H3 Measurement

How do you measure whether a "Third Horizon" of learning — an education system organized around flourishing rather than sorting — is actually emerging, and where are the biggest gaps? American education is in motion: AI, post-COVID disruption, and declining public trust are pushing schools, funders, and learners to ask for something different. Standardized test scores were built to answer a narrower question than the one the field is now asking, and the measurement infrastructure to answer the new question is partial, uneven, and under-implemented. This KB maps who is working on closing that distance.

The shape of the field, so far

The current entry point is LearnerStudio's December 2025 Measures, Metrics, and Indicators (MMI) for the Third Horizon brief — an 18-month synthesis built from 53 leading thinkers, twelve youth co-designers, and an indicator scan that grew to over 2,000 items in partnership with The Study Group. The brief is the foundation source for this KB; nearly every named actor below appears in it. Three of its most-cited partner organizations — EdInstruments, Mathematica, and Search Institute — have now been ingested directly from their published sources, substantiating (and in a couple of places gently complicating) the brief's second-hand descriptions.

Three claims from the brief structure most of what follows:

  • There's more measurement work happening than expected — but it's lopsided. The MMI scan surfaced over 2,000 vetted H3-aligned indicators. The field is not starting from scratch. But roughly 73% of those measures cluster at the Individual learner level, leaving the upper four levels of learning context — relationships, learning units, networks, and the broader ecosystem — comparatively undermeasured. The lens for that finding is the five-level framework the MMI project developed.
  • The implementation gap is the central diagnostic. Even where good measures exist, they don't reach educators — locked behind paywalls, not designed for practitioner use, over-reliant on surveys, or built in isolation from learning systems. The brief calls this the "critical choke point" preventing both educational innovation and the strategic deployment of capital, and most of its R&D recommendations are aimed at closing it.
  • Rigor is not the trade-off it's assumed to be. The brief pushes back on the framing that H3 measures must trade psychometric quality for relevance. Some early H3 measures already demonstrate validity; the active conversation is about expanding the definition of quality to also include practicality, usability, and relevance.

Who is doing what

The actors named in the brief sort roughly into four functional tiers:

  • Cataloging and curation — EdInstruments at Brown (searchable database of measurement tools) and Mathematica (Education-to-Workforce Indicator Framework — 99 recommended indicators).
  • Developing new assessments — the Urban Institute's Student Upward Mobility Initiative and the Learning Engineering Tools Competition (over $10M awarded). Both are funder-driven mechanisms for getting new measurement tools built.
  • Advancing methods and rigor — The Study Group (methods for assessment-in-service-of-learning), ETS and the Carnegie Foundation (psychometric rigor for competency-based measures), and Search Institute (the Developmental Relationships Framework — one of the few mature non-Individual-level frameworks named).
  • Capacity-building and practitioner translation — Transcend and the Learner-Centered Collaborative. These orgs sit at the seam where the implementation gap is most contested.

The MMI project's own funder of record is Lemnis. The participant roster names ~50 additional orgs across foundations, think tanks, capacity-builders, and learning-network operators — most of which are not yet promoted to entity pages. The brief's two co-authors — Babak Mostaghimi (LearnerStudio) and Jason Weeby (Periscope Advising) — are tracked as person entities.

How the actors map onto the levels

The MMI brief's "73% Individual" finding describes the measures surveyed in its scan. Mapping the actors surveyed in this KB onto the same five levels — see Actors across the five-level framework for the full breakdown — produces a corroborating but starker picture:

  • Individual: crowded. Six-plus actors with mature work overlap (EdInstruments, Mathematica E-W, Urban Institute SUMI, Learning Engineering Tools Competition, ETS, Study Group). The work at this level isn't create more — it's consolidate, translate, and close the implementation gap.
  • Relationships: Search Institute carries the level largely alone. The Developmental Relationships Framework is the named exemplar; whether its validated survey instruments are as accessible as the MMI brief implies is an open question — the framework page itself doesn't surface them.
  • Unit: no shipped frameworks. Transcend and the Learner-Centered Collaborative substitute capacity-building work for the missing measurement infrastructure.
  • Networks: empty. No surveyed actor's central contribution is Network-level measurement. R&D area #2 (Advance System-Level Research) is the brief's response.
  • Ecosystem: partial. The system conditions tier of Mathematica's E-W Framework does real work here, but as a secondary tier of a framework with a different primary axis.

