Post 1 in a series specifically focused on primitives for higher education. If you're new, three prior posts set up the frame.

The bundle

Most higher education in America is sold as a bundle. One institution packages instruction, assessment, credentialing, student support, physical plant, social life, and administrative overhead into a single product, priced as a single product. For many learners, that bundle is worth every dollar. For millions of others, it's inaccessible, unaffordable, or simply more than they need to get the credit they actually came for.

Most reform efforts in higher education try to make "the bundle" cheaper or more accessible. Online degree programs. Competency-based approaches. Shorter completion times. Better financial aid design. While many of these are worth doing, they by and large rely on the same underlying architecture: a student is still buying a whole package, from one institution, at an all-in price.

There's a different move available, and it's the one I want to work through in this series. Instead of reforming the bundle, it’s possible to build something else: small independent components that a wide range of individuals and institutions can freely combine into pathways to college credit.

What's a primitive?

I'm going to call these small independent components primitives, following terminology I first encountered through Ben Thompson. A course. A standardized exam. A scholarship. A tutor (whether human or AI). These are all primitives in the sense I mean the term.

The concept of primitives isn't new. The video game designer Steve Grand argued that complex intelligent behavior could emerge from simple computational building blocks without top-down design. As Brad Stone documented in The Everything Store, Jeff Bezos and the early AWS team then took the idea and applied it to their own infrastructure problem: rather than building monolithic technology systems, they built small, simple technology components and offered those to developers, initially within Amazon and then as a new product offering.

The result was Amazon Web Services, which provided technology building blocks from which anyone could build anything they imagined. Many organizations’ core infrastructure (including ours at Modern States) runs primarily on AWS, and those that don’t often run on something similar, just built by another company after AWS proved out the model. The powerful idea is that Amazon didn't plan for the specific use cases of any of those companies in particular; they were emergent.

I want to take these concepts and apply them to higher education, in particular as a way I’m thinking about what Modern States does and how to improve higher education and systems in general.

When I think about primitives, I’m thinking of components with four main characteristics. They are:

1. Modular: each does one job and can be used independently.

2. Standardized: each connects to other components through predictable interfaces.

3. Composable: the same pieces can be assembled into many pathways.

4. Swappable: better pieces can replace weaker ones without breaking the system.

Modern States has been building or using education primitives since 2017. The organization builds some primitives and buys others. Today the stack includes:

• Free online courses that we build;

• College-Level Examination Program (CLEP) exams that College Board builds and that we buy;

• Scholarships to pay for CLEP exams (that we interchangeably call vouchers) that we award and largely fund;

• Student support offerings (like AI tutors) that we integrate;

• Tools (like our CLEP acceptance database) that we design to inform or cultivate the broader ecosystem.

Each of these is a primitive in the sense I mean it.

"Primitives" versus "unbundling"

Unless you squint really closely, it’s easy to consider primitives just part of a process of unbundling, especially because I’ve juxtaposed what we do with “the bundle” of traditional higher education. But I think creating primitives is a better description of what's actually going on, and the distinction matters.

Unbundling is subtractive or destructive. You start with an existing institutional bundle and take pieces out of it, one at a time, so that the parts can be sold or delivered separately. A lot of higher-ed unbundling has taken exactly this shape: take a university course, strip it from the rest of the university, and offer it as a standalone product. The resulting course still largely relies on the existing university infrastructure, though.

Where unbundling is about breaking down, primitives are about building up. They are built from scratch as standalone infrastructure that didn't previously exist as such. A CLEP prep course, in the Modern States sense, is a new atomic unit, designed to stand alone for whoever wants to use it, but more powerfully to connect to the CLEP exam. In this example, the primitives are designed to produce something that already exists and is legible within the broader environment, college credit, but in a new way.

Part of the reason why legibility matters is that the bundle persists for real demand-side reasons. The degree solves a transaction-cost problem in the labor market: it compresses a candidate's underlying competence into a signal employers already understand. Most efforts to replace it run into network-effect problems familiar from any other standard. I'll work this argument out in more detail later, but the implication for what follows is that primitives are most useful, at least in the beginning in higher education, if they compose into the existing legible credential rather than around it.

At Modern States, we offer one version of primitives combined into a more powerful configuration: individual learners can create Modern States accounts (an identity primitive), take a course at their own pace (a learning primitive), earn a scholarship (a finance primitive), take a CLEP exam (an assessment primitive), and hopefully earn college credit. But other institutions that use the primitives we've created can do their own composing: some embed our content and courses into existing learning pathways, or wrap the core learning experience with additional supports. A state system, a library, an employer, or an individual learner can take the same pieces and assemble them differently. As the builder, Modern States can't possibly anticipate every composition, and we don't need to. The architecture is what matters; the applications are emergent.

Where I'm going next

The rest of this series tries to sharpen the framework, stress-test it against how we actually operate, and work out why this feels like such a different way of operating. I'll walk through what a primitive is in more technical detail and lay out the stack as I’m currently thinking about it: a data and identity foundation, the core composable primitives themselves, and the articulation layer where a passed exam becomes credit on a transcript at a specific institution (the layer we only partially own, and that probably matters more than any other).

From there the posts get more applied (for everyone's sake, because it's dense otherwise). Reference architectures and implementation models, which is the language I’m starting to use for the ways partners put the primitives together. The first-best-customer pattern, which is the reason we deploy our own primitives at scale rather than handing them off. And finally an important question for how all this evolves: whether the end point is ubiquity, the way TCP/IP or the shipping container are ubiquitous, or whether the analogy stops at AWS.

The goal is to improve my own thinking, but also to figure out whether there are other individuals or organizations doing similar work already, who might be enlisted on this journey, and whether there's a real place to go. I think there is, and I think that primitives enable the type of quick wins that my first piece framed out and which are so important for a new approach.

Sources & references

On primitives and AWS as a model

• Brad Stone, The Everything Store: Jeff Bezos and the Age of Amazon (Little, Brown, 2013). Documents the Grand-to-Bezos transmission and the early design choices that shaped AWS.

• Ben Thompson, Stratechery. The contemporary articulation of how primitive-building shapes platform strategy. The "Amazon Tax" (2016) is the clearest entry point.

On unbundling in higher education

• Kevin Carey, The End of College: Creating the Future of Learning and the University of Everywhere (Riverhead, 2015). A book-length version of the unbundling argument; the foil against which the primitives frame becomes sharper.