Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 1 / 14
Keep information distinct while joining it together
Micah's June 2022 Git-recorded note separates binding from the distinctions needed for information. Bach and Sorensen's reviewed MCH version likewise coordinates differentiated models and builds represented structure through contrast; the shared function does not equate physical synchrony with model consistency.
Micah / SAN · 2022-06-08
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-06-08 · Git source record
the problem is that synchronization or coherence is non-informative, its a solution for binding, for convergence, coupling of patterns, but not for distinction making, decoherence, deconvergence, decoupling of of ideas & concepts
Exact excerpt; original wording retained.
Joining signals is not enough: their differences must remain readable. SAN describes a receiving oscillator detecting changes against its own ongoing pattern.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Keep distinguishable patterns available within coordinated activity; compare or select contributions rather than erase all differences through undifferentiated binding.
Difference kept visible: No June definition of semantic constraint violation or MCH implementation of this oscillator mechanism; no identity between physical coherence and representational consistency. Neither document proves the neuroscience.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 2 / 14
Keep information distinct while joining it together
Micah's June 2022 Git-recorded note argues that coordination must still permit distinctions. CIMC's reviewed 2026 whitepaper coordinates distinct partial models through conflict-sensitive adjustment. This is a partial functional correspondence, not identity of physical and representational coherence.
Micah / SAN · 2022-06-08
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-06-08 · Git source record
the problem is that synchronization or coherence is non-informative, its a solution for binding, for convergence, coupling of patterns, but not for distinction making, decoherence, deconvergence, decoupling of of ideas & concepts
Exact excerpt; original wording retained.
Joining signals is not enough: their differences must remain readable. SAN describes a receiving oscillator detecting changes against its own ongoing pattern.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Coordinate differentiated components while retaining information-bearing distinctions, using selective interaction rather than merely pooling signals.
Difference kept visible: Physical synchrony, oscillator baseline and semantic model consistency remain different variables. No demonstrated equivalence between June phase deviations and requests/confirmations or constraint reduction.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 3 / 14
Check the brain’s own representations for disagreement
Micah's July 2022 Git-recorded discussion proposes checking internal representations by sampling and cross-referencing them to notice inconsistency. MCH's reviewed version likewise makes conflicts between partial models a focus of conscious attention; the common subchain is comparison and discrepancy detection, not an equation of rarity with contradiction.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
Is there a way for a natural brain to check its own results, to check for pattern representation accuracy. one way to do that would be to sort of do to take multiple samples like sort of like with with and cross reference them, it's like, okay well I'm gonna attack multiple times and if something is true then it's gonna be true.
Exact excerpt; original wording retained.
SAN proposes taking and cross-checking samples of the brain’s own representations. Disagreement can then be noticed instead of simply passed along.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Maintain or obtain multiple representations -> compare/cross-reference them -> notice inconsistent content -> bring the discrepancy into a process that organizes the represented scene.
Difference kept visible: Rarity != inconsistency; noisy deviations may fail the recognition threshold. July does not provide MCH's explicit global constraint objective, conductor architecture or specific repair algorithm.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 4 / 14
Check the brain’s own representations for disagreement
Micah's July 2022 Git-recorded proposal cross-checks representations and notices inconsistency. CIMC's reviewed 2026 whitepaper likewise detects conflicts among partial models before coordinating them. The shared checking operation survives different implementations; rarity alone is not the match.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
Is there a way for a natural brain to check its own results, to check for pattern representation accuracy. one way to do that would be to sort of do to take multiple samples like sort of like with with and cross reference them, it's like, okay well I'm gonna attack multiple times and if something is true then it's gonna be true.
Exact excerpt; original wording retained.
SAN proposes taking and cross-checking samples of the brain’s own representations. Disagreement can then be noticed instead of simply passed along.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Compare information from multiple representations/modalities -> detect disagreement -> recognize the discrepancy as relevant to the system's constructed account.
Difference kept visible: Global optimization, simultaneous model constraints, suppression of unresolved content and distributed negotiation are target-specific elaborations not recovered as a single July algorithm.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 5 / 14
Build an internal world, then register it again
Micah's September 2022 Git-recorded account builds experienced self and world from contrasts in a continuing field and makes rendered patterns available to the brain again. MCH's reviewed Genesis passage also builds represented space, objects and self through differences, alongside reflexive perception. This supports a shared construction narrative, not the same neural mechanism.
Micah / SAN · 2022-09-17
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-09-17 · Git source record
The brain is effectively doing computational rendering & then reading it’s own output pattern, from neuronal outputs, as that pattern is looped or played back to the neural network of the brain which detects it, with neuronal inputs, and re-renders (outputs it again) with new sensor input data.
Exact excerpt; original wording retained.
SAN proposes that differences in ongoing activity form a represented self and world. The brain receives its own rendered patterns again and updates them with new input.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: An initially undifferentiated representational field/substrate gains contrasts -> those distinctions organize an internally constructed world and self -> the system registers its own represented/perceptual activity.
