When Texas Says No | Local Insights

01 · System Trend

More fights, same odds.

Annual open-records ruling volume rose sharply through 2025. The full-release rate stayed near five percent.

Annual volume vs. full-release rate

Bars: logical rulings. Line: full-release rate. Outcomes are resolved for every year shown; 2026 is year-to-date through Aug. 4.

The annual data table replaces the chart on small screens.

Logical rulings Full-release rate 2026 year to date

Read the annual chart data as a table

Year	Logical rulings	Full release	Any withholding
2015	26,786	4.5%	93.4%
2016	28,717	4.0%	93.9%
2017	29,238	4.2%	94.0%
2018	32,071	4.4%	93.7%
2019	36,592	4.2%	94.3%
2020	32,288	4.7%	93.8%
2021	36,802	4.2%	94.4%
2022	39,926	4.9%	93.9%
2023	42,899	4.5%	94.4%
2024	42,597	4.8%	94.1%
2025	45,357	5.2%	93.7%
2026 YTD	32,748	5.7%	92.9%

More people entered the system. Their odds of getting everything stayed almost flat.Full release remained roughly 4–6% each year, including 2024–2026 YTD.

02 · Agency Lookup

Find the agency.

Search 519 machine-normalized governmental-body profiles from the corrected corpus. Name variants may remain.

Agency

03 · Agency Outcome Comparisons

How agency outcomes compare.

The public comparison is strongest when it stays descriptive: who gets full release least often, who gets it most often, and how peers differ.

Large agencies with the lowest and highest full-release rates

Agencies with at least 500 classified rulings in the corrected 2015–Aug. 4, 2026 corpus. Malformed extraction labels and duplicate name variants are excluded.

Lowest full-release rates

Highest full-release rates

Agency types at both ends

Share of rulings in which OAG allowed the agency to withhold at least some information.

Some withholding most common

Law enforcement

96.7%

Municipalities

94.9%

Counties

94.0%

Some withholding least common

Universities / colleges

83.9%

ISDs / school districts

85.2%

State agencies

89.6%

04 · Resident Need

What are you trying to get?

Within the topic classifier, outcome patterns vary sharply by the kind of record described in a ruling.

Record type

05 · Common Provisions

Provisions associated with withholding.

When a provision appears, how often did OAG authorize some withholding?

These are corpus-level associations, not causal effectiveness scores or a continuous trend across the 2023/2024 measurement seam. The missed-deadline finding under §552.302 is presented separately in Section 6 because it is a procedural, pro-release provision rather than a substantive exception. §552.305 is a notice procedure, not a substantive exception.

06 · Human Stakes

When open records gets personal.

These are the human consequences hidden inside the rules. Each example is drawn from a reviewed ruling or a verified corpus finding.

A power of attorney became the reason not to answer

OR2022-11362 · Texas Title Insurance Guarantee Association
The request sought one aggregate figure: the total Guarantee Fees collected during a particular timeframe. The requester said he was acting for an incarcerated person under a power of attorney.

OAG concluded that the authorization made him the inmate's agent under §552.028—and ruled that the association did not have to comply with the request.

Why it matters: the ruling never found the dollar total confidential. It ended the inquiry based on his status as the incarcerated person's agent.

Read OR2022-11362 →

They miss the deadline. Some information can still be withheld.

4,465 rulings with §552.302 in the extracted provision field
Texas law creates a presumption that late agencies must release records. In this provision-tagged cohort, 50.2% of rulings still authorized some withholding.

Concrete example: OR2015-00508, concerning a late security-tape submission by the Texas Department of Criminal Justice.

Why it matters: missing the statutory deadline creates a presumption of release, not a guarantee that the public will receive the records.

The “same” request can end three ways

City of Fort Worth incident requests · 276 rulings
Rulings with the same public description and §552.101 produced 66 full withholds, 197 partial releases, and 10 full releases.

Why it matters: the public-facing description does not reveal the underlying facts that produce sharply different outcomes, making similar requests difficult to predict.

Read one full-release ruling →

07 · Inside The Machine

Point estimates differ after observable controls.

This exploratory model measures from a zero baseline: each point is the estimated difference from the expected outcome among comparable rulings. Withholding and release are shown separately because a third outcome—partial release—matters. Uncertainty intervals were not produced, so the estimates do not establish statistically reliable signer differences.

Adjusted full-withhold point estimate

Left of zero: less full withholding than expected. Right of zero: more. Profiles shown are the range endpoints plus every signer used in the three matched comparisons at right. Point estimates only; no uncertainty intervals are available.

Adjusted full-release point estimate

This is not the inverse of the first chart. A ruling can be mixed: some information released, some withheld. Point estimates only; no uncertainty intervals are available.

Read the selected signer estimates as a table

Anonymized profile	Full withhold difference	Full release difference
Signer A	−4.79 pp	+14.24 pp
Signer B	+12.64 pp	−2.19 pp
Signer C	+10.92 pp	Not shown
Signer D	+6.68 pp	−1.45 pp
Signer E	−12.43 pp	−1.85 pp
Signer F	−6.44 pp	−17.94 pp

The point estimates span a wide range after comparable-case controls.Exploratory, noncausal model · 39 eligible signer profiles · no uncertainty intervals.

The assignment black box

The public cannot see how OAG assigns, drafts, reviews, or supervises these rulings. The observed pattern could reflect discretion, specialization, nonrandom assignment, templates, or final review.

Matched-pair checks:
Signer C vs Signer E: +18.90 pp full withhold.
Signer D vs Signer E: +10.51 pp full withhold.
Signer A vs Signer F: +28.61 pp full release.

The public-interest demand is not an attorney ranking. It is disclosure of the system that produces these differences.

08 · Method & Data

What this page measures.

A public analysis needs its limits to be as visible as its findings.

Scope

426,021 logical OAG open-records rulings, 2015 through Aug. 4, 2026. “Full release” means the ruling was classified as requiring release of all requested information; it is not the success rate for all Texas public-information requests.

Sources & downloads

Limitations & response

Exception fields change measurement method at the 2023/2024 boundary and are not used as a continuous trend. The §552.302 figure is a provision-tagged cohort across that seam, not a statewide missed-deadline rate. Entity labels can have name variants. Signer results are exploratory associations, not causal or statistically established findings; public labels are anonymized, and uncertainty intervals were not produced. OAG response status: Local Insights did not request an agency response before publication. This analysis uses the OAG's public ruling database and links readers to the underlying records; corrections and clarifications are invited.