Counterfactual Store vs. Markov Chains: From Branching State to Reach, an Engine for Possible Futures
The Counterfactual Store does resemble a Markov chain at the conceptual level. Both deal with states and the possible transitions between them. But a Markov chain is mainly a mathematical model for probabilistic transitions: given that I’m in state A, what’s the probability the next state is B, C or D? And the classic Markov property says the next-state probability depends on the current state, not the full history.
The Counterfactual Store is deterministic branching, not probabilistic transition. If the current application state is A, you create a hypothetical A′, change something, propagate the consequences, inspect the result and throw it away without touching A. No probability needs to be involved anywhere.
Still, the Markov comparison points toward a more radical primitive. Combine the two ideas and make possible futures first-class data.
StateSpace: a graph of what can happen
Imagine a StateSpace engine. Instead of a database holding one current state, it holds a graph of possible states and the transitions between them. Some transitions are deterministic, some conditional, some probabilistic. You could ask:
NOW: what state am I in?NEXT: what states can follow?PATH target: how could I reach a desired state?RISK condition: through which transitions could I reach an undesirable state?WHATIF action: what possible states does this action produce?
That’s not really a conventional database anymore. It’s closer to an executable state-space primitive.
Take a manufacturing process in the state machine=running, material=available, order=active. The engine knows that material→empty leads toward machine→waiting. Replenishment leads back toward running. Machine failure leads toward maintenance. Instead of storing only what happened, it represents what can happen from the present state.
Reach: the reachable future as a primitive
There’s a startup-level idea hidden in there: a tiny engine that continuously maintains the reachable future of a system. Call it Reach.
An application supplies its current state plus transition rules. Reach keeps answering one question: what states are reachable from here?
That could sit under workflow validation, infrastructure safety, manufacturing, cybersecurity attack paths, dependency analysis and AI agent safety. Before an autonomous agent acts, Reach could check whether the action opens a path toward a prohibited state.
Conceptually, the primitive is beautifully small:
STATE + RULES → REACHABLE STATES
Markov chains could be an optional extension where transitions carry probabilities, but they wouldn’t define the architecture. Graph search, constraints and state transitions would be the core.
Reach is more interesting than the Counterfactual Store because its identity is clearer. A database tells you what is. A temporal database tells you what was. Precomputing keeps answers ready. Reach tells you what can happen.
That’s different enough to join Distill and Precomputing in the serious-screening pile. And the four-question screen goes straight to the uncomfortable question: who already spends money because they don’t know which states are reachable?
That answer decides whether Reach is a primitive with a business behind it or just an elegant computer science project.