Difference kept visible: No target endorsement of specific tonic frequencies, anatomical assignments, phase-wave tomography or excitatory/inhibitory implementation. Re-reading outputs alone is not proof of MCH's synchronous second-order phenomenology, and the sources remain hypotheses.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 6 / 14
Build an internal world, then register it again
Micah's September 2022 Git-recorded account has neural activity render self/world patterns and receive its own output again. CIMC's reviewed whitepaper similarly proposes self-organization of the substrate's own activity and reflexive perceptual representation. The common subchain is internal construction and re-entry, with its stricter perceptual constraints left separate.
Micah / SAN · 2022-09-17
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-09-17 · Git source record
The brain is effectively doing computational rendering & then reading it’s own output pattern, from neuronal outputs, as that pattern is looped or played back to the neural network of the brain which detects it, with neuronal inputs, and re-renders (outputs it again) with new sensor input data.
Exact excerpt; original wording retained.
SAN proposes that differences in ongoing activity form a represented self and world. The brain receives its own rendered patterns again and updates them with new input.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Operate on the substrate's own activity -> organize differentiated representations of self/world -> make the perceptual process internally available -> iteratively maintain organized experience.
Difference kept visible: No whitepaper-specific Genesis contrast-to-space narrative equivalent to MCH20, and no stated tonic baseline/burst/inhibition mechanism. Perceptual same-level encoding and epistemic transparency stay unproved in this SAN passage.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 7 / 14
Look inside an artificial mind, not only at its answers
Micah's July 2022 Git-recorded discussion calls for criteria grounded in what artificial neural networks do internally, distinct from intelligence alone. MCH's reviewed version also makes internal organization central to consciousness testing. Its stronger argument against a behavior-only test is not established by this July passage.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
We need a guidebook for when an artificial neural network becomes conscious.
Exact excerpt; original wording retained.
SAN connects the search for artificial consciousness to what neural systems do internally. It distinguishes this question from intelligence alone and connects it to ethical treatment.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Use a theory of internal organization to identify candidate artificial consciousness, distinguish it from intelligent output alone, and guide what is constructed or assessed.
Difference kept visible: No validated July consciousness diagnostic, no complete necessary-and-sufficient test, no demonstrated exclusion of all behavior-only methods; MCH also frames its tests as a research program with failure possibilities.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 8 / 14
Look inside an artificial mind, not only at its answers
Micah's July 2022 Git-recorded proposal connects artificial-consciousness criteria to internal neural operations and ethical treatment. CIMC's reviewed whitepaper also couples construction with analysis of internal organization and alternative explanations. The shared research method does not establish a validated test.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
We need a guidebook for when an artificial neural network becomes conscious.
Exact excerpt; original wording retained.
SAN connects the search for artificial consciousness to what neural systems do internally. It distinguishes this question from intelligence alone and connects it to ethical treatment.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Ground artificial-consciousness criteria in a proposed internal mechanism -> examine candidate networks or construct test systems -> assess implications for their treatment.
Difference kept visible: No tested discriminator, converged definition of consciousness or validated internal signature in July. CIMC itself says neither behavioral performance nor weights alone can directly identify consciousness.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 9 / 14
An artificial mind develops; it is not just a snapshot
Micah's November 2022 Git-recorded proposal distinguishes immediate experiential states from longer structural stages and applies both to artificial minds. MCH's reviewed Genesis account likewise makes developmental organization important to constructing self/world models. The shared developmental framing does not equate their mechanisms.
Micah / SAN · 2022-11-28
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-11-28 · Git source record
It's a concept that means we, as robots, can have computed short term state experiences, with synaptic phase changes, and long term stage changes, with (real or simulated) synaptic growth or decay which takes longer to happen.
Exact excerpt; original wording retained.
SAN distinguishes short-lived experiences from slower changes in neural organization. It applies that distinction to the development of artificial minds.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Treat an artificial mind as something whose internal organization develops over time, not solely as a snapshot of present performance; design learning conditions that can alter that organization.
Difference kept visible: States/stages is explicitly attributed to AQAL or Spiral Dynamics (11). Synaptic phase, growth/decay thresholds and implementation hardware are not MCH's developmental mechanism; no backdating this mature wording to June or to an older heading.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 10 / 14
An artificial mind develops; it is not just a snapshot
Micah's November 2022 Git-recorded proposal connects long-term neural changes to threshold-like developmental stages in artificial minds. CIMC's reviewed whitepaper asks whether developmental trajectories show predicted transitions alongside new internal organization. This is a shared developmental test direction, not an equation of synaptic phase with developmental phase transitions.
Micah / SAN · 2022-11-28
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-11-28 · Git source record
Imagine that Synaptic Connections via Growth LTP & Decay LTD, over a long term, results in eventual threshold changes that we can identify as Stages, or major changes in live perspective that only happen after a very long time, or only happen after a very big life event.
