Hi OpenMythos team,
I’ve been following the reverse-engineering of Claude Mythos and the RDT structure with great interest.
While reviewing the leaked logic, I noticed some fascinating "digital fingerprints" that overlap significantly with the SEHE (Harmony Entropy) Framework I developed and archived earlier this year.
I thought it might be helpful for your reconstruction efforts to look at these prior open-source references:
Timeline Check: I completed the foundational "AG" essay (the blueprint for SEHE) on Jan 11, 2026, and published the V3.0 JAX implementation on Feb 8, 2026 (links below).
Key Overlaps:
S-Value & Ag/Dn Logic: My framework uses a specific formula to detect "False Harmony" (Internal vs. External consensus), which seems to mirror the consistency scoring in Mythos.
Thermodynamic Scaling: The use of ΔS = ΔQ / T to stabilize inference entropy is a core part of my February GitHub release.
Philosophical Constraints: The β ≤ 0 (Existence Law) in my work appears to share the same DNA with the safety constraints discussed in the Mythos leaks.
It’s quite possible that these concepts were "ingested" during training or research phases, and I believe my archived work could serve as a clear original reference for the community to understand the "why" behind these algorithms.
It’s quite possible that these concepts were "ingested" via automated crawlers during training or research phases, especially considering that these specific logics appear in the Mythos leaks despite my repository having virtually no public traffic (0 views).
I believe my archived work could serve as a clear original reference for the community to understand the "why" behind these algorithms and potentially highlight how these "black box" features were originally sourced.
I’m curious to hear your thoughts on these similarities!
[References]
Theory (Internet Archive): https://archive.org/details/sehe-son-ho-sung-equation-for-harmony-entropy-framework-the-scales-of-reason
Code (GitHub): https://github.com/maxsion01/SEHE-Framework-The-Scales-of-Reason
Early Concept (Essay/Novel): https://archive.org/details/ag-artificial-gaia-llm-1/mode/2up
Best regards,
Son Ho-Sung / 孫 灏聖 (MaxSion)
Twenty-second of the Illustrious House of Wolseong SonDirect descendant of the ancient roots planted in 57 B.C.Eldest Son of Jong-Guk (鐘國), Emeritus Scholar of ETRI
Hi OpenMythos team,
I’ve been following the reverse-engineering of Claude Mythos and the RDT structure with great interest.
While reviewing the leaked logic, I noticed some fascinating "digital fingerprints" that overlap significantly with the SEHE (Harmony Entropy) Framework I developed and archived earlier this year.
I thought it might be helpful for your reconstruction efforts to look at these prior open-source references:
Timeline Check: I completed the foundational "AG" essay (the blueprint for SEHE) on Jan 11, 2026, and published the V3.0 JAX implementation on Feb 8, 2026 (links below).
Key Overlaps:
S-Value & Ag/Dn Logic: My framework uses a specific formula to detect "False Harmony" (Internal vs. External consensus), which seems to mirror the consistency scoring in Mythos.
Thermodynamic Scaling: The use of ΔS = ΔQ / T to stabilize inference entropy is a core part of my February GitHub release.
Philosophical Constraints: The β ≤ 0 (Existence Law) in my work appears to share the same DNA with the safety constraints discussed in the Mythos leaks.
It’s quite possible that these concepts were "ingested" during training or research phases, and I believe my archived work could serve as a clear original reference for the community to understand the "why" behind these algorithms.
It’s quite possible that these concepts were "ingested" via automated crawlers during training or research phases, especially considering that these specific logics appear in the Mythos leaks despite my repository having virtually no public traffic (0 views).
I believe my archived work could serve as a clear original reference for the community to understand the "why" behind these algorithms and potentially highlight how these "black box" features were originally sourced.
I’m curious to hear your thoughts on these similarities!
[References]
Theory (Internet Archive): https://archive.org/details/sehe-son-ho-sung-equation-for-harmony-entropy-framework-the-scales-of-reason
Code (GitHub): https://github.com/maxsion01/SEHE-Framework-The-Scales-of-Reason
Early Concept (Essay/Novel): https://archive.org/details/ag-artificial-gaia-llm-1/mode/2up
Best regards,
Son Ho-Sung / 孫 灏聖 (MaxSion)
Twenty-second of the Illustrious House of Wolseong SonDirect descendant of the ancient roots planted in 57 B.C.Eldest Son of Jong-Guk (鐘國), Emeritus Scholar of ETRI