Unit of Analysis
RTM does not measure a single number or a single match. The unit of analysis is a network of information, sources, roles, time, and observer, in which the relations among the data matter no less than the data themselves.
Reflective Time Model (RTM) — Eran Harpaz
The Reflective Time Model (RTM) is a research framework for examining the possibility that information about reality is not merely a passive record of a world that has already been completed, but part of a broader informational structure in which records, events, identities, memories, observations, dates, representations, and states of knowledge can participate in a single system of constraints spanning time, source, and role.
RTM is a general model of information about reality. It is not limited to any particular type of object or event. It can examine information about people, events, texts, numbers, times, places, images, dates, identities, outcomes, questions created during an investigation, and changes in record or interface states.
The everyday distinction between a “person,” “event,” “time,” “text,” “object,” “memory,” or “record” is a useful human way of describing the world. In RTM, each of these can function as an information state or a measurement point within the same system. The question is not only what each item is in isolation, but how different information items connect, constrain, reconstruct, and at times may even update relative to one another when history is examined as a single structure.
The encoding is the measurement tool; the recurring architecture and the coherence of the information system across time are the measured phenomenon.
RTM at a Glance
Chronology
The model distinguishes between information that was frozen in advance and information that was created, revealed, or became accessible later. The order in which information enters the system is part of the data, not a minor technical detail.
Control
The same Search Contract must apply both to the real finding and to the comparison worlds. If the real world is allowed a particular degree of freedom in representation, search, depth, or operation, that same freedom must also be available to the alternatives.
Interpretation
RTM separates three different questions: do the relations exist; is the architecture anomalous relative to appropriate alternatives; and what physical mechanism, if any, produces it. Resolving the third question is not a prerequisite for testing the first two.
What Is RTM?
RTM examines a recurring pattern in which information items from different sources and times do not appear merely as a collection of isolated matches, but begin to form a shared system of constraints.
This distinction is central. A single numerical match may be accidental. Even two matches do not necessarily indicate a system. In a world containing countless names, dates, times, texts, numbers, and forms of representation, it is expected that exact relations will occasionally be found even without any deeper regularity.
RTM therefore does not ask only: “Is there a relation?”
It asks a stricter question: Do several different relations begin to jointly determine the same system?
Such a system can contain different paths reaching the same target, branches that reconverge, reconstruction of event data from other data within the same event, reconstruction of identity or observer, recurrence of internal nodes, loops, fixed points, transitions between representations, and preservation of semantic role.
Information created in one place can reappear from another source. Information frozen at an early stage can acquire structural significance at a later stage. A question formulated during the investigation can itself enter the network. A record can change, and the transition between states can become an additional datum.
None of these components alone defines RTM.
The phenomenon begins when several of them operate together as part of the same architecture.
Structural Overdetermination
One of the central principles in RTM is Structural Overdetermination.
This means that the same target, event, identity, or structure does not depend on a single path. If one path were removed from the network, other parts could still continue to constrain or reconstruct the same target.
Overdetermination can arise through different paths, different source families, different temporal layers, direct and indirect relations, different semantic roles, different representations, and different blocks of dependence.
This creates an important distinction between the number of equalities and the amount of independent structure that remains in the network.
Ten equations may in practice be ten ways of writing the same constraint. By contrast, a smaller number of paths arising from genuinely different constraints can carry more structural information.
RTM therefore does not assign strength to a finding merely because it contains many lines.
The question is what remains after duplicates are removed, dependent relations are merged, Dependency Compression is applied, and the network is deliberately challenged.
An alternative explanation must reproduce the architecture — not merely generate numerical equalities.
The Phenomenological Law of RTM
RTM calls the recurring pattern:
The Phenomenological Law of Cross-Temporal Information Organization
Within the operational domain tested, heterogeneous information systems repeatedly display precise, role-preserving, dependency-aware architectures in which information items from different sources and times reconstruct, constrain, and complete one another as part of a coherent global structure.
The claim here is first and foremost phenomenological: it describes the form of organization observed at the level of information and record.
In observer-dependent cases, the full organization is also linked to the appropriate information profile of the observer and is not necessarily preserved when that profile is replaced by an alternative profile.
It is therefore important to distinguish between describing the regularity itself and explaining it.
One can ask whether the architecture exists and whether it is anomalous before knowing why it exists.
