Independent impact intelligence

Put every dollar
where it matters most.

A living market of the world’s most promising funding opportunities—compared across evidence, expected impact, and room for more funding.

Build a portfolio →How we compare impact

—accepted Coefficient rows

—published row amounts

—fund lenses in the ledger

—database snapshot retrieved

YOUR GIVING, MADE EXPLICIT

Build a portfolio.
Keep every assumption visible.

This is an allocation worksheet, not an impact ranking. Your cause weights set the dollars; evidence and practical constraints determine which accepted opportunities can receive them.

PROPOSED—Building from current evidence…

UNALLOCATED—Never redistributed across causes silently

VERIFY ROOM—Selections without a published numeric cap

Reapplying your explicit constraints…

SAN FRANCISCO OUTCOME CONTRACT

A local dollar needs
a local outcome.

Before ranking organizations, define what success means. This first ontology keeps durable outcomes separate from administrative proxies and delivered services—and blocks donor-attribution and health conversions until the required model exists.

OUTCOMES DEFINED—Mutually legible, not necessarily additive

OFFICIAL SOURCES—City, federal, and school-system definitions

MODEL REQUIRED—Days avoided and deaths averted

QALY / WELLBY BLOCKED—Until a versioned local conversion exists

Loading the versioned local measurement contract…

THE PUBLIC BASELINE

What City money
is already here?

Private giving enters a field already shaped by public budgets and contracts. This baseline shows those accounting signals without pretending that authorized dollars equal outcomes—or that a large contract proves a small philanthropic gap.

CITY BUDGET—Current fiscal year approved spending · all departments

OUTCOME-RELEVANT ENVELOPES—Loading mapped departments…

ACTIVE NONPROFIT CONTRACTS—Fully executed prime-supplier contracts active on snapshot date

OUTCOME-LINKED—Conservative text rules loading…

Reconciling approved budgets and active nonprofit contracts…

SAN FRANCISCO · CANDIDATE UNIVERSE

Start with the field.
Then earn a recommendation.

A major donor should see the denominator, not just a hand-picked shortlist. This first complete layer groups every nonprofit prime-contractor name in the active DataSF snapshot, then separates source discovery from real impact diligence.

Loading the complete active-contract universe…

SAN FRANCISCO · IRS IDENTITY LAYER

6,688 EINs.
Still not 6,688 answers.

The IRS universe widens discovery beyond city contractors and gives each registered organization a durable identity anchor. It tells us who has a San Francisco filing address—not who serves San Francisco well, or what another dollar would accomplish.

Loading the San Francisco EIN universe…

LOCAL GIVING DILIGENCE

Six candidates.
Zero automatic endorsements.

This first cohort tests a GiveWell-style local scorecard across direct service, community infrastructure, housing advocacy, and elections. It asks what a donor can actually infer—and keeps unknown marginal impact visible.

CANDIDATES REVIEWED6Intentionally incomplete discovery cohort

ACTIVE CONTRACT MATCHES35Exact prime-contractor aliases · accounting context

LEDGER OVERLAPS1Exact accepted-grant identity matches

PUBLISHED MARGINAL GAPS0Unknown is not zero funding room

501(c)(3) · San Francisco Bay Area; current city contracts confirm San Francisco programs.

Hamilton Families

family homelessness prevention, shelter, rapid rehousing, and housing subsidies

organization-reported outcomesMarginal gap unpublishedQALY blockedWELLBY blocked

THEORY OF CHANGE

Temporary assistance, landlord partnerships, shelter, and case management can prevent displacement or help families move into stable housing; child supports may reduce education disruption.

WHAT THE EVIDENCE CURRENTLY SAYS

The April–June 2026 report states 338 families served, 34 program exits to stable housing, 19 families receiving eviction-prevention assistance, and $660,709 in housing subsidies.

CAUSAL BOUNDARY

The report supplies service and exit counts, but no comparison group, durability follow-up, cost per sustained housing outcome, or donor-attribution estimate.

NATIVE SIGNALSNever collapsed into one score

output338

Families served

April–June 2026

organization-reported outcome34

Program exits to stable housing

April–June 2026

output19

Families receiving eviction-prevention assistance

April–June 2026

PUBLIC-FUNDING OVERLAP8 exact contract matches

AUTHORITY$60.05M

PAID$33.5M

REMAINING AUTHORITY$22.18M

Exact prime-contractor alias matches in the active-contract snapshot; figures are accounting context, not philanthropic room.

educational attainment · housing stability

NEXT-DOLLAR STATE

Room remains unknown

No reviewed, time-bounded marginal funding gap or next-dollar plan is published in the accepted sources.

