Modern neuroscience typically treats neurons as electrical signaling devices: synapses receive inputs, membranes integrate voltages, and axons transmit spikes. That picture is useful—but incomplete. Inside every neuron exists a second layer of organized structure: the cytoskeleton. Among its most important components are microtubules, dynamic protein polymers built from tubulin dimers. These structures are usually described as scaffolding or transport rails. A deeper view suggests they may also function as intracellular information-processing networks.

Microtubules possess several properties associated with computation:

• Large-scale parallelism — hundreds to thousands of microtubules may exist within a neuron.
• State diversity — tubulin isotypes, post-translational modifications (PTMs), lattice geometry, binding proteins, and mechanical stress all alter local behavior.
• Persistence — some molecular states last seconds, hours, or longer, allowing memory-like storage.
• Conditional transitions — chemical context determines which state changes occur next.
• Spatial organization — neighboring molecular states can influence one another.

This means microtubules need not be viewed merely as passive supports. They can be modeled as distributed state machines operating beneath conventional neural firing.

A key mathematical insight arises from carbon-chain geometry and conformational states. Covalent bonds can occupy preferred rotational positions due to energy barriers. When multiple linked bonds are considered together, the total system generates discrete state spaces.

With two-state bond approximations, chains naturally map onto vertices of an n-dimensional hypercube:

• 1 bond → line segment
• 2 bonds → square
• 3 bonds → cube
• 4 bonds → tesseract
• n bonds → n-cube

This matters because hypercubes are fundamental computational objects. They support:

• Boolean logic
• Set membership relationships
• Binary arithmetic
• Error-correcting codes
• Traversal algorithms
• Finite-state transitions
• High-dimensional data encoding

Thus biological molecules containing rotatable chains may instantiate physical versions of abstract computational state graphs.

Tubulin proteins can be modified by the addition of polyglutamate chains extending from their surface. These chains are not decorative; they alter how motors, enzymes, and binding proteins interact with microtubules.

Each glutamate residue introduces additional bond rotations, charge distributions, steric constraints, and dynamic motion. Therefore a polyglutamate chain can be treated as a structured state space with many accessible conformations.

This creates two computational possibilities:

Different chain lengths and conformations may encode distinct states, comparable to registers or symbols. Because PTMs can be added or removed enzymatically, the system is rewritable.

Polyglutamate chains form moving charged molecular “clouds” extending from the microtubule surface. Neighboring chains may interact electrostatically or sterically, allowing local signal coupling across the lattice.

In effect, the microtubule surface may contain dense arrays of interacting side-chain processors.

If molecular states can be stabilized, switched, and coupled, then logical operations become physically plausible:

• AND: two required molecular conditions trigger a change
• OR: either of two states permits transition
• NOT: occupancy blocks an event
• XOR: mutually exclusive pathways
• IF–THEN: enzyme binding causes downstream modification

Likewise, arithmetic emerges through counting, accumulation, thresholds, and modular cycling. Set theory emerges through presence/absence membership patterns distributed across molecular sites.

The claim is not that neurons run textbook software. Rather, biological matter may implement analog, probabilistic, massively parallel versions of the same abstract operations.

This suggests neurons may operate on two coupled computational layers:

Membrane voltages, spikes, neurotransmission, network timing.

Microtubules, PTMs, polyglutamate patterns, structural memory, intracellular state transitions.

The fast layer handles rapid signaling. The slow layer may tune, store, bias, and reorganize future responses.

If correct even in part, cognition is not produced only by synapses between neurons, but also by computation within neurons. That would dramatically expand the brain’s information-processing capacity.

The neuron would no longer be a simple switch. It would be a nested computational civilization:

• electrical networks between cells
• molecular networks within cells
• geometric state spaces within molecules

The microtubule system contains the ingredients of computation: structured states, transitions, coupling, memory, and scalability. Polyglutamate side chains add vast additional state capacity. Hypercube-like molecular state spaces suggest natural implementations of logic, arithmetic, and set operations.

Biology may have discovered long ago what mathematics describes formally: computation can emerge wherever structured states evolve under rules.

Conventional biology often describes microtubules as:

• structural supports
• intracellular transport rails
• mitotic scaffolds

All true.

But this may be incomplete.