The precise physical mechanism that produces it is a separate question.
What Is Actually Measured?
RTM does not test whether “a particular number is special.”
The measured object is an architecture of relations.
The architecture includes sources, anchors, roles, dates, timestamps, representations, internal nodes, paths, loops, dependence among paths, the order in which information enters, record states, observer identity, and information created during the investigation.
Each component derives meaning from its position within the system.
For example, the same numerical value is not necessarily equivalent to itself in every context. If in one case it represents time and in another identity, RTM preserves the distinction between the roles. A path that reaches the correct value through the wrong role is likewise not necessarily equivalent to a path that preserves the correct semantic mapping.
The question therefore is not only: “How many matches were found?”
but rather: “What complete system of constraints is obtained, and how difficult is it for an alternative system to reproduce it under the same rules?”
Numerical Information as a Measurement Ruler
In the current implementation of RTM, Hebrew text can be converted to a numerical value according to a fixed rule, alongside predefined numerical representations.
The same string under the same rule always yields the same value.
This makes it possible to map names, sentences, dates, times, numbers, titles, places, identities, and event data into a common measurable language.
The transition to a numerical language allows information items of different kinds to be compared using a uniform system of rules.
But it is important to emphasize:
The encoding is not the phenomenon itself.
The encoding is the ruler.
The phenomenon being tested is the organization that emerges among information items after they are converted into a common language. RTM therefore does not rely on the assumption that any number is inherently meaningful. The research significance lies in the relations, roles, chronology, and architecture formed among the data.
The Observer as an Information Structure in Time
The observer is a central component of RTM, not peripheral metadata.
Operationally, depending on the protocol, an observer profile can include first name, surname, full name, year of birth, age, family relations, place of residence, apartment number, frozen biographical anchors, as well as information, questions, and actions created during the investigation.
At the conceptual level, however, RTM examines a broader possibility: the observer need not be only a “person at a particular moment.” The observer can also be understood as an information structure extended through time — identity, memories, records, knowledge states, choices, access to information, and the relations among all of these.
In observer-dependent findings, identity is not necessarily merely the target one is trying to reach. At times it is itself resolved from the network; and in stronger cases direct links to identity can be removed while its reconstruction still emerges through indirect relations.
RTM therefore distinguishes between two levels:
The observer as a measurable variable
The observer can be replaced through Observer Replacement, allowing one to test whether the same architecture is preserved when the identity profile changes.
This is an empirical test.
The observer as a possible part of the mechanism
RTM examines whether observation, choice, wording, memory, access to a record, or the creation of new information may participate in organizing the system itself.
This is a mechanistic hypothesis.
The two levels are related but not identical. Measurable observer dependence is not by itself proof of any particular physical mechanism.
However, if the observer is an information structure extended through time, then continuity of identity, memory, and records also becomes part of the physical question rather than merely an external frame from which the data are viewed.
Information Is Part of Reality, Not Merely a Description of It
RTM does not treat a document, memory, photograph, message, observation, or record as an abstract layer outside reality.
Each exists through some physical substrate: a brain, screen, paper, server, sensor, signal, object, or another system that carries information.
An important distinction follows. “Information about reality” is itself part of the reality in which it is stored, transmitted, and made accessible to an observer. A record state or memory state is therefore not only a description of another event; it is also a present state of the system.
In RTM, the implication is that records, memories, representations, and knowledge states can be included in the same research framework as the events to which they refer.
The model does not separate in advance “the world” from “the information through which we know the world.” It examines the relations between them as part of one system.
Frozen Anchors
Not every datum receives the same weight.
There is a fundamental difference between a number printed on an object before the investigation began and a number selected after the target is already known; between an automatically generated time and a time chosen from a range; between a message received externally and wording selected after testing alternatives; and between a fixed historical date and data that can be altered retrospectively.
RTM therefore records Frozen Anchors.
The earlier information was fixed, the more external its source, the better documented it was, and the less it could be altered after the result was seen, the greater its experimental force.
The idea is simple: the less freedom remains after seeing the result, the harder it is to explain the structure by retrospective fitting alone.
A Question Created by the Observer Is Information
In some cases a question or sentence is formulated before its value is calculated.
In such a case the wording itself is not merely an external tool used by the researcher to describe the finding. It is a new information event within the chronology of the case.
The text, the choice of words, the time at which it was written, the person who wrote it, and the state of knowledge that existed at that moment are created together.
