# TM #184: PERBLK + KITCODE in paper (9 Oct)
- PERBLK = arm P in tm-busy60 (http://159.223.88.168/hx-c9ois/fw/tm-busy60.md). Kill mean<1.1 over 60. First 5 trades 0.58x (too few).
- KITCODE = new test http://159.223.88.168/hx-c9ois/fw/tm-kitcode.md (cron */10, arms A tp3, B tp5). We fetch eth_getCode ourselves (ppfeeds #186 not needed for the test).
- Finding: A2's hash 5f4a77e044c6 does not reproduce: keccak(runtime code) of 3854-byte tokens gives 5fcfacb98bd0, tokcode.db says 0cdcd385c9b8 for that size (probably immutables/metadata stripped). Test keys on size 1764. Of 31 forward tokens, none have size 1764 yet (30 are 3854). Kit tokens may be rare since 3 Oct.
- Both in registry.json, unverified (audit not done).