigated alone, has been accepted before, for the cases thought least
tolerable to get wrong. It has been accepted in money and not in law. Georgia appropriates
$900,000 a year to the Center, and it is the only state of fifty with no mechanism for appointing
counsel to a person under sentence of death: no right, no appointing authority, no compensation
provision. Twenty-four states require the appointment, three permit it, and twenty-one have no
death penalty. (GPS fifty-state survey, counsel in capital post-conviction, September 2026; the
appropriation is Judicial Branch § 6.7 of the FY2027 General Appropriations Act, HB 974.)

The deadline is already before this Committee

At the 10 July hearing, a member of this Committee raised the four-year period on his own
initiative. Rep. Tyler Paul Smith observed that Georgia’s four-year period runs alongside a
one-year federal limitation: “it seems to me there may be an issue on the horizon here that we’re
incentivizing state [habeas filings], but a lot of defendants, especially indigent, they’ll never
be able to take advantage of the federal habeas because they have passed the statute.” He asked
the witness then before the Committee, Kristin Verrill, Executive Director of the Georgia
Innocence Project, “do you think that’s the interplay here as part of the discussion we should
have?”

She answered: “I think it’s something we should try to address, yeah.“

What Rep. Smith had in mind, the exchange does not say. He identified the interaction of the two
limitation periods as an issue and asked whether it belonged in the Committee’s discussion; the
witness agreed it did. He did not say the four-year period should be repealed, and we do not claim
this exchange as support for our ask. We state the federal point only as he stated it, and make no
claim of our own about how the two clocks interact.

We cite the exchange for one fact: the four-year period reached this Committee on 10 July, raised
by one of its own members without prompting from us. And the defendant he described, indigent,
past the federal statute while Georgia’s was still running, is a person running out of courts.
Repeal does not touch the federal period. What it does is keep Georgia’s own courthouse open.

IV. What the last decade of published outcomes shows

Georgia published habeas outcomes once and stopped. The Judicial Council’s Annual Report: Georgia
Courts carried the Supreme Court’s dispositions of habeas certificate-of-probable-cause
applications through the FY2013 volume, reporting data through calendar 2012. The table appears
in no later volume we have been able to read; we examined FY2014 through FY2019, and the FY2017 volume
exists only as a page-image file and is the one we could not verify.

And the decade it did publish was a decade of decline. From the Council’s own tables, the grant
rate ran at 10.0% in 2007 and 2.0% in 2012, and in that final published year the Court
granted 9 of 449 applications while dismissals more than doubled, from 68 to 171.

(The commonly cited aggregate for FY2003–2012, 263 granted of 3,942, a 6.7% grant rate, comes
from the petitioner’s brief in Wilson v. Sellers. The figure is not printed in the Council’s own
volumes; our reconstruction from the primary reports cross-verifies six of the ten years at 6.4%.
We cite the per-year figures above, which we have taken directly from the reports.)

Nor can outcomes be recovered from published decisions. Appeal from a habeas denial requires a
certificate of probable cause from the Supreme Court (O.C.G.A. § 9-14-52), and in the ordinary case
it is refused by unpublished order. In Redmon v. Johnson, 302 Ga. 763 (2018), the Court described
its own practice: it “denies summarily the applications of another 20 habeas petitioners today,
as we have denied thousands of such applications in the past (while granting a few each year,
including two today).”

It is tempting to call this a discretionary gate. Redmon says it is not: the label “means only that a full appeal is
not provided as of right in every habeas case. If the Court determines … that a habeas case has
arguable merit, then the Court has no discretion: the application must be granted.” Our point is
not that the Court refuses meritorious appeals. It is that the refusals are unpublished, so the
disposition of the ordinary habeas case leaves no public record, which is why the outcome data
cannot be rebuilt from decisions.

And the requirement runs one way. § 9-14-52 provides that “[i]f the trial court finds in favor
of the petitioner, no certificate of probable cause need be obtained by the respondent as a
condition precedent to appeal.” The
State may appeal a grant as of right. The petitioner must ask permission to appeal a denial.

And the absence is not a publication gap. The data is not gathered. Asked directly, the Judicial
Council’s Office of Research and Data Analysis confirmed in writing on 7 August 2026: “We do collect
data regarding Manner of Disposition, but it is not reported by all courts for all years. We do not
collect any data regarding case outcomes.” Whatever this Committee recommends, it will recommend
into a system whose results no one measures: not withheld, not aggregated differently, not
collected.

