A friend of mine teaches earth science at a public high school. We were grabbing lunch a few months back and he started venting about the interactive tools he's forced to use in class. "They all look like they were built in 2006," he said. And he's right. Most of them were. Flash-era color palettes, clunky drag-and-drop interfaces, layouts that barely function on a phone. His students take one look and check out.

That conversation stuck with me. I'd just shipped Molecular, my 3D chemistry learning tool, and the response from teachers had been genuinely encouraging. The whole point of Molecular was that students engage more when the visuals don't feel like homework. So when my friend described the state of earth science ed-tech, I already knew what I was going to build next.

The problem is real

I spent a week surveying what's out there. PhET Interactive Simulations from CU Boulder is the gold standard, and for good reason. The pedagogy is excellent. But the visuals are flat 2D diagrams. NOAA's Carbon Cycle Interactive is informative but static. ExploreLearning Gizmos and Concord Consortium's earth science resources are solid curriculum tools, but none of them look or feel like something a teenager in 2026 would voluntarily spend time with.

This isn't a knock on those teams. They built functional tools with the tech and budgets they had. But we're in a moment where browser-based 3D rendering is genuinely good, and students are growing up with Fortnite and Roblox. The bar for "visually engaging" has moved. Educational software hasn't kept up.

So I built Strata

Strata is a 3D web app where students manipulate the forces that shape Earth. It's the earth science counterpart to Molecular. Same design language, same tiered education system (Beginner, Standard, Advanced), same shell architecture. They're designed as siblings.

The app has three modules, each targeting a core earth science concept.

Tectonics visualization mode

Tectonics is the flagship. Users scrub a timeline across six geologic eras, from Pangaea at 250 million years ago all the way to a projected future 50 million years out, and watch the continents transform in real time. This isn't just repositioned modern outlines. I'm using real paleoshoreline data from Scotese & Wright's 2018 PALEOMAP PaleDEMs. Each era has true continent geometry, and plate polygons morph between eras using spherical linear interpolation. You can watch Pangaea tear apart, see the Atlantic Ocean open up, and track India's collision with Asia. On a phone. In a browser.

Atmosphere mode

Atmosphere is a 24-hour day cycle with general atmospheric circulation. The sun tracks across the sky on a tilted 23.44-degree axis. A season selector (Equinox, June Solstice, December Solstice) shifts the Intertropical Convergence Zone, temperature gradients, and wind belts. You can toggle layers for convection cells, a temperature heatmap, and cloud cover. Hover over any point on the globe and get local conditions. It's the kind of thing that makes Hadley cells click for a student who's been staring at a textbook arrow diagram.

Earth Systems mode

Earth Systems is the one I'm most excited about. It's a carbon cycle sandbox built on a mass-conservation engine that moves carbon between four reservoirs: atmosphere, ocean, biosphere, and lithosphere. Two forcing levers let students control fossil fuel emissions and land use changes. There are scenario presets for quick exploration. But the real hook is the visual feedback. As atmospheric carbon rises, the planet degrades. Land browns out. Clouds turn hazy and smoggy. It's a dying-planet visual that communicates consequences in a way no bar chart ever could. Students can scrub through years of simulated change and actually see what "400 ppm vs. 800 ppm" looks like painted across a globe.

The visual ambition is the point

I want to be direct about this: the game-quality rendering isn't polish. It's the entire thesis. If the visuals don't grab a 15-year-old in the first three seconds, the pedagogical depth behind them doesn't matter. They'll never get there.

The hub screen is a photoreal rotating Earth with PBR surface materials, an atmospheric rim simulating Rayleigh scattering, animated cloud layers, and on desktop at ultra settings, a cutaway revealing crust, mantle, outer core, and inner core. It's not a screenshot. It's running live in the browser. Zoom into a module and it transitions seamlessly into the learning environment.

The tech stack makes this possible without sacrificing performance. Three.js via React Three Fiber handles rendering. Next.js provides the app shell with server components for the static parts and client components for the 3D canvas. Zustand manages state. Tailwind CSS and shadcn/ui handle the overlay UI. An AI tutor powered by Vercel AI Gateway provides context-aware help, meaning it knows what the student is looking at and what settings they've changed.

I built a three-tier fidelity system so visuals scale with hardware. Desktop ultra gets HDRI lighting, bloom, ACES tone mapping, and screen-space ambient occlusion. A balanced tier drops the heavier post-processing. Mobile lite strips down to essentials while still looking good. This is the same approach I use in Structle, my bridge-building puzzle game, and Solar, my solar system propulsion simulator. If you're building 3D for the web and not thinking about adaptive fidelity, you're leaving most of your audience behind.

What I learned from Molecular

Molecular taught me a few things that directly shaped Strata. First, the tier system works. Beginners need less information and simpler controls. Advanced students want data density and precision. Trying to serve both with one interface fails. Second, mobile-first isn't optional for education. Students are on phones. Period. Every interaction in Strata is designed for touch first, mouse second. Third, teachers care about curriculum alignment more than features. The modules in Strata map to specific NGSS standards because that's what determines whether a teacher can actually justify using the tool in class.

Building Molecular also gave me the shell architecture I'm reusing. The hub, the tier selector, the AI tutor integration, the responsive canvas layout. Having that foundation means I could focus my time on the hard parts: paleoshoreline data processing, atmospheric circulation modeling, the carbon mass-conservation engine.

Why earth science, why now

Interactive effects of carbon levels over time on planet Earth

The carbon cycle module's relevance speaks for itself. But beyond climate, earth science is one of those subjects that suffers most from static teaching materials. Plate tectonics is a fundamentally spatial, temporal process. You can't understand it from a diagram. You need to see plates move over deep time. Atmospheric circulation is three-dimensional and cyclical. A cross-section diagram with arrows labeled "Hadley Cell" does not communicate what's actually happening.

These are concepts that demand interactive 3D visualization. The technology to deliver that in a browser, at 60fps, on a phone, exists right now. It just hasn't been applied to this space with any real visual ambition.

What's next

Strata and Molecular are the first two pieces of something larger. I don't have a grand unified product name for it yet, and I'm not sure I need one. But the pattern is clear: take a science discipline where students are stuck with outdated tools, build something that looks and feels like it belongs in 2026, and give teachers a reason to swap out the old Flash-era interactive for something their students will actually use.

I've got a shortlist of disciplines. I've got a shell architecture that scales. And I've got a teacher friend who keeps texting me suggestions.

That's usually how the best projects start.

Strata: Building a 3D Earth Science Tool That Students Actually Want to Use