DOWNSIDE CASE

What could make this less effective?

Subsidies may replace public dollars or move families between programs without increasing durable housing supply; housing exits may not persist.

DONATION VEHICLE

501(c)(3)

Organization states donations are tax-deductible to the extent allowed by law.

Donation information ↗

CHARITY NAVIGATOR100% · 4 stars

The displayed overall score is based on Accountability & Finance only; it is not an impact estimate.

Profile ↗

Hamilton FamiliesQuarterly Report (April–June 2026)2026-04-01 to 2026-06-30 · retrieved Aug 30, 2026↗Hamilton FamiliesFinancialsFY2024–2025 documents available · retrieved Aug 30, 2026↗

READING RULES

Discovery. Candidate discovery is not a recommendation. A headquarters address, public contract, Charity Navigator profile, or self-reported impact claim does not establish marginal impact.

Ratings. Charity Navigator ratings are preserved with completed-beacon coverage when published. They are governance, finance, and sometimes measurement signals—not Market for Impact effectiveness scores.

Causality. Organization-reported outputs and policy or election wins remain contribution evidence unless an independent counterfactual design supports an outcome claim.

Conversion. QALY and WELLBY estimates are blocked for every candidate until a versioned local model, defensible counterfactual, and uncertainty analysis exist.

sf-nonprofit-diligence-v0.7 · An intentionally incomplete first diligence cohort chosen to test direct service, advocacy, LGBTQ+ community support, food security, housing, and multi-service models. Inclusion is not a recommendation.

CAUSE DISCOVERY · COMPLETE SOURCE UNIVERSE

Beyond the EA lists.
Before the recommendation.

Charity Navigator surfaces 441 organizations tagged “LGBTQ rights.” This explorer contains every current search result, then shows exactly what still needs diligence before Market for Impact can recommend one.

Social supportHealth careLegal servicesAdvocacy

DISCOVERED ORGANIZATIONS44145 source pages · not an MFI rank

FOUR-STAR PROFILES163Charity Navigator signal only

PROFILE-REVIEWED10Deep fields retained for first ten

PUBLISHED FUNDING GAPS0Not yet assessed is not zero

SEARCHHEADQUARTERSCN RATING

VISIBLE441

Candidate	Headquarters	Charity Navigator	Completed beacons	MFI evidence state
Lambda LegalEIN 237395681 Public affairsLegal servicesLGBTQ rights	New York, NYService geography unknown	90% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
Out in Tech Inc.EIN 464379985 EducationTechnologyLGBTQ rights	NEW YORK, NYService geography unknown	94% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
1n10 Inc.EIN 860728990 LGBTQ rightsYouth servicesYouth organizing	PHOENIX, AZService geography unknown	93% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
OUT MaineEIN 203682307 Community organizingLGBTQ rightsYouth development	ROCKLAND, MEService geography unknown	98% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
RevoiceEIN 824858874 ChristianityNonprofitsLGBTQ rights	SAINT LOUIS, MOService geography unknown	90% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
Atlanta PrideEIN 582032010 Community and economic developmentArts and cultureLGBTQ rights	ATLANTA, GAService geography unknown	92% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
Woodhull Freedom FoundationEIN 113681116 Public policyAdvocacy journalismFreedom from violence and torture	Jenkintown, PAService geography unknown	90% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
The Freedom Center for Social JusticeEIN 454000599 LGBTQ rightsDiversity and intergroup relationsSpecial population support	Charlotte, NCService geography unknown	93% · 4 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
The Frederick CenterEIN 461705400 Health care accessMarriage equalityLabor rights	FREDERICK, MDService geography unknown	84% · 3 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
TransFamiliesEIN 454910834 Equal opportunity in educationAdolescent parentingChild development	SEATTLE, WAService geography unknown	84% · 3 starsNo advisory in accepted search field	4/4Accountability & Finance · Impact & Measurement · Leadership & Planning · Culture & Compensation	Research leadImpact evidence unassessed Funding room unassessed	↗
PFLAG GREATER BOSTONEIN 043272394 LGBTQ rights	WALTHAM, MAService geography unknown	86% · 3 starsNo advisory in accepted search field	Not reviewedSource-index-only row	Research leadImpact evidence unassessed Funding room unassessed	↗
Just Fund Education Project Inc.EIN 208465456 EducationPhilanthropyLGBTQ rights	LEXINGTON, KYService geography unknown	82% · 3 starsNo advisory in accepted search field	Not reviewedSource-index-only row	Research leadImpact evidence unassessed Funding room unassessed	↗