If such new information later connects to anchors that were already frozen, RTM treats it as part of the chronological data of the system.
This must, however, be clearly separated from adaptive search.
A one-time formulation fixed before calculation is not equivalent to selecting a formulation after testing many alternatives.
The distinction matters because the question is not only what the text says, but when it was created and under what state of knowledge it was created.
From Closure to Network
A closure is an exact relation in which several values reach a value with a defined role.
But RTM almost never stops at a single closure.
The output of one path can become the input of another. The same node can arise from a second source. A number can be converted into a verbal representation and re-enter the network. A later path can return to an earlier node. Several paths can converge on the same point and then branch again.
When this happens, the unit of research changes.
It is no longer an equation.
It is a graph.
The graph contains the sources, anchors, nodes, paths, roles, loops, information-entry times, and dependencies among the relations.
The central question is therefore not whether a striking closure exists, but whether that closure is embedded in a broader architecture.
RTM Architecture Map
01 · Event Self-Reconstruction
Event self-reconstruction
Documented fields of the event reconstruct other fields of the same event, before or alongside convergence on identity.
02 · Multi-Anchor Identity Resolution
Multi-anchor identity resolution
One bounded unit can reconstruct several identity or family components as part of the same network.
03 · Role Preservation
Role preservation
Not only does the correct number return, but the correct role returns: father as father, sender as sender, time as time, place as place, and a person in the role in which that person appears in the source.
04 · Cross-Path Convergence & Branching
Cross-path convergence and branching
Materially different paths reach the same node, from which additional paths can open.
05 · Recursive Re-Entry
Recursive re-entry
The output of one stage returns later as a new input and creates depth in the network.
06 · Loops · Fixed Points · Symmetry · Checksums
Loops, fixed points, symmetry, and checksums
A path returns to its source or to an earlier node and creates an internal verification mechanism.
07 · Cross-Source Node Re-Entry
Cross-source node re-entry
A node produced in one source returns from another source and connects units that were measured separately.
08 · Question → Answer Reconstruction
Question reconstructing its answer
A formulated or documented question can enter the same network from which the person, event, time, or datum it asks about is obtained.
09 · Cross-Representation Closure
Cross-representation closure
A number, number words, spoken digits, date, name, and text can move between defined representations and meet within the same system.
10 · Record-State Transition
Record-state transition
An earlier and a later state of the same record are preserved separately, and the change between them itself becomes data.
11 · Chronological Carryover
Cross-stage carryover
Information frozen at an early stage returns later in a new and precise role.
12 · Live External Input
Live external input
A message, timestamp, photograph, result, or another field arriving from outside enters the network under a chronology that cannot be rewritten retrospectively.
13 · Distributed / Latent Observer Reconstruction
Distributed observer reconstruction
Even after direct links to identity are removed, the indirect constraint system can continue to resolve the same profile.
14 · Observer-Linked Checksum Architecture
Observer-linked checksum architecture
Different subsystems — event, time, identity, family, and internal nodes — verify one another as part of a single network.
What Makes a Network Strong?
The anomaly in RTM is not the mere existence of an equality.
A network becomes more informative as several non-trivial features accumulate together: external and frozen sources, multiple anchors, reconstruction of several roles, paths that are not the same equation written differently, recurrence of internal nodes, loops, self-reconstruction, observer and role specificity, rigid chronology, and transitions between sources.
But even this accumulation is not sufficient by itself.
RTM asks what remains after Dependency Compression, after Knockout, and after comparison with Null worlds operating under the same Search Contract.
In other words, strength does not arise only from the appearance of the network, but also from its ability to survive systematic attempts to dismantle it.
History as an Information State
In RTM, the “past” is examined not only as an event that has already occurred, but also as the present information state from which we know what occurred.
A historical document, photograph, object, record, memory, or old result is not located, at the time of testing, inside the original moment in which it was created. It exists now as an information state referring to another time.
Alongside it exists the observer’s present information: identity, knowledge, moment of access, choice, measurement action, and a registry of anchors that have already been frozen. Later, new fields can enter — information created, revealed, or made accessible only after other parts of the system were already fixed.
RTM therefore examines history not only as a sequence of events but also as aninformation history: a system in which earlier and later states, records about them, and the observer accessing them can all be part of the same chronological object.
