Information → body → information → future

Convergence is
not equivalence.

Two developmental histories reached almost the same visible body. Then, left alone, their futures separated again. The organism had forgotten in appearance—but not in state.

39organisms
4held-out identical actions
780new futures
6,240new future frames
01 / The finding

The same-looking body can still occupy a different developmental state.

Under the closed information history, target branches fell from +1.405 SD at the end of feedback to −0.043 SD after release—then separated again to −0.607 SD over 128 untouched steps.

-0.609 SDclosed-history branch future under same actions-0.609 [-1.478, +0.149]
+0.326 SDopen-history branch future under same actions+0.326 [-0.008, +0.651]
+0.935 SDopen minus closed future interaction+0.935 [+0.017, +2.028]
+0.762 SDsame interaction with no new pulse+0.762 [-0.351, +1.999]

The latent difference was dynamical, not merely visual: it described where the body would go next.

02 / The long trajectory

One history relaxed. The other crossed through itself and rebounded.

Target-branch separation across three phases

-0.5+0.0+0.5+1.0+1.5open gateclosed gatefeedback endrelease +128future +128

Positive means the expansive-target branch still looks more expansive. The closed-gate history crosses zero and keeps moving.

The controller had not simply written a body. It had written a trajectory through body-space.

When the two closed-history branches looked nearly equal, one was still moving through the shared appearance toward a different future. Near-convergence hid direction and momentum.

03 / The same future challenge

Fresh identical actions preserved the hidden split—but did not create it.

Branch future under four held-out actions

-1.8-1.1-0.4+0.3+1.0openclosed16326496128steps after apparent convergence

Across four unused actions, the open and closed information histories develop opposite target-branch futures. Their +128 interaction is +0.935 [+0.017, +2.028].

Extra separation caused by action, beyond no pulse

-0.3-0.1+0.1+0.3+0.5openclosed16326496128steps after apparent convergence

The pulse-specific amplification is modest: open +0.171 [-0.045, +0.387]; closed -0.002 [-0.165, +0.185]. The autonomous trajectory is the principal result.

04 / One organism

At the handoff, almost identical. Later, measurably different.

Source 51 is an illustration selected because its two closed-history branches were almost equal at the handoff (−0.015 SD) and separated over the next untouched future (−2.327 SD). The visible difference remains subtle—which is precisely the point.

Source 51 compact-target history at apparent convergence
compact-target historyapparent convergence
Source 51 expansive-target history at apparent convergence
expansive-target historyapparent convergence
Source 51 compact-target history at 128 more untouched steps
compact-target history128 more untouched steps
Source 51 expansive-target history at 128 more untouched steps
expansive-target history128 more untouched steps
05 / What changed

We need to model organisms as moving states, not frozen shapes.

Information historysets an early mode of causal responsiveness
Body feedbackpushes the organism along an imposed morphological direction
Visual convergencedifferent trajectories pass through similar-looking forms
Future divergencetheir hidden direction reappears in what the body becomes next

This gives us a sharper target than “predict the final creature.” We should infer the hidden developmental vector: not what the body is, but where its internal dynamics are carrying it.

06 / Next direct swing

Read the direction before the form turns.

Take states that are morphologically matched but historically different. Probe their short local futures, infer a developmental tangent from those futures alone, and ask whether that tangent forecasts the later direction of body change across unseen organisms.

That would connect the information geometry directly to organism emergence: a hidden causal vector preceding the visible turn.