Anomalies and research leads

Quid-pro-quo signature chains

For each (lobbyist L, candidate R, client C, bill B) tuple, the detector surfaces chains where all four of these happen at once: (1) R's campaign paid L; (2) L represents C; (3) C contributed to R; (4) R sponsored or voted-Yea on B (with bonus weight if C testified Support on B). Each individual link is innocent in isolation; the bundled correlation is the corruption signal. Citation: Stigler 1971 ("The Theory of Economic Regulation"); Olson 1965 ("The Logic of Collective Action"); NRS 281A.420 (conflict-of-interest); NRS 218H (lobbyist disclosure); NRS 294A (campaign-finance disclosure).

43 chains surfaced; 4 bound to specific bill action. The bill-bound subset (shown below) is the strongest set of leads — those in the un-bound subset typically pair a PAC payer (no vote record) with a contributing client and merit follow-up against committee-chair voting records as that data lands.

Contribution-and-bill temporal clusters

For every quid-pro-quo chain above, the auditor pulls the contributor's actual contribution dates from cf_contributions and the bill's introduction date from NELIS. Findings surface only where at least one contribution falls within ±90 days of the bill drop. The temporal tightness is what converts an abstract co-occurrence into a documented sequence-of-events. Citation: NRS 281A.420 (conflict-of-interest); NRS 218H (lobbyist disclosure); NRS 294A.220 (campaign-finance disclosure).

1 of 4 QPQ chains have at least one contribution within window. Total of 1 qualifying contributions. Temporal correlation is observed; causation is not asserted. Vote-alignment with donor interest is reported separately (see "client position" / "candidate action" columns) so readers can distinguish patterns that reward the donor from patterns that don't.

Arabella-network grants landing in Nevada

For each Arabella-network 501(c) hub (Sixteen Thirty Fund, New Venture Fund, Hopewell Fund, Tides Foundation/Center/Advocacy), the auditor pulls the most-recent 990 Schedule I from IRS e-file public XML and surfaces every grant whose recipient (a) lists a Nevada state address, OR (b) matches the normalized name of a NV-active entity in our donor / lobbyist-client data. The pooled-fund pattern obscures original-donor identity: original donor → fiscal-sponsor 501(c) → fiscal-sponsored project → spending. Schedule I forces partial visibility. Citation: 26 USC 501(c)(3)/(c)(4); 26 USC 6033 (annual return / public disclosure); GAI / Capital Research Center / NPRI methodology.

76 grants surfaced across 52 unique NV-side recipients. Per-hub breakdown:

• Sixteen Thirty Fund (EIN, 202412): 318 total grants → 4 with NV state address, 7 matching a NV-active name.
• New Venture Fund (EIN, 202412): 854 total grants → 7 with NV state address, 9 matching a NV-active name.
• Hopewell Fund (EIN, 202412): 256 total grants → 2 with NV state address, 4 matching a NV-active name.
• Tides Foundation (EIN, 202412): 2,110 total grants → 21 with NV state address, 14 matching a NV-active name.
• Tides Center (EIN, 202412): 461 total grants → 1 with NV state address, 4 matching a NV-active name.
• Tides Advocacy (EIN, 202412): 192 total grants → 2 with NV state address, 1 matching a NV-active name.

Name-match findings include some intra-network flow (e.g. Sixteen Thirty → Tides Advocacy) — that is itself the documented pattern, not a false positive. Recipients listed below the network include 501(c)(3) and (c)(4) entities active in Nevada civic, voter-engagement, criminal-justice-reform, energy-policy, and worker-organizing space.

Arabella-network transfers to political committees

Form 990 Schedule C Part I-A requires every 501(c)(4)/(c)(5)/(c)(6) to list every Section 527 political organization it transferred funds to (recipient name, EIN, address, dollar amount). The auditor pulls this from the same e-file XML used elsewhere on this page and surfaces transfers whose recipient (a) lists a NV state address, OR (b) name-matches a NV-active PAC in our cf_groups data. The 501(c)(4)→527 pipeline is the legal donor-anonymization route: 501(c)(4) accepts dark-money donations (no donor disclosure), then transfers to a 527 PAC for electoral spending. Citation: 26 USC 501(c); 26 USC 6033; FEC reporting; NRS 294A.