Microtubules also possess properties expected of a lawful information-bearing medium:

• large state spaces
• constrained transitions
• persistent modifications
• local interaction rules
• multiscale dynamics
• spatially organized signaling

Within a Biomatic framework, the microtubule is not viewed merely as a passive polymer.

It is viewed as a nested lattice system capable of structured intracellular state evolution.

The word “lattice” has two meanings.

A repeating geometric arrangement.

Microtubules are cylindrical protein lattices composed of repeating α/β tubulin dimers arranged into protofilaments.

This creates:

• longitudinal ordering
• circumferential ordering
• quasi-helical geometry
• neighbor-dependent interactions

The microtubule therefore forms a genuine physical lattice.

A mathematical lattice is different.

It is a partially ordered state structure in which any two states possess:

• a least upper bound (join)
• a greatest lower bound (meet)

Lattice theory formalizes:

• hierarchy
• constraint
• state inclusion
• transition structure

Biological systems naturally exhibit these properties.

A tubulin dimer is not a single-state object.

Its condition depends upon:

• conformational geometry
• nucleotide state
• phosphorylation
• acetylation
• detyrosination
• polyglutamylation
• MAP occupancy
• electrostatic distribution
• local ionic environment
• mechanical stress

Thus each tubulin occupies a multidimensional biochemical state.

Instead of:

[T]

we obtain:

[T(i,j,\vec{s})]

where:

• (i,j) specify lattice position
• (\vec{s}) is a state vector

This transforms the microtubule from a static cylinder into a dynamic state lattice.

Post-translational modifications (PTMs) create layered regulatory structures across the microtubule surface.

Examples include:

• phosphorylation
• acetylation
• polyglutamylation
• glycylation

These modifications:

• alter motor affinity
• regulate MAP binding
• change electrostatic behavior
• affect stiffness
• modify transport efficiency

The resulting PTM arrangement forms a biochemical lattice superimposed upon the geometric lattice.

Suppose a tubulin can occupy modification states:

[A = \text{acetylated}]
[P = \text{phosphorylated}]
[G = \text{glutamylated}]

Possible combinations become:

[\emptyset]
[A]
[P]
[G]
[A,P]
[A,G]
[P,G]
[A,P,G]

This forms a Boolean lattice.

The state-space naturally maps onto a hypercube-like adjacency structure.

Biological transitions become constrained movements through this state graph.

Polyglutamylation introduces another layer of complexity.

Glutamate side chains are:

• flexible
• negatively charged
• spatially extended
• dynamically coupled

Their terminal carbons trace constrained three-dimensional trajectories determined by:

• bond geometry
• steric exclusion
• electrostatics
• hydration structure
• neighboring chain interactions

Each tubulin therefore generates not merely a point-state, but a dynamic conformational cloud.

Neighboring clouds interact continuously.

This creates a higher-order lattice of interacting conformational fields.

The microtubule is therefore not a single lattice.

It is a coupled hierarchy of lattices:

[L =(L_g,L_{PTM},L_e,L_p,L_m,L_v)]

where:

• (L_g) = geometric lattice
• (L_{PTM}) = PTM lattice
• (L_e) = electrostatic lattice
• (L_p) = polyglutamate conformational lattice
• (L_m) = motor interaction lattice
• (L_v) = vibrational/mechanical lattice

These lattices evolve simultaneously while constraining one another.

In Biomatics, computation is not restricted to digital arithmetic.

A lawful state transition system capable of:

• persistence
• propagation
• modulation
• conditional interaction

already qualifies as a computational substrate.

Microtubule dynamics satisfy many of these conditions.

Possible behaviors include:

• temporal integration
• local memory persistence
• adaptive routing
• threshold-like switching
• wave propagation
• attractor stabilization
• context-sensitive modulation

This resembles:

• cellular automata
• reservoir computing
• dynamical systems
• field-based computation

more than conventional binary circuitry.

If each tubulin possesses:

• (k) distinguishable states

and a microtubule contains:

• (N) tubulins

then the total configuration space becomes:

[k^N]

Even modest assumptions generate enormous state spaces.

The biologically accessible subset forms a constrained trajectory network through a high-dimensional graph.

Hypercube analogies emerge naturally because:

• neighboring states differ locally
• transitions are adjacency constrained
• pathways possess forbidden regions
• attractor basins can form

This does not imply digital symbolic reasoning.