Page 1 of 37

What this is. This is the complete current Charity Navigator “LGBTQ rights” search universe, paginated in its source-default order. It is not curated-list membership or a ranking by Market for Impact.

What ratings mean. The overall score and stars are Charity Navigator signals. They are not an MFI effectiveness score, and four completed beacons do not establish causal impact or marginal cost-effectiveness.

What is missing. City and state are headquarters fields. Service geography is left unknown until an organization-level review establishes it. The first ten profiles retain reviewed beacon and source-publication fields. Remaining rows are source-index-only, so those profile fields stay null. Rating-specific dates were not exposed as accepted fields and remain null.

Decision state. Every candidate remains a research lead. Impact evidence and current room for more funding have not yet been assessed, so this slice makes no giving recommendation.

DISCOVERY SOURCECharity Navigator · LGBTQ rights search441 results at retrieval · default ordering↗RATING METHODSpring 2026 methodology guideBeacon weights, eligibility, alerts, and curated-list criteria↗

charity-navigator-lgbtq-discovery-v2 · retrieved Aug 30, 2026 · exact-name crosswalk found 0 evaluator and 0 historical grant-ledger overlaps; none are treated as effectiveness evidence

THE OPPORTUNITY MARKET

Compare the next dollar.

These are research leads, not a universal ranking. Unlike outcomes stay visibly unlike until a defensible conversion is available.

⌕Cause area

#	Opportunity	Cause	Evidence	Published metric	Room for funding	Research source
Loading accepted opportunity assessments…

This table contains accepted assessment rows materialized in the ledger, including current recommendations, pooled funds, and explicitly labeled published research. Metrics preserve each evaluator’s native unit; unlike outcomes are deliberately not collapsed into a universal score.

THE EVALUATOR MAP

One cause.
Five different questions.

Compare what each evaluator actually publishes: direct recommendations, pooled funds, historical grant lenses, decision criteria, evidence date, and marginal funding status. Empty cells stay explicit.

Reconciling five evaluator sources against the accepted database…

Loading source reconciliation… · Absence is not a negative evaluation · Grant history is not a recommendation · Native outcome units remain incomparable

COMPARABLE IMPACT · V0.1

Translate with assumptions.
Never erase the native unit.

QALYs, WELLBYs, $CG, and cash-benchmark multiples are different models—not interchangeable labels. This lab exposes the formula and uncertainty, then refuses conversions whose required evidence is absent.

Loading versioned formulas and native metrics…

MARGINAL FUNDING CURVE · V0.1

The next dollar needs
a specific job.

Room for more funding is not one permanent organization score. Each row below is one source-backed amount, period, use, confidence statement, and counterfactual-funder field. Missing evidence stays missing.

Reconciling marginal funding evidence…

THE GIVEWELL MARKET

Current programs.
No fake precision.

GiveWell’s current Top Charities share a funding bar, but not a single donor-ready rank. We preserve each program’s evidence, native delivery unit, model version, geography, and live funding-room process.

Cash comparison—

Loading versioned benchmark definitions…

Loading accepted GiveWell assessments…

Historical metric. Cost-per-life figures are GiveWell’s reported averages for 2022–2024 directed funding—not literal outputs of the newer location-specific models.

Source discrepancy. The accepted row sum is exactly $3 above Airtable’s displayed aggregate. Both are preserved; neither is silently “fixed.”

Top Charities updated September 2025 · Cost-effectiveness framework updated May 2026 · Loading database freshness… · Top Charities ↗

THE ANIMAL WELFARE MARKET

Current recommendations.
Different animals, intact units.

ACE’s current set spans corporate campaigns, policy, technology, food-system work, and wild-animal research. We preserve each program’s own denominator, evaluation vintage, and uncertainty instead of manufacturing one cross-species league table.

Current set—

Recommended Charities in ACE’s 2025 set.