1 NV-side transfer findings. Per-hub national context:

• Sixteen Thirty Fund (202412): 32 PAC transfers totaling $21,424,300 (0 with NV state address; 0 matching a NV-active PAC name).
• New Venture Fund (202412): 0 PAC transfers totaling $0 (0 with NV state address; 0 matching a NV-active PAC name).
• Hopewell Fund (202412): 0 PAC transfers totaling $0 (0 with NV state address; 0 matching a NV-active PAC name).
• Tides Foundation (202412): 0 PAC transfers totaling $0 (0 with NV state address; 0 matching a NV-active PAC name).
• Tides Center (202412): 0 PAC transfers totaling $0 (0 with NV state address; 0 matching a NV-active PAC name).
• Tides Advocacy (202412): 18 PAC transfers totaling $192,758 (1 with NV state address; 0 matching a NV-active PAC name).

People serving as officer/director across multiple tax-exempt orgs

Form 990 Part VII Section A requires every 501(c) organization to publicly disclose its officers, directors, trustees, and key employees with compensation. We extract that data from the IRS e-file XML and surface every individual who appears on the leadership of two or more orgs in our snapshot. This is the federal-layer equivalent of the NVSOS shared-officer detector — Olson 1965 coalition-shell signal at the 501(c) layer. Citation: 26 USC 6033 (annual return / public disclosure).

60 orgs analysed; 23 people serve as officer of 2+ orgs (max 3 per person). The standout patterns are: (1) the AFP / AFP Foundation 501(c)(4)/(c)(3) pair — same board on both, the canonical Koch-network governance pattern; (2) the Tides triangle (Foundation/Center/Advocacy) — same CEO on all three; (3) the Arabella twin (Hopewell + New Venture) — shared General Counsel + shared Board chair.

Voted on a matter affecting a known financial supporter

For each (legislator R, bill B, client C) tuple where R voted on B AND C testified on B AND C contributed $1,000+ to R's campaign, the detector surfaces a research lead. Vote-aligned subset (R voted Yea on a bill C supported, or Nay on one C opposed) is the strongest signal — it suggests the campaign-finance interest may have shaped the legislative vote. NRS 281A.420 requires recusal when a "matter would be affected by the official's commitment in a private capacity to the interests of others." Each finding is a research lead — not all overlaps imply an NRS violation, but each merits review against the legislator's actual disclosure filings.

81 total findings; 34 vote-aligned; 12 where contributor gave more than $10,000. The vote-aligned subset is shown below.

Lobbyists who represent both sides of a regulatory question

Heuristic detector: classify each registered lobbyist client by name into {regulator-side, industry-side, neither}. Lobbyists who serve at least one client in EACH camp are flagged. Regulator-side covers state agencies, boards, commissions, judicial associations, government employee unions, county/city governments. Industry-side covers gaming, mining, energy, telecom, healthcare, alcohol/tobacco/cannabis, construction, banking, insurance, and transportation.

97 lobbyists serve clients in both classes across 88 regulator-side and 113 industry-side clients. The classification is heuristic by display-name pattern; false positives exist (e.g., a "Nevada Energy Association" matches both energy industry and regulator name pattern). Each finding is a research lead, not a published claim. Citation: Stigler 1971, Olson 1965, Tullock 1967.

Lobbyist disclosure gap (NRS 218H reports vs detected money flow)

Compares each registered NV lobbyist's official NRS 218H expenditure disclosures against detected actual money flow: campaign-account payments received plus unambiguous personal political contributions. Of 1,151 registered lobbyists, only 61 reported any expenditures (total $255,302). Detected actual money flow across the 134 lobbyists with material activity totals $5,727,063 — a gap of two orders of magnitude.

NRS 218H disclosure is narrowly scoped to gifts and benefits provided to legislators or staff — it does NOT require disclosure of personal contributions or campaign income. The gap is therefore not a violation in itself; it's a structural transparency signal: who relies on lobbying as primary income without it being visible from the official disclosure record? Citation: NRS 218H, NRS 281A.420, POGO + Government Accountability Institute methodology.

Full list — all 134 findings with depth-audit context →

Same-household pairs giving to the same recipients

Pairs of cf_contributors sharing last name + address but with different first names, who give to ≥3 of the same candidates / PACs. NRS 294A explicitly permits spouses to contribute separately, each up to the per-cycle individual cap. The pattern itself is not a violation. The signal is bundling intent: a household maxing out twice on the same race effectively doubles its influence under the same ceiling. 21 likely-spousal pairs surfaced, totalling $3,210,789 in combined giving; 9 additional pairs filtered out as likely alias / nickname duplicates of the same individual.

Citation: NRS 294A.100 individual contribution limits. Heuristic detector: same address + same surname + different first name (length > 1) + ≥3 shared recipients + ≥$1,000 combined. False positives include adult children living at the parent's address, siblings in the same household, and registration-spelling variants. Each pair is a research lead, not an adjudicated finding.

