what happened on the btx network this week, and what we shipped.

on 23 september the btx chain split at block 227,313. one branch starts with a block whose proof of work doesn't check out: its header claims a matmul result that meets the target, but recomputed on cpu and gpu it doesn't. the btx developers confirmed it. every node that actually checks blocks rejects that branch. it still grew, because some miners kept building on it and some mirrors followed signers who signed it without checking.

the valid chain is the one where block 227,313 is d5f0e92f. btx-pool mines on it, and btxscan and the wallets read from it.

for easyNode users this hit one group: nodes that follow signatures instead of checking blocks themselves (M5 macs, PCs without an nvidia card, native windows). they refuse the bad branch, but none of the keys they trusted was signing the valid one, so they sat at 227,312.

easyNode 0.6.30 fixes that. it trusts the one key that signs the valid chain and connects to a node that holds its signatures. before release we tested it on a stuck node: 219,000 to the tip in 34 minutes, on the right chain.

the app updates itself, or grab it at easybtx.com/node

next up: more independent signers, so no single machine carries the network's mirrors.