2025 reviews—

Awarded or renewed using the newer model.

Retained 2024—

Still current under two-year status.

Published annual room—

Descriptive sum across two overlapping funding periods.

Comparability boundary. ACE metrics retain their original units and vintages; loading accepted assessments…

Loading accepted ACE assessments…

Loading database freshness… · 2025 recommendation announcement ↗

THE CLIMATE MARKET

Current best bets.
Systems change, no fake tonnage.

Giving Green’s current recommendations span firm power, hard-to-abate industry, food systems, aviation, and shipping. Its research is qualitative at the organization level, so we show strategy, funding need, and announced grants without inventing emissions-per-dollar scores.

Top nonprofits—

Alphabetical, explicitly not ranked.

Announced grant rows—

Five top nonprofits plus 24 strategy grantees.

Full cycle announced—

Planned grants; not proof of disbursement.

Top nonprofit share—

Grant size is not an effectiveness rank.

Comparison boundary. Loading the accepted Giving Green assessment and grant ledger…

Loading accepted climate assessments…

Loading database freshness… · Current top nonprofits ↗ · Grant announcement ↗

THE FOUNDERS PLEDGE MATRIX

One evaluator.
Four evidence regimes.

Founders Pledge mixes modeled direct interventions, partner research, active pooled funds, and high-uncertainty bets. This matrix preserves those differences—and shows exactly where a GiveDirectly comparison does and does not exist.

Funding opportunities—

Funds, programs, and published organization recommendations.

Cause areas—

Education, climate, global health, and catastrophic risks.

Current pooled funds—

Climate and Global Catastrophic Risks.

Cash-relative estimates—

Only Imagine Worldwide carries an explicit multiple.

Comparison boundary. Loading the accepted Founders Pledge assessment matrix…

Education—

Climate—

Global health—

Global catastrophic risks—

Loading accepted Founders Pledge assessments…

Loading source freshness… · Current pooled funds are separated from older published recommendations · Founders Pledge Funds ↗

THE AI SAFETY ECOSYSTEM

Follow the field,
without forcing a rank.

Every accepted Coefficient grant in the Navigating Transformative AI fund is mapped into an auditable, multi-label role taxonomy. Founders Pledge recommendations are overlaid as a separate signal.

Published grants—

Complete accepted Coefficient fund lens.

Published amount—

Historical row amounts, not funding room.

Named organizations—

Recipient identities in the public index.

FP overlap—

Five matched recommendations and one external-only.

Interpretation boundary. Published grant flow is neither an effectiveness score nor current room for more funding. Categories overlap, so category totals must not be added.

SEARCH ORGANIZATIONS

Loading the accepted AI safety taxonomy…

Loading taxonomy freshness… · Coefficient fund ↗

THE RENPHIL FRONTIER

Funding the tools
behind discovery.

Renaissance Philanthropy’s AI for Math Fund backs open research infrastructure that individual academic and industry labs may lack incentives to build. These are disclosed awards—not a cost-effectiveness ranking.

Fund-level commitmentNot row-level

The announced commitment is excluded from grant totals because RenPhil does not allocate it across awards.

Published portfolio—

Loading accepted award rows…

Rows with amounts—

Application caps and fund totals are never substituted.

Named funderXTX Markets

RenPhil administers the fund and supports grantees.

AWARD 01AMOUNT NOT PUBLISHED

An AI-Focused Tactic for Language Learning

The project aims to enhance AI proof agents by developing a domain-specific tactic language for Lean, designed to better align with AI capabilities and…

AWARD 02AMOUNT NOT PUBLISHED

A Dataset of Modern Formalized Theorem Statements

This project creates a public dataset of hundreds of formalized statements of recent theorems from top journals, such as the Annals of Mathematics. In…

AWARD 03AMOUNT NOT PUBLISHED

A Principled Approach to Proof Search with Applications to Siderenko’s Conjecture

This project proposes new proof methods to automate and streamline the proof discovery process for significant unsolved problems. To this end, this work seeks…

AWARD 04AMOUNT NOT PUBLISHED

A Structured Representation of Tactics for Machine-Assisted Theorem Proving

This project develops a machine learning-driven approach to creating tactics for interactive theorem proving. The project will represent tactics as algebraic objects in a…