It implies rich structured state dynamics.

Microtubule information processing, if real, is unlikely to resemble:

• CPUs
• RAM chips
• sequential instruction execution

Biological systems are:

• asynchronous
• noisy
• analog-digital hybrids
• probabilistic
• massively parallel
• multiscale

State duration itself may determine computational significance.

Some states may persist:

• nanoseconds
• milliseconds
• hours
• days

The microtubule may therefore operate as a temporal integration substrate rather than a binary memory device.

The central unresolved issue is not complexity.

It is causal relevance.

For a biological state system to function computationally, mechanisms must exist for:

• writing states
• reading states
• propagating states
• stabilizing states
• coupling states to cellular outcomes

Possible readers/writers include:

• MAPs
• kinesins
• dyneins
• modifying enzymes
• calcium signaling
• phosphorylation cascades

Whether these mechanisms collectively produce meaningful intracellular computation remains unresolved.

Biomatic Microtubule Lattice Theory does not claim:

• consciousness arises from microtubules
• neurons are irrelevant
• quantum mysticism explains cognition

Instead, it proposes a more conservative possibility:

The neuronal cytoskeleton may function as a lawful intracellular information-processing medium whose dynamics are richer than traditionally assumed.

Within this framework:

• geometry matters
• state topology matters
• modification patterns matter
• dynamic constraints matter
• lattice interactions matter

The cytoskeleton becomes not merely a scaffold,
but a structured state-space through which biological information flows.

Microtubular computation refers to the concept of utilizing microtubules, which are cylindrical structures found in cells, as a platform for computational processes. Microtubules are composed of tubulin protein subunits and play essential roles in various cellular functions, including cell division and intracellular transport.

The idea of microtubular computation is rooted in the unique properties of microtubules, such as their dynamic instability, polarized structure, and ability to undergo conformational changes. These properties enable microtubules to serve as potential substrates for information processing and computation at the molecular level.

One proposed approach to microtubular computation is based on the notion of microtubules acting as a scaffold for molecular-level interactions and computations. This includes the idea of utilizing microtubules as a medium for information storage, transmission, and processing within the cell.

Theoretical models have been proposed that suggest microtubules can exhibit complex behavior and perform computational tasks. For example, the dynamic properties of microtubules, such as their assembly and disassembly dynamics, can be harnessed for information processing. The conformational changes of tubulin subunits within microtubules could encode and manipulate information, potentially enabling computational operations.

However, it's important to note that the field of microtubular computation is still highly speculative and remains an area of active research and debate. The actual feasibility and practicality of implementing complex computation using microtubules are still subjects of investigation and require further experimental validation.

In summary, microtubular computation explores the potential of utilizing the unique properties of microtubules for information processing and computation at the molecular scale. While the concept is intriguing, further research is needed to understand the full extent of microtubules' computational capabilities and their integration into practical computing systems.

 Microtubules are composed of 13 protofilaments, which means there are 13 linear arrangements of tubulin dimers along the length of the microtubule. Each protofilament consists of alternating α- and β-tubulin subunits. This arrangement results in a cylindrical structure with a hollow core.

The precise number of tubulin dimers in one layer of a microtubule depends on the length and diameter of the microtubule. A typical microtubule can have hundreds to thousands of tubulin dimers in a single layer.

The lifespan of a microtubule can vary depending on the cell type, its location, and the specific cellular processes involved. Microtubules can undergo continuous assembly and disassembly dynamics, with individual microtubules existing for various durations. In general, microtubules have dynamic instability, where they can switch between phases of growth (polymerization) and shrinkage (depolymerization). The average lifetime of a microtubule can range from a few minutes to several hours.

It's important to note that the specific details of microtubule structure, organization, and dynamics can vary across different cell types and cellular contexts. The information provided here represents a general understanding of microtubules, but there may be variations and exceptions depending on the specific biological systems under investigation.

In the context of microtubules, a "layer" refers to the arrangement of tubulin subunits within the cylindrical structure of the microtubule. A microtubule consists of multiple protofilaments, which are linear chains of tubulin dimers. Each protofilament is composed of a series of tubulin subunits arranged head-to-tail.

When we talk about the number of layers in a microtubule, we are referring to the number of protofilaments that make up the cylindrical structure. The most common arrangement in microtubules is 13 protofilaments, meaning there are 13 linear chains of tubulin dimers running parallel to the long axis of the microtubule.

