The 5-paragraph answer: base-3 died with Chrysalis — binary everywhere, the doof is the homage · content/ternary-note.md

Ternary, in Short

Date: 2026-09-12 · The 5-paragraph answer to "ternary systems?" — the full technical saga (Curl-P, Troika, Setun, qutrits, with sources) is Ternary Systems.

1. What it was. IOTA's original design ran on base-3: data lived in "trits" (three-state digits, grouped into "trytes" of 27), hashed with a custom 243-trit sponge called Curl-P, on the theory that base 3 is the most economical integer radix (the 1850 Steiner math: base 3 sits within 0.46% of the theoretical optimum e, vs binary's 5.7% penalty). It was also a nod to the Setun, the Soviet ternary computer of 1958 — the only one ever built in series.

2. Why it died. Three compounding reasons. The radix economy was theoretical: real CMOS transistors are binary, and no ternary hardware ecosystem ever materialized — trits were emulated on binary chips, paying the 5.7% penalty in practice while gaining none of it. The custom crypto aged badly: Curl-P survived a 2019 broken-hash disclosure badly (MIT DCI's collision attacks, ePrint 2019/344 — the episode that cost IOTA the cryptographers), and IOTA itself patched around ternary with a binary Keccak variant (KERL) years earlier. And the product moved on: the Chrysalis upgrade (April 2021) removed ternary from the protocol entirely — standard binary transactions, standard EdDSA signatures — and the later Rebased fork we build on is pure binary/Sui-style Move.

3. Our verdict. Binary everywhere: X25519, AES-256-GCM, ed25519, sha256, Move objects. Ecosystem, tooling, and auditability beat radix arithmetic every time — that sentence is in our whitepaper's honesty box on purpose, because "we chose the weird math" is exactly the criticism that hobbled IOTA. No trit-based address or hash scheme will ever ship.

4. The one homage that stays. The doof — 1 Qals = 1,000,000 doofs — is IOTA's million-multiple convention (1 MI = 1,000,000 i) carried over: humans read friendly units, machines count the smallest. That's the entire inheritance; the base-3 dream itself stays respectfully dead.

5. The someday line. The one live thread is qutrits in quantum error correction — 2025–2026 arXiv work is the first credible "ternary done right," because physics gives the third level for free. Relevant to us only if the compute marketplace ever brokers quantum-QEC capacity; it's a watchlist item, not a roadmap. Full survey: Ternary Systems.