The question is not only what came first and what happened later, but whether all these information layers jointly maintain a coherent architecture that is not well explained when they are separated from one another.
Global Coherence of History
RTM’s strong hypothesis does not have to be phrased as “the future causes the past.”
A deeper formulation is that history as a whole can be examined as a single solution in which earlier and later states are jointly subject to a system of global constraints.
In such a structure, each point in time need not derive its full meaning only from the point preceding it. One can examine a system in which both early and late boundary conditions participate in determining the complete coherent structure.
RTM’s question is therefore not only:
What caused what?
but also:
What complete history satisfies all the constraints together?
This shifts the description from a local chain of causes and effects to the examination of a cross-temporal structure in which relations among parts of history are tested as a whole.
Updating the Information State
From this follows RTM’s most radical hypothesis.
If a record, memory, observation, and knowledge state are part of the physical information system, and if the complete history is subject to global constraints, one can examine whether the entry of a later constraint does more than merely add a new layer of information — whether it also participates in a coherent update of the overall information state.
Such an update could, in principle, be expressed in relations among data, in representations, in records, in information accessible to the observer, and, in the strongest version of the hypothesis, also in a memory state.
In such a model, the system need not preserve alongside the new state an “external copy” of every previous state from which the observer could identify that a change occurred. The accessible state after the update may simply be the system’s current coherent state.
This is an L6-level mechanism hypothesis. It is not required in order to measure RTM findings and is not presented as an already proven result.
Is This “Rewriting the Past”?
In everyday language, part of this possibility can be described as “rewriting the past,” but RTM uses the phrase carefully.
The claim need not be that a physical event that already occurred was erased and then occurred again differently.
The possibility being examined is that the information state that defines history for the system — records, memories, representations, relations, and knowledge states — may participate in a dynamic process of coherent updating.
In this sense, the “past” we are able to test is always also the present state of information about the past.
RTM leaves the ontological question open: whether this is only an update of the effective representation of history, an observer-relative structure, cross-temporal boundary conditions, retrocausality, or another mechanism not yet formulated.
The Corpus as a Cumulative Experiment
The RTM corpus is not presented as a collection of unrelated examples.
It is structured as aPressure Ladder.
The idea is that each layer changes a central variable that could have served as an alternative explanation for the previous layer.
The source can move from a broad archive to an independent historical event, from such an event to a single family object, from there to a physical record that changes state, to live information, to a later public outcome, and to distributed reconstruction of the observer.
The source type, scale, chronology, and possibility space therefore change — but the question remains the same:
Does the same family of architectures continue to appear as the source conditions change?
Pressure Tests and Controls
RTM deliberately tries to break the findings.
This is not a separate layer from the phenomenon, but an attempt to create competitors for it.
Observer Replacement
In Observer Replacement, the person or identity profile is replaced while the rest of the structure is held fixed as far as possible.
Role Replacement
In Role Replacement, we ask whether the same values continue to work when their roles are changed.
Temporal Specificity
In Temporal Specificity, the day, time, year, or temporal point is changed and the effect on the architecture is examined.
Exhaustive Enumeration
When the alternative space is finite, Exhaustive Enumeration is used to traverse the entire space.
Monte Carlo / Adaptive Search
When the space is larger, Monte Carlo / Adaptive Search allows the Null worlds to search for themselves under the same rules and search budget.
Knockout / Ablation
In Knockout / Ablation, strong parts of the network are deliberately removed and what remains is tested.
Dependency Compression
In Dependency Compression, multiple formulations of the same constraint are not allowed to count as though they were independent evidence.
Whole Finding
In Whole Finding, the unit being compared is the complete network — not a collection of separate p-values.
Connecting Numbers Is Easy. Reconstructing an Architecture Is Hard.
With sufficient arithmetic freedom, relations can be created among many numbers.
This is exactly why RTM does not stop at the question: “Can some value be reached?”
The stricter question is:
Under the same rules, can the correct source, correct role, correct observer, correct nodes, correct chronology, and correct relations be reconstructed within the same network?
The gap between these two questions is central.
A general ability to connect numbers is not equivalent to an ability to reconstruct a specific system of constraints.
Chronology Is Part of the Data
In RTM, temporal order is not a footnote.
A source can exist years earlier, be frozen in a particular state, be exposed to the observer only at a later stage, enter the investigation, become linked to new information, change state, and be examined again after the change.