Exact excerpt; original wording retained.
SAN distinguishes short-lived experiences from slower changes in neural organization. It applies that distinction to the development of artificial minds.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Observe an artificial system's developmental trajectory for changes in internal organization, including threshold-like transitions associated with new capacities.
Difference kept visible: CIMC's phase transitions are not specified as synaptic phase shifts or LTP/LTD, and November does not establish the target's coherence-plus-metarepresentation transition. General states/stages is acknowledged background from AQAL/Spiral Dynamics.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 11 / 14
Perception can include the act of perceiving
Micah's July 2022 Git-recorded discussion explicitly locates perceiving perception in neural operations, alongside internally rendered patterns received again. MCH's reviewed version also characterizes consciousness through perception that perception occurs. The partial reflexive correspondence does not establish identical timing or representational structure.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
the one process in the machine is always seeking to know, it's always seeking to know is perceiving perception.
Exact excerpt; original wording retained.
SAN explicitly discusses perceiving perception as a neural process. Its account includes internally rendered patterns being received again.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: A neural/computational process registers its own perceptual activity, with internal representations available to further perception rather than an unexplained outside viewer.
Difference kept visible: A following receiver alone could process content without representing that perception occurs. July does not establish MCH's immediate non-inferential synchronous form, and its self-as-direction (187) is not MCH's full observing-self model.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 12 / 14
Perception can include the act of perceiving
Micah's July 2022 Git-recorded proposal explicitly treats perceiving perception as neural activity. CIMC's reviewed whitepaper likewise makes reflexive perception central, while adding same-level encoding and transparency constraints not established by the July receiver description.
Micah / SAN · 2022-07-06
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-07-06 · Git source record
the one process in the machine is always seeking to know, it's always seeking to know is perceiving perception.
Exact excerpt; original wording retained.
SAN explicitly discusses perceiving perception as a neural process. Its account includes internally rendered patterns being received again.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: The perceptual process becomes internally available as part of what the system registers; perception of perception is proposed as a function of the substrate's own activity.
Difference kept visible: Same-level metarepresentation, non-inferential synchrony, epistemic transparency and global-coherence linkage remain unresolved. The target itself says its precise format and constraints are still under development.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · MCH · 13 / 14
Choose the detail a simulation needs to preserve
Micah's November 2022 Git-recorded implementation proposal models individual dendrites, synapses and their changing dynamics. MCH's reviewed version likewise raises the possibility that simulating consciousness needs more cellular detail than simplified neuron-to-neuron messages. Neither establishes the required resolution experimentally.
Micah / SAN · 2022-11-28
Git-recorded wording
→Joscha Bach and coauthors · 2026-01-15
MCH · selected version
Micah Blumberg / SAN
2022-11-28 · Git source record
The essential physics process could be simulated in just matrix calculations on a TPU, with each dendrite, each neuron, each synapse, and each cortical column being unique or random in it's original configuration
Exact excerpt; original wording retained.
SAN proposes modeling dendrites, synapses, neurons and larger assemblies with changing spatial and temporal responses.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: A simulation meant to realize consciousness may need to retain computational detail below a simplified whole-neuron unit; preserve relevant component organization rather than assume a coarse network is sufficient.
Difference kept visible: Dendrites and synapses are sub-neural structures, not evidence that every intracellular biochemical process must be simulated. MCH makes a possibility statement, not a proven requirement; neither side demonstrates sufficiency or feasibility.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →
Micah Blumberg / SAN vs Joscha Bach and coauthors · CIMC · 14 / 14
Local interactions can build higher-order organization
Micah's November 2022 Git-recorded proposal links context-sensitive local neural components and larger assemblies in a simulated mind. CIMC's reviewed whitepaper also studies how local interactions produce higher-order organization. Its cited pages do not establish the stronger claim that simulation must extend inside neurons.
Micah / SAN · 2022-11-28
Git-recorded wording
→Joscha Bach and coauthors · 2026-04-27
CIMC · selected version
Micah Blumberg / SAN
2022-11-28 · Git source record
The essential physics process could be simulated in just matrix calculations on a TPU, with each dendrite, each neuron, each synapse, and each cortical column being unique or random in it's original configuration
Exact excerpt; original wording retained.
SAN proposes modeling dendrites, synapses, neurons and larger assemblies with changing spatial and temporal responses.
Read the dated source and contextWhat matches, what differs, and how it was checked
Shared operation: Local interacting computational units, arranged across scales, can be studied or simulated as the basis for higher-order organization.
Difference kept visible: CIMC18-19 neither requires nor explicitly proposes simulating inside a neuron. Processing units, biological cells and agent communities are comparison scales, not evidence of subcellular fidelity. Retain the stronger route open, not established.
The 2022 dates are verified Git source-record dates. Current public access to the exact bytes was checked; first public availability on those dates is not independently established.
All three comparison lenses and the separate longitudinal lens →