01 / Possibility and assurance
One successful future tells us something. It doesn't tell us everything.
Imagine two forests facing a disturbance. I can identify a successful continuation for each. At first, that seems encouraging: in both cases, recovery remains possible.
But let me now widen our view. Suppose the first forest has two admitted responses and both succeed, while the second has a successful response and another response in which its defining organization fails.
Successful continuation is present.
Successful continuation is present here, too.
This is an intuitive two-path illustration of an information boundary. It is not the exact counterexample constructed in Paper I and does not assign probabilities to the illustrated responses.
Both forests answer yes when all I ask is whether a successful continuation exists. Yet the second forest carries a failing response that a success-only description has left out.
If I intend to evaluate robustness against a declared challenge, that missing information may be decisive. A representation that reports only successful continuation cannot always settle a question that also depends on what can go wrong.
Two different logical claims
At least one. Every admitted pathway.
The symbol ∃ means “there exists”; ∀ means “for every.” The first expression requires one successful history. The second describes a stronger illustrative condition over a specified history collection L. It is not the complete formal definition of robustness in Paper I.
- ∃
- “There exists”: at least one successful pathway can satisfy the first statement.
- ∀
- “For every”: the second statement must hold for each pathway in the specified collection.
- L
- The specified collection of histories over which the universal claim is evaluated.
- Successful(γ)
- The declared success predicate applied to a particular history γ.
The gap between these two statements is worth keeping in mind. Knowing that something can continue does not establish how it responds across the full set of challenges we may care about.
02 / A proved information boundary
Paper I gives this distinction an exact mathematical witness.
In the robustness portion of Paper I, I construct systems that agree on their positive successful-persistence information but differ in their robust-persistence judgments.
Success-only information versus robustness
Equal positive capacity. Different robust judgments.
Here the superscript + marks the positive successful-persistence representation. This is a formal comparison between declared systems, not an empirical finding about any actual forest.
This establishes a sharp limitation: positive successful-continuation information does not, in general, determine robust persistence. A representation that includes how a system fails, which challenges are admitted, or what happens under a control policy may be needed to answer richer robustness questions.
I regard a result of this kind as valuable because it tells us where information has to be added. Later in the series, richer response structures will explicitly retain both successful and unsuccessful outcomes.
03 / A different limitation
What if we observe only the present?
Let's return to the finite forest model. Its baseline state, B, and its perturbed state, P, both presently realize the defined forest organization.
Now hold the scenario and time horizon fixed: mild conditions, short horizon. Their represented capacities differ, and one of our requirements can detect that difference.
| What I evaluate | Baseline · B | Perturbed · P |
|---|---|---|
| Presently realizes the organization? | Yes | Yes |
| Represented viable capacity | {stable, exposed} | {exposed} |
| Stable response available? | Yes | No |
| Health under stable-response requirement? | Yes | No |
Both present states realize the organization. That single observation gives them the same answer. But their prospective Health judgments differ once I evaluate capacity against the stable-response requirement.
The implication is direct: present realization alone is not a sufficient observation for every prospective Health judgment. Two states can agree on what we observe now and differ in what the declared future requirement demands.
04 / The information needed for a judgment
What does a sufficient observation actually mean?
We often say that a measurement tells us something about a system. In this theory, I want that claim to have a precise meaning: the information supplied by the measurement must determine the health question we're asking it to answer.
Call an observation of state x by the symbol O(x), and call the health judgment H(x). The observation is sufficient for that judgment when equal observation values guarantee equal health answers:
Observation sufficiency
Equal observation values must preserve the answer.
The arrow ⟹ means “implies.” The double arrow ↔ means “if and only if”: the truth values of the two Health judgments agree. Here H is the particular declared Health predicate being evaluated.
Notice how demanding this is. If O gives the same value for two states while H says true for one and false for the other, the observation has erased a distinction that matters. It is insufficient for that question.
To make the intuition familiar, imagine a hypothetical physiological observation that records only whether a person is conscious. Matching consciousness readings would tell us little by themselves about the two people's capacities under a specified future physiological challenge. I use this as an illustration of information insufficiency, not as a validated medical application of the framework.
Our formal forest example provides the exact mathematical case. The same present-realization result fails to distinguish B from P, while the declared prospective requirement does.
05 / Changing how we describe a system
A new description should preserve what the judgment depends on.
There is another important part of Paper I that I want to bring into view: presentation invariance. A system may be described in different ways. Its states could have numerical identifiers in one representation and descriptive names in another.
Changing a label alone should leave a properly translated Health judgment unchanged. More substantial changes can also preserve it, but we must carry the relevant structures faithfully.
The two descriptions agree on which states realize the specified organization.
Corresponding states and contexts retain appropriately matching viable-response capacity information.
The declared requirements are translated so that their adequacy answers agree.
Under the appropriate compatibility conditions, Paper I proves the Health judgment is preserved. The result supports faithful changes of presentation. It also gives us a way to examine whether a simplification has discarded something necessary for the original question.
Papers III and IV return to this problem with more detailed results about representations, sufficiency, information loss, and conservative enrichment.