There is therefore a principled difference between a static pattern and Chronology-Constrained Reconstruction.
In a chronological structure, it is not enough to know what data exist at the end of the process.
One must also know when each item existed, when it became accessible, and when it could be used.
Chronology constrains the search and is therefore part of the measured object.
RTM’s Hierarchy of Claims
To avoid conflating measurement with interpretation, RTM separates six layers.
L1 · Source and Documentation
Is the record authentic, complete, and documented?
L2 · Encoding and Relations
Are the values and calculations correct and deterministic?
L3 · Architecture
Does a genuine network remain after dependencies and duplicates are merged?
L4 · Selectivity
Does the structure survive against defined alternatives of identity, time, role, and source?
L5 · The Phenomenological Law
Does the same family of architectures recur systematically within the tested domain?
L6 · Physical Mechanism and Ontology
What mechanism of reality, if any, produces the regularity?
This separation is essential because a claim at one level does not automatically answer the level above it.
Failure to resolve L6 does not erase a result at L1–L5.
And equally:
A strong L5 result is not by itself proof of any specific L6 mechanism.
The Mechanism Proposed by RTM
RTM examines the possibility of:
Self-Consistent Cross-Temporal Information Organization
that is, self-consistent organization of information across the history of the system.
Under this hypothesis, an earlier state and a later state are not necessarily two completely isolated units of information. They may both be parts of the same global solution.
In such a framework, the complete description of the system need not be constructed only as a sequence in which each state is explained unidirectionally by the state that preceded it. One can also examine a description in which different parts of history jointly satisfy a single system of constraints, including early and late boundary conditions.
Instead of viewing time only as a chain of local updates, RTM asks whether there is a layer in which the coherence of the complete history is itself a constraint.
If so, information that becomes accessible later can be structurally related to information recorded earlier, not necessarily because it “sends a message backward,” but because both belong to the same information history examined as a whole.
This is a proposed mechanism, not a final conclusion.
Why “Reflective Time”?
We experience time as a sequence:
past → present → future
RTM asks whether beneath this local experience there is an additional layer of organization: a global information history in which different parts of time do not merely follow one another, but can constrain, reconstruct, and complete one another.
Later information can complete a structure whose sources are earlier. An earlier record can acquire a precise role only after later information appears. A node created at one stage can return from another stage. The observer, memory, and record state can participate in the same system.
If so, past, present, and future are not examined merely as three separate information stores, but as different regions of a single history.
In this sense, time is Reflective: parts of history reflect, constrain, and complete one another as part of one structure.
Relation to Existing Research — Not Proof of RTM
RTM is a separate framework, and its findings are not derived from the quantum literature. However, several concepts that the model places under examination appear explicitly in existing research, expressed in different physical languages.
Effective Past and Revised Histories
Within his framework, Henry Stapp describes an “effective past” that is recreated with observation, “revised effective histories,” and even the effective elimination of records associated with rejected parts of an effective past. At the same time, he emphasizes that the actual event that already occurred does not necessarily change.
Records and Memory Can Change
Veronika Baumann and Časlav Brukner analyze a Wigner’s Friend scenario in which a later measurement changes the observer’s memory register. In their model, after the change the observer need not retain knowledge of the previous result or even awareness that the memory changed.
Information and Memory Are Physical
Andrea Di Biagio and Carlo Rovelli formulate information as something physical stored in variables and correlations. They explicitly include memories, records, and present observations among the variables that constitute the perspective from which the world is known.
A Time-Delocalised Observer and Global Constraints
Bethany Terris proposes describing an informational observer as time-delocalised: a structure whose record coherence extends through time. She examines coherent history through initial and final boundary conditions and describes a logical structure determined by global constraints rather than only a local causal chain.
Facts Need Not Be Absolute
Wigner’s Friend experiments and analyses by Proietti, Bong, and others place constraints on maintaining all classical assumptions about absolute facts, locality, and freedom of choice simultaneously. More recent work also examines explicit formulations of time symmetry and no-retrocausality.
Classical Reality as a Network of Records
Quantum Darwinism describes how the environment redundantly encodes information about a system, allowing different observers to access different fragments of the environment and agree on the same classical reality. Here too, accessible objectivity emerges through a distributed structure of information.