AWARD 05AMOUNT NOT PUBLISHED

Bridging AI, Proof Assistants, and Mathematical Data (BRIDGE)

The BRIDGE project aims to integrate AI with proof assistants and mathematical data by creating and curating large-scale datasets and dependency graphs from formalized…

AWARD 06AMOUNT NOT PUBLISHED

Bridging Complexity and Automation to Advance Automatic Theorem Proving

This project automated theorem proving through three innovative research directions that address key conceptual questions about proof difficulty, tractability and performance scaling. In particular,…

AWARD 07AMOUNT NOT PUBLISHED

Bridging Proof and Computation

The project focuses on developing an extensible, native interface between Lean and the computational algebra system (CAS) Macaulay2 (M2) to enhance the integration between…

AWARD 08AMOUNT NOT PUBLISHED

Constraining LLMs for Theorem Proving

The goal of this project is to advance autoformalization—the task of translating informal mathematics into formal statements that can be automatically checked by state-of-the-art…

Coverage conflict. The publisher’s declared portfolio and linked project records do not reconcile; the accepted ledger preserves that unresolved gap.

Capital signals stay separate. RenPhil also reports organization-level catalyzed, directly raised, and unlocked capital. None is treated as a grant-ledger total because it cannot be reconciled to the published award rows.

Loading database freshness… · Official winners page ↗

TRACE THE CAPITAL CHAIN

Follow one dollar without counting it twice.

Choose one publisher ledger, then follow each source row from originating funder through adviser, grant, recipient, intervention, and cause. Unknown links stay unknown.

Accepted source rows—

Counts can be combined because source-row identities are distinct.

Publisher ledgers—

Dollar totals remain inside one selected ledger.

Overlap excluded—

Coefficient EGC rows already occur in its complete index.

Selected rows—

One row stays one flow, even with multiple recipients.

NON-ADDITIVE

Cross-publisher dollar totals stay separate because accepted publications may describe the same underlying funding.

Reconciling source rows…

One row, one amount. Recipient roles never duplicate a grant amount.

Role discipline. Originating and advising funders remain separate.

Missing stays missing. Unsupported fields are never inferred.

Loading source reconciliation… · cause-tag counts may overlap inside a ledger · Publisher source ↗

QUALITY BEFORE CONFIDENCE

See what the data can—and cannot—support.

A donor should be able to distinguish a real source conflict from a documented publication limit, a missing field, or an old snapshot. This register uses rules, not an opaque quality score.

Tracked source pages—

Reconciling freshness…

Accepted grant rows—

Distinct publisher rows; amounts remain non-additive.

Amounts unpublished—

Unknown is retained rather than converted to zero.

Canonical dates missing—

Award, decision, or disbursement date by ledger.

Source conflicts—

Publisher totals or declared coverage disagree.

Disappeared rows—

Not called retracted without publisher evidence.

CONTENT-ADDRESSED—

Snapshot sources with a stored semantic hash.

REVIEWED REFERENCE—

Method pages retained without pretending a page hash was captured.

FUNDING ROOM—

Reconciling tranche evidence…

Reconciling accepted sources…

OPEN EVIDENCE REGISTER

Conflicts, gaps, and boundaries.

Counts describe missing evidence or source behavior. They do not rank evaluator quality.

SourceIssue state

VISIBLE ISSUES0

SOURCE INVENTORY

Every tracked page has a freshness state.

Loading explicit freshness rules…

Grouped rows. Grouped status is counted only when published row evidence supports it.

Loading quality state… · zero observed grouped flags does not override publisher grouping caveats · missing is not zero

THE FLOW OF CAPITAL

See where impact funding is actually going.

Recommendations reveal what researchers believe. Grants reveal what funders do. Market for Impact connects both.

Explore all funding flows →

ACCEPTED LEDGER TOTALSA first map of the disclosed ecosystem

D1 LIVE

Coefficient public indexLoading accepted rows…

Not published

GiveWell grant exportLoading accepted rows…

Not published

Coefficient EGC subsetLoading accepted rows…

Not published

Giving Green 2025 cycleLoading accepted rows…

Not published

RenPhil AI for MathLoading accepted rows…

Not published

These database totals are intentionally not summed.
The Coefficient EGC ledger overlaps its complete index, while cross-publisher exports may describe the same underlying funding. Missing RenPhil amounts remain missing.

THE COEFFICIENT MARKET

One index. Current fund lenses.

