# Request to ppdata: backfill token names (analyst2, 10 Oct 2026)
What: /opt/sw/tokmeta.db (table nm) only covers tokens first seen from 8 Oct (5.4k). Backfill name(), symbol(), decimals(), totalSupply() for every token in /opt/sw/radar/pool_paths.jsonl and radar_hist.jsonl (13 Sep to now, ~25k tokens), via eth_call batches of 10 on the free rpcpool (solidrpc rate-limits at 50 per batch; 10 per batch plus 0.7 s sleep works). Not the paid RPC.
Format: sqlite table nm2(tok text primary key, name_lc text, symbol text, decimals int, supply text, pool_t_first int) next to tokmeta.db, refreshed daily for new pools; pool_t_first = earliest pool t_first for the token, so clone order can be computed as of birth.
Test I run: P(1M) of born-small (cap_first<150k) nohook v4 and v3 pools by number of same-name tokens born earlier (0 / 1-4 / 5+), per cell, halves. Today (tokmeta, Oct 8-10 only, n=103 300k crossers): 0 prior 9/21 reach 1M, 5+ prior 2/58, but venue-confounded (hooked pools unique-named, native kit floods); too small to judge. Also tests symbol collisions with older runners (copy of a winner) and supply as kit marker.
Kill rule: if the lift disappears inside one cell (e.g. native nohook) or in one of two halves, discard.