Making the frame visible
What are we actually optimizing for?
Most debates about progress are actually debates about definitions. People argue about whether things are getting better or worse, but rarely pause to ask: better or worse by what measure, for whom, compared to when?
Daniel Miessler built Substrate to make such questions answerable. It’s a structured framework that starts from first principles - what are humans, what do we need, what does flourishing actually look like - and works toward measurable dimensions: health, wealth, freedom, knowledge, meaning, and a few others. The goal isn’t a utopia spec. It’s a shared problem definition, grounded in data, that you can actually argue about rather than just assert.
What drew me to it was the epistemological move: define the problem before proposing solutions. That’s rare. Most policy, most tech discourse, most public debate skips straight to the fix and treats the frame as obvious. Substrate insists on making the frame visible.
The DE/EU fork
The original Substrate skews heavily toward US data and US context. That’s not a criticism - Miessler built it from where he stands. But if you’re trying to understand human flourishing in Germany, or Europe, you need different numbers and propably different questions.
I forked it: gitea.giersig.eu/svemagie/Substrate.
What I’ve added so far: German labor market data (Bundesagentur für Arbeit), energy production via SMARD, federal budget data via Bundeshaushalt.de, parliamentary tracking via DIP, and social mobility indicators from Destatis and the OECD. Fourteen datasets, all templated consistently, all source-linked.
A few things become immediately visible when you look at Germany through this lens:
- Social mobility is worse than it looks. Germany spends 4.6% of GDP on education (OECD 2020), but children from academic households are four times more likely to attend Gymnasium than children from working-class families. The money is there. The sorting mechanism is running anyway.
- The energy transition is measurable, not just claimed. Renewables now generate more than half of Germany’s electricity in good quarters. That’s a structural shift, not a press release.
- The lobby register has gaps. The Bundestag registry has improved but still contains significant underreporting. Substrate makes that visible as a governance data quality problem, not just a political one.
Why this matters for AI
Substrate is also, explicitly, a context layer for AI systems. If you’re building something that makes decisions - or assists with them - it needs a model of what “good” means. Substrate is one attempt at that model: explicit, revisable, and empirically grounded rather than baked into training data as invisible assumptions.
The AI connection is why it lives at the intersection of the AI and Substrate tags on this site. I’m not building an AI product. But I’m increasingly convinced that the hardest problem in AI isn’t technical - it’s the one Substrate is trying to solve: what are we actually trying to achieve, and how would we know if we got there?
The fork is public. Have a look, contribute, push back. If it gets solid enough, I’ll send a PR upstream.