Causal sets in five minutes

Of all the background frameworks this programme borrows from, the one I find myself defending most quietly is causal set theory, because it is the only approach to quantum gravity that arrives with both a dynamics and a measure, which is to say it tells you what happens and how much to weight each happening. Here is the whole idea, and then why a multiverse programme wants it.

The idea

Strip spacetime of everything except causal structure. What remains is a set of events with a partial order: x precedes y if a signal can go from x to y. No metric, no manifold, no coordinates. A causal set is such a set with one extra property, local finiteness: between any two events there are only finitely many others. That is discreteness, stated as bookkeeping rather than as a lattice.

The remarkable fact is how much comes back. Given only the order relation, you can recover dimension, topology and, in a precise sense, volume: the number of events in a region is its spacetime volume, up to one fundamental length per event. The Hauptvermutung, the principal conjecture of the field, is that order plus count is the whole of geometry. A continuum spacetime is the continuum limit of a causal set, the way a fluid is the limit of molecules.

The dynamics nobody else has

Knowing the objects is half a theory. The other half is a rule for growing them, and causal sets have a serious candidate: classical sequential growth. Events are added one at a time; each new event is born to a random past, with probabilities that respect causality, meaning the new event can influence its future but the rule cannot conspire with the future to shape its own birth. The Rideout-Sorkin scheme makes this precise, and it comes with a measure: a definite probability for every possible history, from which you can compute expectations of anything.

Pause on how unusual that is. String theory gives you objects without a unique dynamics. Loop quantum gravity gives you kinematics whose dynamics is contested. Causal sets give you a growing universe with a normalisable measure over its histories, and questions like how many universes of each shape appear are well-posed. For a programme whose central claim is about many realities, having a rigorous measure over a growing multiverse of histories is not a detail. It is the rarest commodity in the field.

Why this programme leans on it

Three reasons. First, the substrate story: if branches of a multiverse are computed, the substrate has a discrete causal structure, and causal sets are the most conservative version of that, discreteness with no baggage. Second, growth: the framework’s branches are not static worlds but sequentially rendered histories, and sequential growth is the only off-the-shelf mathematics for exactly that. Third, the counting: the fact-budget argument in the computational series, the one with two hundred and forty-four digits, is at bottom a causal-set count, one event per Planck four-volume in a causal past.

The honest costs

The costs are real. The continuum limit is hard: nobody has exhibited our full four-dimensional world with its matter content emerging cleanly from a causal set, though pieces exist. The dynamics, as stated, is classical, and promoting it to a genuine quantum measure is where much current work lives; the framework’s use of it is at the classical skeleton level. And the approach’s austerity is a feature for derivations and a vice for phenomenology: there is very little in a bare causal set that a particle collider could see.

The programme’s position is that this is the right level of commitment: enough structure to count facts and grow branches, not so much structure that the multiverse claim smuggles in its own conclusion. The next two posts draw the lines this austerity has to respect: one from a three-slit experiment, one from a pair of photons.