Rebuttal of the vendor's fallback correction

James O'Beirne replies to Coinkite's "extremely important correction" to Jack Mallers, held at coldcardwallet-2085541034243600805, which argued there was no weak-entropy fallback and that Yasmarang was MicroPython's general-purpose RNG reached through a link-time error. O'Beirne calls that hairsplitting and misdirection, says the device absolutely did fall back to weak entropy, and says the Yasmarang implementation placed in libngu was weaker than MicroPython's because it used hardcoded constants, linking a pinned commit and line range in switck/libngu ngu/random.c. He adds that he warned Coinkite about this issue and that at least one other person did four years before him. This is a direct exchange between the vendor and the auditor whose disclosure history the record already carries; the file he cites is itself held at libngu-random-c, so the comparison can be checked against the source rather than taken on assertion. The warning history and the characterisation of the correction are his own.