V. The statutory safety valve opens, and what it takes to open it

Georgia’s habeas statute contains a provision that softens the deadline. It is real, it works, and
we want to show you precisely what working looks like, because that is the argument.

The late-discovery trigger. § 9-14-42(c)(4) starts the clock from “the date on which the facts
supporting the claims presented could have been discovered through the exercise of due diligence.”

It has saved petitions, including very late ones. In Watkins v. Ballinger, 308 Ga. 387 (2020),
a conviction became final in 2003 and the petition at issue was filed in 2017, fourteen years
later, alleging that a juror had conducted her own out-of-court investigation and that the State
had concealed exculpatory evidence. A unanimous Court, Justice Peterson among those joining, held the
allegations “sufficient to satisfy the requirements of OCGA §§ 9-14-42 (c) (4) and 9-14-51,” and
reversed the dismissal. In 2022 the Court affirmed the grant of habeas relief on the
juror-misconduct claim. Ballinger v. Watkins, 315 Ga. 369 (2022). And in Shelton v. Lee, 299
Ga. 350 (2016), the habeas court held a petition filed eight years after finality timely under
(c)(4), a finding the State did not challenge on appeal, where
appellate counsel had abandoned the case.

So the answer to “the safety valve covers hard cases” is not that it never opens. It is what
opening costs. Twenty-two years of decisions contain eight citing that subsection, two of them the same
litigation at two stages. (CourtListener, courts ga and gactapp, run 2026-08-05; a broader
search on the provision’s operative phrase adds two more; parameters supplied on request.) Mr. Watkins needed a lawyer, an innocence organization, two trips
to the Supreme Court of Georgia, and twenty-one years between his conviction and the affirmance of
his relief. That is the machinery a man with thirty minutes at a library terminal is being told is
available to him.

The strongest answer to this section: in Mitchum v. State, 306 Ga. 878 (2019), the Court
pointed to (c)(4) as evidence that the four-year limitation “does not make habeas corpus any less of
an adequate remedy.” That is a fair point, and Watkins supports it. Our
response is not that the provision is dead. It is that a provision cited in eight decisions across twenty-two
years, by the best-resourced petitioners in the system, is not a general answer for the population
the deadline actually closes out, and Part V-B identifies who that is.

One distinction matters for drafting. In Stubbs, the Court cautioned that “[w]e have not always
been precise in the way we have described the four events discussed in OCGA § 9-14-42 (c) that
could cause a petitioner’s one- or four-year statute-of-limitations clock to start running.”
(c)(4) is a trigger date, not a tolling rule. It governs when the clock starts; it does not
reopen a case once four years have run against a claim the petitioner knew about all along.

The miscarriage-of-justice provision, and why it is not a safety valve for this deadline.
§ 9-14-48(d) provides in mandatory terms: “In all cases habeas corpus relief shall be granted
to avoid a miscarriage of justice.”

It does open. In Turpin v. Hill, 269 Ga. 302 (1998), the Court allowed a
belated claim through it over a partial dissent that objected in exactly those terms.

But it is not addressed to lateness. That sentence sits within § 9-14-48(d), the
procedural-default subsection. It excuses a petitioner who failed to raise a claim at trial or
on appeal; it says nothing about one who filed his petition after the limitation period ran. No
Georgia court has held that it reaches a petition dismissed as late; the question is never reached,
because the case ends at the limitation.

And where it does apply, the standard is severe. Valenzuela v. Newsome, 253 Ga. 793 (1985), holds
the term “is by no means to be deemed synonymous with procedural irregularity, or even with
reversible error“; it “demands a much greater substance, approaching perhaps the imprisonment
of one who, not only is not guilty of the specific offense for which he is convicted, but, further,
is not even culpable in the circumstances under inquiry.

V-A. Georgia has already decided that finality must sometimes yield

The four-year bar contains an exception, and the General Assembly wrote it. § 9-14-42(c) applies

“within four years in the case of a felony, other than one challenging a conviction for which a
death sentence has been imposed or challenging a sentence of death…”

Capital cases are exempt from the deadline entirely. A person under sentence of death may bring
a habeas petition however many years have passed.

So the principle is not in dispute. Georgia already accepts that in some class of case the
interest in correcting a wrongful conviction outweighs the interest in repose, and that no period of
years is short enough to justify enforcing a judgment that should not stand. The Committee is not
being asked to accept a new proposition. It is being asked where the line falls.