Donor → recipient pairs dominated by in-kind contributions

System-wide, in-kind contributions (services, polling, mailers, lent staff time) make up only ~3% of NV campaign-finance dollar volume. Pairs where ≥50% of a donor's flow to a single recipient is in-kind are research leads: the in-kind dollar value is self-reported by the giver, with no third-party clearinghouse confirming it. 200 pairs surfaced; $19,135,478 in flagged in-kind dollars.

NRS 294A.0035 defines in-kind contributions and requires a fair-market valuation; NRS 294A.120 mandates disclosure. Self-reported valuation is the inherent vulnerability — a $50K disclosed mailer might cost $10K to actually produce, or might equally cost $200K. Cross-reference with cf_expenditures from the same recipient back to the same donor (or a related entity) to detect bidirectional flow patterns. Self-pair findings (e.g., Eddie Hamilton → Eddie Hamilton) are the same-name false-positive pattern documented under #184.

State vendors who also contribute politically

Cross-reference of Nevada state spending data (checkbook.nv.gov, FY 2019–2025) against (a) registered lobbyist clients with NCC contributions and (b) entity donors of $1K+ to NV candidates. Each row is a legal entity that BOTH receives state-vendor payments AND gives politically — the structural rent-seeking signal under Buchanan/Tullock 1962. 2,648 matches across 154 lobbyist clients and 2,494 entity donors, totalling $58,016,841,945 in state vendor payments.

Each finding is a research lead — corporate-name normalisation has limits, and being both a vendor and a donor is constitutionally protected speech and lawful business. The pattern itself is the signal: who relies most heavily on state contracts while also contributing to the elected officials whose budgets fund those contracts? Citation: NRS 333 / NRS 332 (state purchasing), NRS 294A (campaign-finance disclosure), Buchanan/Tullock 1962.

Registered lobbyists and candidates drawing a state PERS pension

Cross-reference of the Nevada Public Employees' Retirement System (PERS) retiree set (FY 2019–2025) against the registered-lobbyist roster and the cf_candidates table. Surfaces every name that appears in both: a former state employee or legislator now drawing a state pension while simultaneously appearing in lobbyist or candidate records. Citation: NRS 218H (lobbyist disclosure of state employment); Olson 1965 captured-coalition framework. PERS data sourced via socrata-gateway from nvof-open.data.socrata.com.

101 of 1,151 registered lobbyists name-match a PERS pensioner; after filtering out common-name collisions, 76 are low-collision-risk findings (16 high-risk excluded). 2,245 of 8,934 candidates also match. Top lobbyist pension shown below: $276,156/yr. Each is a research lead — verify against authoritative state-employment records before publishing.

Year-over-year budget growth and largest-dollar lines

Analysis of Nevada's OPENBUDGETEXPENSES dataset (one row per agency × budget-account × category × fund × fiscal-year). For lines above $10K we surface every consecutive-year jump of 3× or more — the typical fingerprint of a one-time program injection or a stimulus pass-through. Each one is a research lead, not a published claim of misuse: COVID-era ESSER and ARPA pass-throughs legitimately produced enormous one-year spikes, and large supplemental appropriations also pass through line items.

49,515 rows loaded across FY2019, FY2020, FY2021, FY2022, FY2023, FY2024, FY2025, FY2026. Total expended: $166,223,143,920. Lines flagged: 1,010. Citation: Buchanan/Tullock 1962 ("Calculus of Consent"), Olson 1965, Stigler 1971. Source via socrata-gateway from nvof-open.data.socrata.com.

Politicians and committees that paid registered lobbyists

Cross-reference of cf_expenditures (Nevada Secretary of State campaign expenditures) against the registered-lobbyist roster. Surfaces every payment from a candidate's or committee's campaign account to a registered Nevada lobbyist by tokenwise-normalised name match. 204 registered lobbyists received 3,109 payments totalling $5,122,794.

Why this matters: a politician's campaign committee paying a registered lobbyist directly, while that same lobbyist represents clients with business before the politician's committee or agency, is the structural conflict pattern that NRS 281A.420 was written to prevent. Payments not reported as compensation by the lobbyist would also implicate NRS 218H disclosure violations. Citation: NRS 281A.420 conflict-of-interest; NRS 218H lobbyist-disclosure.

⚠ Same-name flags indicate a candidate's name matches a registered lobbyist's name — often a self-payment by a candidate (campaign paying its own committee chair) and not actually a politician-paying-a-lobbyist relationship. Verify each before treating as a lead.