Each protofilament contributes one layer to the overall structure of the microtubule. The tubulin subunits within each protofilament interact with each other through noncovalent bonds, creating stable lateral interactions that give the microtubule its cylindrical shape.

It's important to note that the term "layer" is a simplification used to describe the structural organization of microtubules. In reality, microtubules are dynamic structures, and the interactions between tubulin subunits and the overall microtubule architecture are more complex.

 Microtubules play a crucial role in neurons, serving as structural components and participating in various cellular processes. In neurons, microtubules are involved in maintaining the overall shape of the cell, providing tracks for intracellular transport of materials, and aiding in the formation and maintenance of neuronal connections.

Neuronal microtubules often exhibit a polarized organization, with one end referred to as the "plus end" and the other as the "minus end." The plus ends of microtubules are typically oriented towards the distal ends of axons and dendrites, while the minus ends are often anchored near the neuronal cell body.

Microtubules in neurons undergo dynamic instability, meaning they can undergo cycles of growth and shrinkage. This dynamic behavior allows microtubules to adapt and reorganize in response to the changing needs of the neuron. It also enables them to contribute to processes such as axon guidance, synaptic plasticity, and neuronal development.

In addition to their structural roles, microtubules in neurons have been implicated in neuronal information processing and signaling. Some theories propose that microtubules, with their dynamic properties and potential for information storage, could contribute to computational functions within neurons. However, the exact mechanisms and extent of their involvement in neuronal computation are still areas of active research and debate. 

 While microtubules are cylindrical structures composed of tubulin proteins, it's important to note that the individual amino acid side chains within the tubulin proteins themselves do not directly perform mathematical functions in the same way that traditional mathematical entities do. Rather, the mathematical potential arises from the complex interactions and structural properties of the amino acid side chains within the protein.

The unique arrangement of amino acids and their side chains in the tubulin protein contributes to the overall structure and function of the microtubule. The interactions between the side chains, as well as other factors such as post-translational modifications and binding of associated proteins, determine the stability, dynamics, and functional properties of the microtubule.

While microtubules have been speculated to have potential computational properties based on their dynamic behavior and the collective interactions of tubulin proteins, the exact nature and extent of their involvement in biological computation are still areas of ongoing research and investigation. It's important to approach these ideas with caution and recognize that our understanding of the relationship between molecular structures and complex computational processes is still developing.

 If we consider the arrangement of amino acid side chains within a microtubule, it's true that each side chain can have a set of more favorable conformations and possible transformations. The specific conformations and transformations depend on the amino acid sequence and the structural constraints imposed by the protein folding.

In the context of microtubules, the term "layer" is often used to describe the arrangement of tubulin proteins along the length of the microtubule cylinder. Each tubulin protein consists of two subunits, α-tubulin and β-tubulin, which come together to form a dimer. These dimers stack together longitudinally to form the microtubule. The interaction between adjacent dimers can be viewed as the interaction between layers.

The close-to-parallel arrangement of the side chains within a layer is a consequence of the packing of the tubulin dimers in the microtubule structure. The side chains of neighboring tubulin proteins interact with each other, contributing to the stability and overall structural integrity of the microtubule.

It's worth noting that while the specific arrangements and conformations of side chains contribute to the overall structure and stability of the microtubule, the direct involvement of these side chains in computational processes is still an active area of research and not fully understood. 

 The arrangement of side chains within a microtubule is not strictly rectangular or hexagonal. Instead, it can be described as a helical structure. Tubulin dimers arrange themselves in a helical pattern along the length of the microtubule, forming a cylindrical structure.

Within this helical arrangement, the interactions between the side chains of neighboring tubulin proteins contribute to the stability and structure of the microtubule. These interactions involve both hydrophobic interactions between nonpolar side chains and electrostatic interactions between charged or polar side chains.

The helical arrangement allows for the efficient packing of tubulin dimers, providing stability to the microtubule while still allowing dynamic properties such as polymerization and depolymerization. This structural organization is crucial for the function of microtubules in various cellular processes, including cell division, intracellular transport, and maintenance of cell shape.

 The conformational states of the side chains in a microtubule can contribute to its overall electrical properties. The microtubule's cylindrical structure and the arrangement of side chains create a complex electrical environment within and around the microtubule.