And the line is currently drawn at execution. A man serving forty years for a crime he did not
commit has four years. A man sentenced to die has no limit. Whatever is said in defense of that
distinction should at least be stated expressly.

V-B. What the deadline forecloses, and what it does not

We begin with the strongest answer to this section, because it is a real one.

Georgia does provide a route for claims resting on scientific change: the extraordinary motion for
new trial, O.C.G.A. § 5-5-41, which has no limitation period for newly discovered evidence. In
Smith v. State, No. S25A0548 (Ga. 15 Oct. 2025), the Supreme Court of Georgia addressed a father convicted in 2003
of his infant son’s death on a Shaken Baby Syndrome diagnosis, who argued in 2021 that the medical
consensus had “dramatically evolved.” On remand, on 21 July 2026, Judge Kimberly A. Gallant of the
Superior Court of Gwinnett County ordered a new trial. (Elizabeth Weill-Greenberg, “After More Than
20 Years in Prison, Georgia Man Wins New Trial in ‘Shaken Baby’ Case,” The Appeal, 4 August 2026;
see also Atlanta Journal-Constitution, July 2026.) The four-year bar did not foreclose him, and
that is worth stating plainly.

But look at what that route cost, and what it does not cover.

Twenty-three years. The Supreme Court of Georgia observed that one expert’s opinion “could not
have been offered at the time of trial, let alone in the 1990s”, which is to say the evidence that
won him a new trial did not exist when his four years were running. He has not been acquitted and
he has not been released; he has been given back the trial he should have had.

Two things about how he got there.

He could not use habeas. By 2021 that door had been closed to him for roughly nine years. His
only available vehicle was the extraordinary motion, a remedy Georgia courts describe as
disfavored and hedge with six separate requirements.

And he was represented. Eight expert witnesses, six days of testimony, two appeals to the
Supreme Court of Georgia and two remands is not a case a person assembles alone. The comparison
this Committee should hold in mind is between what that took and what § I describes: no lawyer,
thirty minutes a week, four years.

That is the point, and it cuts our way. Georgia already recognizes that some claims cannot
fairly be time-limited, and has provided an untimed vehicle for them. The question for this
Committee is why that recognition stops where it does.

What the extraordinary motion cannot carry

Constitutional claims. In Mitchum v. State, 306 Ga. 878 (2019), a unanimous Court held that
post-appeal constitutional claims “could be pursued only through habeas corpus.” So the untimed
route is available for newly discovered evidence and closed to constitutional violation, and habeas, which carries constitutional claims, closes after four years.

A person who discovers, in year six, that the State suppressed exculpatory evidence, or that a
juror was tainted, or that counsel labored under an undisclosed conflict, holds a claim the
extraordinary motion will not take, and a habeas door that depends entirely on the discovery
triggers.

We note the provisions that may answer part of this. § 9-14-42(c)(2) starts the clock from the
removal of “an impediment to filing… created by state action in violation of the Constitution or
laws of the United States or of this state”,
which may reach some suppression claims, and is the one category with a genuine statutory answer.
Whether it permits a petition filed years after the four-year period is not settled, and we do
not assert otherwise. § 9-14-42(c)(4) starts it from when facts “could have been discovered through
the exercise of due diligence”, but the Supreme Court cautioned in Stubbs that this is a trigger
date and not a tolling rule, and in twenty-two years it has been cited in eight
decisions.

The gap, stated precisely

The bar is not aimed at claims that arrive late. It is aimed at people who cannot act in time.

Section 9-14-42(c)(4) asks a single question: could these facts have been discovered through
diligence? If they could not, the petitioner gets a later start; that is Watkins. But the
statute asks nothing at all about whether he could have done anything with facts he already had.

That is the gap, and it is the whole of it. A man who knew from the day of sentencing that his
lawyer never investigated his alibi has no discovery problem. His facts were available to him
immediately. What he lacked was a transcript, a law library, a lawyer, and the ability to draft a
petition that states a constitutional claim, and none of that starts his clock later, because none
of it is discovery. He is not late because the evidence was hidden. He is late because the State
gave him four years and no means of using them, and the exception the State points to is not
written to see him.