Isolation Forest top anomalies

An entity scored as anomalous by Isolation Forest has a feature pattern far from the bulk of its peers. Lower (more negative) decision_function scores are more anomalous. Features include client count, contribution totals, federal-overlap, NVSOS officer count, NELIS exhibit count, government/surveillance flags, and graph centrality measures. Algorithm: Liu, Ting, Zhou, "Isolation Forest," ICDM 2008.

Donor-base concentration (Herfindahl-Hirschman Index)

HHI is the U.S. DOJ's antitrust market-concentration metric. HHI > 0.25 = "highly concentrated" market. We apply the same threshold to political donor bases: a candidate or committee with HHI > 0.25 is funded predominantly by one donor cluster. Filtered to entities with $10K+ in receipts. 362 highly-concentrated candidates and 469 highly-concentrated PACs.

Contribution change-points (PELT detection on weekly $-flow)

For each recipient with $50K+ historical receipts and 20+ weeks of activity, weekly contribution flow is analysed for change-points — moments when the mean of the series shifts significantly. Sudden surges or drops in giving warrant investigation: they often correspond to bill activity, candidate filings, scandal-driven donor mobilisation, or coordinated giving campaigns. Citation: Killick, Fearnhead, Eckley, "Optimal Detection of Changepoints With a Linear Computational Cost," J. American Statistical Association 107(500), 2012 (PELT method).

39 recipients with detected change-points · 50 total change-points across all flagged recipients.

Shell-entity formation timing — clustered nonprofits at same agent

Recently-formed (since 2021) politically-named Nevada nonprofits clustered by registered agent. When 3 or more politically-named nonprofits form within 90 days of each other at the same registered agent's address, that's the Olson 1965 coalition-shell pattern via temporal correlation: nominally-independent advocacy organisations created in close succession at a single legal coordinator. Method citation: combines NVSOS formation-date data with the shared-agent detector. Pattern citation: Mancur Olson, The Logic of Collective Action (1965).

6 formation clusters detected across 3 registered agents.

Benford's Law deviation (calibrated against Nevada baseline)

First-digit distribution of contribution amounts compared not to theoretical Benford but to the empirical Nevada-statewide first-digit distribution. The baseline calibration removes the universal round-number preference (every NV donor gives in $25/$50/$100/$500 increments, which deviates from theoretical Benford uniformly). What remains is per-entity deviation from typical Nevada behaviour. Higher chi-square = more anomalous. Citation: Tam Cho & Gaines, "Breaking the (Benford) Law," American Statistician 61:3 (2007).

Shared officers across politically-active organizations

Officers (per Nevada Secretary of State business-entity records) who serve as named officer at three or more politically-active Nevada entities. Multiple nominally-independent advocacy nonprofits sharing officers indicate coordinated activity behind a fragmented public face. Citation: Mancur Olson, The Logic of Collective Action (1965). Capital Research Center / InfluenceWatch entity-relationship tracking model. 13 officers with 3+ politically-active overlaps.

Address co-occurrence — multiple entities at one location

Addresses where five or more campaign-finance contributors AND/OR registered agents are recorded at the same physical location. The straw-donor / contribution-laundering pattern: many distinct individuals at one address giving in coordinated amounts to the same recipient is a textbook NRS 294A.150 contribution-limit-circumvention signal. Addresses where contributors AND a registered agent share a single location are the highest-signal hosting pattern. Commercial registered-agent hubs (Corporation Service Company, CT Corporation System, etc., serving 100+ entities each) are excluded from this list — they are commercial RA services, not coordination signals. Citation: NRS 294A.150; FEC contributions-in-the-name-of-another, 11 CFR 110.4.

21361 total addresses flagged · 8710 with both contributors and a registered agent · 2225 with 10+ contributors. Top 30 non-commercial-hub addresses below.

Registered agents serving 5+ politically-active organizations

National commercial registered-agent services (Corporation Service Company, CT Corporation System) appear at the top by design — they handle hundreds of corporate filings statewide. The corruption-relevant signal is in the smaller agents serving niche clusters of politically-active entities at the same address.

Lobbyist-client communities (Louvain detection)

Bipartite community detection on the lobbyist-client representation graph. Each community is an "industry-as-observed" — a tightly coupled cluster of lobbyists and clients that work together more often than with outside actors. No taxonomy imposed; clusters emerge from the registry alone. Algorithm: Blondel et al., "Fast unfolding of communities in large networks," J. Stat. Mech. 2008. Method citation: LobbyView project (MIT).

Detectors added since first publication

Each new detector ships with a dedicated landing page; this section is the index. Click any detector card to see findings + methodology.