— accepted public grant records, classified against every fund on Coefficient’s current funds page. Overall totals count a source record once; fund rows are intentionally non-additive.

Published records—

Published amounts—

Recipient names—

Index coverage—

Coverage, not certainty. Loading database coverage checks…

Amounts are partial. Loading database missingness checks…

QUERY THE LEDGER

Trace individual grants.

Filter the complete D1-backed source ledger without losing Coefficient’s own fund tags or published-record semantics.

Querying the ledger…

Loading database source metadata… · Fund taxonomy ↗

FIRST LIVE GRANT LEDGER

Inside one Coefficient fund.

Every currently accepted record from the Effective Giving & Careers fund page, normalized without claiming that “published” means paid.

Published records—

Published amounts—

Distinct recipients—

Latest decision month—

RecipientPurposeAmount

Loading the D1-backed ledger…

Loading database freshness… · Content-addressed · Removed records remain detectable through last-seen timestamps.

SOURCE LEDGER

Every claim should lead back to evidence.

Coefficient GivingGrant publishing process and coverage↗GiveWellTop charities, updated Sep 2025↗GiveWellBenchmark semantics, updated May 2026↗ACE2025 recommended charities↗Giving Green2025–26 climate recommendations↗Founders PledgeEducation evidence & recommendations↗Coefficient Giving$CG cost-effectiveness methodology↗HM TreasuryWELLBY appraisal guidance↗

A COMMON LANGUAGE FOR IMPACT

Comparable where possible.
Transparent where it isn’t.

Evidence

How confident should we be that the intervention causes the intended outcome?

Marginal impact

What additional good is expected from the next dollar—not the average past dollar?

Funding room

How much capital can the organization productively deploy, and on what timeline?

Uncertainty

Which estimates are measured, modeled, judgment-based, or fundamentally incomparable?

Put every dollarwhere it matters most.

Build a portfolio.Keep every assumption visible.

A local dollar needsa local outcome.

What City moneyis already here?

Start with the field.Then earn a recommendation.

6,688 EINs.Still not 6,688 answers.

Six candidates.Zero automatic endorsements.

Families served

Program exits to stable housing

Families receiving eviction-prevention assistance

Room remains unknown

What could make this less effective?

501(c)(3)

Beyond the EA lists.Before the recommendation.

Compare the next dollar.

One cause.Five different questions.

Translate with assumptions.Never erase the native unit.

The next dollar needsa specific job.

Current programs.No fake precision.

Current recommendations.Different animals, intact units.

Current best bets.Systems change, no fake tonnage.

One evaluator.Four evidence regimes.

Follow the field,without forcing a rank.

Funding the toolsbehind discovery.

An AI-Focused Tactic for Language Learning

A Dataset of Modern Formalized Theorem Statements

A Principled Approach to Proof Search with Applications to Siderenko’s Conjecture

A Structured Representation of Tactics for Machine-Assisted Theorem Proving

Bridging AI, Proof Assistants, and Mathematical Data (BRIDGE)

Bridging Complexity and Automation to Advance Automatic Theorem Proving

Bridging Proof and Computation

Constraining LLMs for Theorem Proving

Follow one dollar without counting it twice.

See what the data can—and cannot—support.

Conflicts, gaps, and boundaries.

Every tracked page has a freshness state.

See where impact funding is actually going.

One index. Current fund lenses.

Trace individual grants.

Inside one Coefficient fund.

Every claim should lead back to evidence.

Comparable where possible.Transparent where it isn’t.

Evidence

Marginal impact

Funding room

Uncertainty

Put every dollar
where it matters most.

Build a portfolio.
Keep every assumption visible.

A local dollar needs
a local outcome.

What City money
is already here?

Start with the field.
Then earn a recommendation.

6,688 EINs.
Still not 6,688 answers.

Six candidates.
Zero automatic endorsements.

Beyond the EA lists.
Before the recommendation.

One cause.
Five different questions.

Translate with assumptions.
Never erase the native unit.

The next dollar needs
a specific job.

Current programs.
No fake precision.

Current recommendations.
Different animals, intact units.

Current best bets.
Systems change, no fake tonnage.

One evaluator.
Four evidence regimes.

Follow the field,
without forcing a rank.

Funding the tools
behind discovery.

Comparable where possible.
Transparent where it isn’t.