Six R&D investment areas

The brief's most concrete output is a six-area menu of "high-leverage opportunities for catalytic investment." In the brief's own framing:

  1. Create shared infrastructure — an open-source indicator database, latitudinal data pilots across all settings where structured learning happens, Learner Education Record protocols, and verifiable micro-credential platforms.
  2. Advance system-level research — practical tools for measuring relational health and ecosystem-level H3 transformation. This directly attacks the level-imbalance gap.
  3. Build better instrumentation — integrated measurement packages for academic plus human skills; performance-based civic-readiness assessments; adaptive digital-literacy assessments (including ethical AI use); embedded real-time tools.
  4. Align funders and innovators — convening organizations, dedicated Implementation Labs to translate rigorous tools into practitioner-ready formats, and an advance market commitment for H3 tools.
  5. Include youth in co-design — beyond design sprints, including UX Labs where young people flag invasive or "creepy" assessment tooling.
  6. Make a market for better tools — state and district policy pilots; educator capacity-building and assessment literacy.

Each maps back to a specific diagnostic surfaced earlier in the brief. Three (Implementation Labs, advance market commitment, knowledge commons) are explicitly aimed at the implementation gap; two (system-level research, instrumentation) at the level imbalance; the others at the trust, awareness, and policy environment around adoption.

Key tensions

The brief frames these as "growing pains" rather than dead ends, and as the natural surface for R&D investment:

  • Holistic skills agreement vs. no consensus measures. Universal field agreement that durable human skills matter; no commonly accepted measurement set.
  • Creation vs. use. Measures exist but aren't designed for easy implementation or aren't on educators' radar — the implementation gap in tension form.
  • Authentic vs. accountable. Portfolios, graduate portraits, and qualitative artifacts sit in real tension with the comparability demanded by state and federal accountability frameworks.
  • Local innovation vs. comparability. Community-driven assessment doesn't roll up cleanly across districts.
  • Private R&D vs. public infrastructure. Private companies lead a lot of next-generation assessment R&D, raising risk that proprietary products consolidate access and power.

What learners are asking for

A March 2025 SXSW design sprint with twelve learners aged 14–18 from Dallas surfaced a coherent set of preferences: assess what learners can do with knowledge rather than recall; favor final products and verbal explanation over traditional documentation; treat learning as process, not destination; give learners voice in defining success; allow mastery-based pathways; recognize learning wherever it happens (including outside school); and integrate well-being. The brief notes that these preferences align with University of Virginia researcher Shereen El Mallah's broader youth-engagement findings.

Where to start reading

Open questions

  • Whether the upper four levels of the five-level framework are actually maturing — beyond the Search Institute example, who has shipped working tools at relationships, unit, network, or ecosystem level?
  • Whether any organization has actually built infrastructure that closes the implementation gap in production — not in concept. The Transcend / Learner-Centered Collaborative capacity-building tier is the obvious candidate to investigate.
  • Whether the "advance market commitment" pattern proposed in R&D area #4 has any real-world precedent in education funding. Adjacent fields (vaccines, climate tech) have used the pattern; education would be new ground.
  • Where the public-infrastructure vs. proprietary-products tension is actually playing out — which next-gen assessment vendors are private, which are nonprofit / public, and what posture does each take on access?
  • Whether the LearnerStudio canon (Learning to Flourish, Human Skills in the Age of AI, the concept note) substantively extends or merely restates the MMI brief's framing. The user has these documents queued for ingest.
  • Whether the systems-change-indicator question has been operationalized anywhere — what does measuring "the health of an H3 ecosystem" actually look like, with real data?