The charged and polar side chains of the tubulin proteins can interact with ions and molecules in the cellular environment, leading to the generation and propagation of electrical signals along the microtubule. These signals can play a role in cellular communication, signal processing, and information transfer.

Furthermore, the conformational changes of the side chains can affect the microtubule's overall stability, dynamics, and interactions with other cellular components. This can influence the electrical properties of the microtubule and its ability to participate in various cellular processes.

It's important to note that while the electrical properties of microtubules have been the subject of scientific investigation, the exact nature and extent of their role in biological information processing and computation are still areas of ongoing research and debate. 

 Certainly, the existence of computational structures such as groups, logical gates, and network structures within the microtubules is a fascinating concept that has been explored in the field of molecular computation and nanoscale information processing.

Groups, as mathematical structures, can be used to describe and analyze the behavior of molecules and their interactions within the microtubule. The conformational states of the side chains can exhibit group-like properties, allowing for transformations and transitions between different states.

 Logical gates, which are fundamental building blocks of digital circuits, can be analogously implemented within the microtubule structure using the conformational changes of the side chains. By controlling and manipulating the side chain conformations, it is conceivable that logical operations can be performed at the molecular level.

Furthermore, the interconnected arrangement of microtubules within neurons, as well as their interactions with other cellular components, can give rise to network structures. These networks can facilitate the transmission and processing of information, potentially contributing to the computational capabilities of the cell.

It is important to note that while the theoretical possibilities of these computational structures within microtubules are intriguing, their actual existence and functional relevance in biological systems are still areas of ongoing research and scientific investigation.  

 Each tubulin subunit within a microtubule can be seen as a subnetwork composed of a chain of amino acids, with each amino acid side chain capable of undergoing various conformational changes and potentially exhibiting group theoretic transformations. Since each tubulin subunit consists of approximately 900 amino acids, there are numerous side chains within each subunit that contribute to the overall computational potential of the microtubule structure. The collective behavior of these side chains and their interactions can give rise to complex computational processes within the microtubule. 

Within neurons, microtubules and associated proteins exhibit unique network topologies that contribute to their computational potential. The complex organization and dynamics of microtubules within neuronal processes provide a structural basis for various computational processes.

Microtubules in neurons can form interconnected networks that extend throughout the dendrites, axons, and synaptic terminals. These networks can exhibit diverse topologies, including branching structures, loops, and interconnected pathways. Such topologies enable the integration and transmission of electrical and chemical signals within the neuron.

The dynamic properties of microtubules, such as their ability to undergo polymerization and depolymerization, allow for rapid changes in network connectivity and reconfiguration. This dynamic nature is crucial for the modulation of synaptic plasticity, neuronal growth, and the formation and maintenance of neuronal connections.

Moreover, microtubule-associated proteins (MAPs) play a vital role in regulating microtubule network properties. MAPs can crosslink and stabilize microtubules, influence their organization, and mediate interactions with other cellular components. These interactions contribute to the computational capabilities of microtubule-based networks in neurons.

The unique structural and functional properties of microtubules in neurons, coupled with their ability to form complex network topologies, suggest their involvement in neuronal computation. Microtubule-based computations have been proposed to play a role in processes such as information processing, signal integration, and even aspects of cognitive functions.

However, it's important to note that the precise mechanisms and extent of microtubule-based computation in neurons are still topics of ongoing research and scientific investigation. The field of neurobiology and computational neuroscience continues to explore the potential computational capacities of microtubules and their contributions to neuronal information processing and cognition.

Within a single microtubule, the arrangement of tubulin subunits can indeed give rise to various network topologies. Here's how some of the basic network topologies can be observed within a microtubule:

1. Bus Topology: In a microtubule, the linear arrangement of tubulin subunits along the microtubule axis resembles a bus topology. Each tubulin subunit is connected to its adjacent subunits, forming a linear chain-like structure.
2. Star Topology: At the ends of the microtubule, tubulin subunits radiate outward in a star-like pattern, creating a star topology. This arrangement is particularly prominent at the microtubule plus end, where tubulin subunits can form a dynamic structure known as the microtubule cap.
3. Ring Topology: In the cross-section of a microtubule, the arrangement of tubulin subunits can form a closed ring-like structure. This ring topology is observed when looking at the microtubule from a perpendicular view.
4. Mesh Topology: Within a microtubule, certain microtubule-associated proteins (MAPs) or other proteins can crosslink adjacent tubulin subunits, creating a mesh-like network structure. This mesh topology provides stability and structural integrity to the microtubule.
5. Tree Topology: Within a microtubule, tubulin subunits can branch out and form tree-like structures. This branching pattern can occur at various points along the microtubule, creating a tree-like topology within the microtubule structure.