V-C. The deadline is redundant

Georgia protects the finality of criminal judgments by several means, all of which survive repeal:

• Procedural default (§ 9-14-48(d)): absent cause and actual prejudice, relief “shall not be
  granted” for claims not properly raised at trial and on appeal.
• Res judicata and the bar on successive petitions (§ 9-14-51).
• The certificate of probable cause (§ 9-14-52): no appeal from a denial without the Supreme
  Court’s permission, which Redmon describes as refused in “thousands” of cases.
• The petitioner’s burden of proof, unchanged.

Georgia had all of these and no deadline for the whole of its history before 2004. Repeal would
not open a door; it would remove one lock from a door with four others on it.

We looked for the reason the 2004 limitation was enacted and could not find one. No bill
caption, purpose statement or committee note we could retrieve explains what problem it was meant to
solve. We do not assert what the General Assembly intended, and we would welcome being shown it.
But the burden of showing what the deadline accomplishes should not fall on the people it
excludes.

VI. Habeas reform is already before this Committee: the Chief Justice put it there

On 3 March 2026, Chief Justice Peterson wrote separately in Sanders v. State, joined by six
other Justices, “to point out the harm that rule has caused our criminal justice system, and to
encourage the General Assembly to change it.” The rule is Georgia’s requirement that ineffectiveness
be raised at the earliest opportunity, which forces substitution of counsel and, under In re
Formal Advisory Opinion 10-1, 293 Ga. 397 (2013), counsel from outside the circuit public
defender’s office.

But the concurrence does not stop at ineffective assistance. It identifies what the Court cannot
fix and the General Assembly must, and two of the three items are habeas:

“…sending ineffectiveness claims to habeas is an improvement only if it is paired with creating
a limited statutory right to counsel in habeas for certain claims. Taking the thousands of
ineffectiveness claims litigated each year in motions for new trial across all 51 judicial
circuits and sending them all to habeas proceedings overseen by the relatively small handful of
circuits that house Georgia prisons would be a serious problem for judicial workloads. And a
shift of this sort would likely require budget shifts.

All of those things — creating a limited right to counsel for some habeas proceedings,
adjusting habeas venue, and budget adjustments — are the sort of thing that far exceeds the
judicial power.“

So the habeas system is already on this Committee’s agenda, placed there by the Chief Justice, with three legislative items named: counsel, venue, and budget.

Repeal of the
four-year limitation is not among his three items. We do not claim his support for it, and nothing
in this section should be read that way. What his concurrence establishes is that the habeas
article is properly before you and that the Court cannot repair it: the four-year period is our
ask, made on the evidence in Parts I through V, and it should be judged on that evidence.

On venue: the Chief Justice named the problem and the solution

He warned that sending “thousands of ineffectiveness claims litigated each year… across all 51
judicial circuits” into “the relatively small handful of circuits that house Georgia prisons” would
be “a serious problem for judicial workloads.” Georgia superior courts currently receive about
nine hundred habeas petitions a year in total, so decoupling would multiply that several times
over.

Our analysis of the State’s own caseload data measures the concentration he described: 91.7% of
habeas filings are in counties holding a GDC facility, and five rural counties receive roughly a
quarter of every petition arising anywhere in Georgia.

The solution is his own. On 10 July he told this Committee there “would be some benefit to
adjusting habeas venue so that that first habeas petition in which ineffectiveness is litigated
happens in the court of conviction rather than in just the superior court for whatever circuit the
prison is in.”

His reason was the record, and the workload was his second point, not his first. Moving venue,
he said, would put the case “closer to where the lawyers and the witnesses and everybody else other
than the incarcerated defendant are”, and “in addition to” that, “it also spreads the work
across the state as opposed to concentrating all of those now much more involved habeas petitions in
the handful of circuits that have prisons.” He also suggested the change could let the district
attorney defend the petition, by making the DA a defendant in his official capacity.

Thousands of claims divided among five circuits is a crisis. The same claims across fifty-one is
ordinary business. The workload objection to opening habeas is therefore an argument about where
these cases are heard, and the answer to it is already before this Committee, in the Chief
Justice’s own venue recommendation. We support that change. We do not ask you for it, and repeal
does not depend on it.

What other states do

From the GPS fifty-state survey. Each of the routing and counsel cells below was read
individually and carries a verified quotation from the governing statute, rule or decision; the
deadline findings come from a separate pass built for that question, described beneath.

