docs(research): Ergon daily adjustment for GLotus via split covenant inputs

[nghiacc]

Explores the ALP author's hint for GLotus: replace felt +1-bit (×2 / 845 d) steps with smooth Ergon daily decay target ← floor(target · 99918/100000), splitting work across covenant inputs that cross-check each other's scripts.

Key findings (all measured with the repo's Spedn toolchain, not estimated):

• Shipped GLotus shape is 195 ops / 470 B — only 6 ops headroom, so even one computation step needs the split.
• One day-step on a 32-bit compact target compiles to 40 non-push ops (t − t·82/100000, double-and-add + OP_DIV, no OP_MUL needed).
• Multi-input splits the per-step cost, never the step count (days since last remint are unbounded) — so the design carries (tipDay, target) state forward with a per-remint cap K = 3, plus fee-only tick txs for stale batons.
• Concrete two-shard layout (compute + mint/PoW), cross-pinned via OP_TXINPUTCOUNT + OP_UTXOBYTECODE (~8 ops/side; hand-assembled since Spedn 5.0 can't emit introspection), with the soundness argument and unchanged cheat surfaces.
• Must still verify on chipnet: 64-bit script numbers and introspection acceptance on eCash (with a pure-31-bit fallback at ~45–50 ops/step).

Docs only — nothing live changes. Full write-up: docs/research/glotus-ergon-multiinput-daa.md.

  

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}