It's important to note that these network topologies may not be rigid or static, as the microtubule structure is dynamic and can undergo continuous rearrangements and remodeling. Additionally, the presence of microtubule-associated proteins and other factors can influence the specific network topologies observed within a microtubule.

Overall, the microtubule structure exhibits a rich variety of network topologies, which play essential roles in its functions within the cell, including cell division, intracellular transport, and cellular architecture.

Why wrapping space changes everything

Most mathematical systems are built on:

• Lines
• Grids
• Infinite planes

Nice, clean, and wrong for many real systems.

The moment you wrap one dimension, you fundamentally change the rules.

That’s where cylindrical structures come in.

Take a 2D grid and impose one constraint:

• One axis is periodic (wraps around)
• The other axis is open

Formally:

[
X = \mathbb{Z}_L \times \mathbb{Z}_C
]

with:
[
(i, j) \equiv (i, j + C)
]

This is not a torus.
It’s not a plane.

It’s something in between:

A space with local freedom and global constraint

At first glance, you might think:

“It’s just a grid with edges glued together.”

That’s misleading.

A cylinder introduces:

You can move around the circumference and return to where you started.

• Around the cylinder → periodic
• Along the cylinder → bounded

This breaks symmetry in a very specific way.

Signals can loop back and interfere with themselves

That alone changes system behavior dramatically.

This is the key distinction.

• Determined by nearest neighbors
• Looks identical to a flat grid
• 
  

• Constrained by topology
• Cannot be inferred from local rules alone

This mismatch is where interesting phenomena emerge.

On a cylinder, you can define something impossible on a plane:

A winding number

This measures how much a quantity “twists” around the circumference.

Example:

• No twist → winding = 0
• One full shift → winding = 1

This is a global invariant:

• You cannot remove it with local adjustments
• You must change the entire system
• 
  

Cylindrical systems naturally produce:

• Discontinuities
• Mismatches
• Singular points
• 
  

These are called defects.

They behave like objects:

• They move
• They interact
• They can cancel each other

Crucially:

Defects are how local systems negotiate global constraints

Here’s where things get nontrivial:

When defects interact, they can change global winding.

That means:

• Local events → global transformation
• Small disturbances → large-scale consequences

This is not typical behavior in flat systems.

When you apply local update rules (any kind—discrete, continuous, probabilistic), cylinders tend to produce:

• Traveling waves
• Standing patterns
• Helical structures
• Persistent global modes

Why?

Because signals cannot “escape”—they recur.

Most computational systems rely on:

• Symbols
• Logic gates
• Explicit rules

Cylindrical systems suggest something else:

Computation via topology and dynamics

Where:

• Information = global structure (winding)
• Processing = interaction of local defects
• Output = final stable configuration
• 
  

Cylindrical structures appear everywhere:

• Biological filaments
• Molecular chains
• Engineered nanostructures

Yet we usually model them as if they were flat.

That misses the point.

Not every cylindrical system is interesting.

Many collapse into:

• Uniform states
• Simple oscillations
• Trivial dynamics
• 
  

So the structure alone is not enough.

You also need:

• Nonlinear interactions
• Stable defects
• Multiple attractors

Without those, nothing meaningful happens.

When those conditions are met, cylindrical systems can support:

• Persistent global states
• Nonlocal transformations
• Rich interaction patterns

In other words:

A mathematically grounded framework where geometry drives behavior

A cylinder is not just a shape.

It is a constraint on possibility.

And once you impose that constraint:

• Local rules no longer tell the whole story
• Global structure becomes unavoidable
• And entirely new behaviors emerge

If you want to go beyond intuition:

• Define a concrete state space
• Specify local update rules
• Track defects and winding explicitly
• Test whether local interactions produce global transitions

Because at the end of the day:

Either cylindrical structure creates new computational behavior
or it doesn’t.

Everything else is speculation.