A Present Defined by the Structure of the Future
In the context of black holes, Emily Adlam emphasizes that the location of an event horizon can be defined using facts about future history, and discusses the possibility that the analogy to Wigner’s Friend points toward global, teleological, or retrocausal explanations.
The significance for RTM is that its conceptual language is not disconnected from existing physics: physical information, relative facts, records and memory, cross-temporal coherence, boundary conditions, and global constraints already appear explicitly in scientific discussion — RTM brings them together within a single framework and places their joint organization under test.
The Radical Hypothesis
Beyond the phenomenological law, RTM places a stronger interpretation under examination.
Reality may not consist only of a sequence of independent states updating forward in time, but of a global informational system in which information belonging to different temporal points, its records, and the observer accessing it are jointly constrained as part of a complete solution.
In its strongest version, RTM asks whether, when new constraints become part of the system, not only the mathematical relations among them reorganize, but the information state from which history is accessible — representations, records, knowledge states, and possibly memory — can also participate in coherent updating.
Within such a framework one can examine early and late boundary conditions, symmetry between earlier and later information, retroactive informational completion, dependence of a record state on a later state, and the participation of observer, choice, and observation in the actualization of information.
This is not a claim that the findings have already proved that “all of reality is rewritten.” It is the strong ontological hypothesis being tested against the phenomenological law and the stringent controls already applied across the corpus.
The evidence for structure and the interpretation of mechanism remain two separate questions.
Symmetry Between Times
RTM’s strong claim is not merely that “the future influences the past.”
Such wording keeps us within an ordinary causal picture and simply reverses the direction of the arrow.
The broader possibility being tested is structural symmetry between earlier and later information: both sides may act as boundary conditions of the same system, without assuming in advance that one has a more fundamental ontological status than the other.
If so, the complete history need not be described only as an initial state that unidirectionally generates everything that follows. It can be examined as a structure in which constraints from both sides of the sequence participate in determining one consistent solution.
The deeper question is therefore:
What complete history satisfies all the constraints together?
This shifts the description from a sequence of points to a description of a complete structure.
From Theory to Findings
This page presents the model and its levels of claim. The cases shown on the findings page are only a limited sample from a much broader corpus of RTM findings; they were selected as representative examples of the structures, source types, and different controls — not as a complete list of findings that have been collected and tested.
The Testable Prediction
One of RTM’s central predictions can be formulated independently of resolving the question of the microphysical mechanism.
Suppose there are observers A1, A2, A3… and record sets R1, R2, R3…, where R1 actually belongs to A1, R2 to A2, and so forth.
If genuine record–observer specificity exists, then on average we expect:
Score(R1,A1) > Score(R1,A2)
and likewise for the remaining pairs.
That is, Matched Observer–Record pairs should display stronger structure than mismatched pairs.
The importance of this prediction is that it can be tested without first deciding why the specificity occurs.
Prospective Controls Already Implemented
RTM’s stringent protocols are not presented as a future stage that has yet to be carried out.
Across the corpus, depending on the case, the work has already used anchors frozen before the outcome appeared, chronological documentation, later external input, observer and identity replacements, matched Null models, Exhaustive Enumeration, Knockout / Ablation, Dependency Compression, and Whole Finding.
In cases where new information became available only at a later stage, the earlier anchors were preserved as frozen anchors so that the later information could not select them retrospectively.
This means that prospectivity, chronology, and comparison against alternatives are not merely methodological proposals of RTM — they are already part of the controls applied to the different findings across the corpus.
RTM in Summary
RTM examines the possibility that information about reality does not always behave as a collection of separate records, but exhibits a cross-temporal architecture of relations and constraints.
In the cases tested, networks appear in which information from different sources, roles, and times reconstructs, constrains, and completes other information as part of one system.
Gematria and numerical representation serve as deterministic measurement tools. The phenomenon under examination is the architecture.
RTM’s radical layer examines whether this architecture reflects a deeper principle: a global information history in which records, observer, memory, and earlier and later information can jointly participate in a coherent state.
The statistics and controls test the structure; the question of the physical mechanism behind it remains open for research.
The Central Statement
RTM presents a cross-temporal information architecture in which records, events, identities, memories, roles, choices, and access times are not measured merely as isolated data: they can reconstruct, constrain, and complete one another as part of a single information history; and in the radical version of the model, the information state from which that history is accessible may itself be part of the system that updates in order to maintain global coherence.