• Twenty-nine states route ineffectiveness claims to collateral review: nineteen send them
  there exclusively, and ten more treat collateral review as the preferred forum. Georgia’s
  proposed destination is the mainstream on either count.
• Of those twenty-nine, eight make counsel mandatory and twelve more require it once a stated
  condition is met. The conditional twelve need a caveat: several condition
  appointment on the court first ordering a hearing. That is appointment after the threshold has
  been crossed, which is not the same as counsel to help cross it, and it is the narrower reading
  of what the Chief Justice said decoupling requires.
• Eleven states impose no post-conviction limitations period at all, and thirty-five more set a
  period a court may excuse.
• Only four states run a deadline that cannot be excused for any reason: Georgia, South Dakota,
  Virginia and Washington. Of those four, Georgia and Virginia also provide no counsel.

A note on how the deadline figures were produced, because the comparison is easy to get
wrong. A single number cannot be compared across states without naming the instrument it
belongs to: Georgia’s four years limits the writ of habeas corpus itself, while most states’
figures limit a post-conviction act. We ran the deadline question as a separate pass for that
reason, naming the operative instrument before the number and comparing only like with like. An
earlier keyword count put the no-limitations states at sixteen; reading each statute individually
reduced it to eleven, and eleven is the figure we stand behind.

VII. What repeal would do, and would not do

Would do. Remove a limitation period enacted in 2004, returning Georgia to the position it
occupied for its entire prior history and aligning the state remedy with a federal clock that is
already running.

Would not do:

• Create a proceeding. Habeas corpus exists. Repeal changes when it may be invoked, nothing else.
• Remove any other requirement. The cause-and-prejudice bar of § 9-14-48(d) remains. The
  certificate-of-probable-cause requirement of § 9-14-52 remains. Res judicata and successive-petition
  rules remain. Repeal removes a clock, not a screen.
• Free anyone. Habeas relief where granted ordinarily returns the case for further proceedings.
• Cost anything. No office, no program, no appointment obligation. (A right to counsel for a
  first habeas petition, the Chief Justice’s stated necessity, is a separate question that does
  carry cost. We support it and can supply costed options on request.)

VIII. Questions we expect

“Four years is already longer than the federal one year.” The two clocks are not the same kind
of clock. The federal year is tolled while a properly filed state petition is pending, and may be equitably
tolled for extraordinary circumstances (Holland). Georgia’s four years is tolled for nothing
(Stubbs). And because the federal clock runs while the state clock is merely available, using
Georgia’s period is what destroys the federal one, the interplay a member of this Committee
identified on 10 July.

“This destroys finality.” Georgia’s habeas statute carried no filing deadline from the Code of
1863 until 2004, and no one suggests those 141 years produced disorder. Every other limit on the writ
survives repeal: procedural default, the successive-petition bar, and the merits themselves.

“Floodgates.” Repeal enlarges the group that may file. That is its purpose, and a paper claiming otherwise would be arguing against its own case.
What we can say is the scale. Superior courts receive roughly nine hundred habeas petitions a
year now. Georgia already runs the experiment: capital petitions are exempt from this deadline
entirely, and no one describes that docket as a flood. Every new petition still has to clear
procedural default, § 9-14-51, and the merits; repeal opens the courthouse door, not the cell.

“The safety valves already cover hard cases.” They open. We set out above exactly how
often. The late-discovery trigger appears in eight decisions in twenty-two years, and it saved
Watkins after fourteen, with a lawyer, an innocence organization, and two appearances before the
Supreme Court. The miscarriage-of-justice provision has granted relief (Turpin v. Hill, 269 Ga. 302
(1998)), but it sits in § 9-14-48(d) and answers procedural default, not lateness. And in Stubbs
the Court found “only one case in which this Court even discusses” equitable tolling, and none
applying it in the habeas context. The valves are real and they are narrow. Our argument is
about who is left outside them.

“If counsel were provided, would this still be a problem?” It would be a smaller one, and we
support providing counsel. But no proposal before this Committee would give counsel automatically
to every person filing a habeas petition: every version on the table is limited to first
petitions, indigent petitioners, and cases surviving to a hearing. For everyone outside whatever
program is created, the deadline and the thirty minutes remain exactly as they are.

“Why not just fix ineffective assistance?” Because decoupling routes those claims into this
deadline. Decoupling determines where the claim is heard. Repeal determines whether it can be
heard at all.

Georgia Prisoners’ Speak is the public voice of The GDC Accountability Project, Inc., a Georgia 501(c)(3) nonprofit. Corrections: accountability@gps.press.