Taumorrowlogic you can check

Built on the Tau language · Germany · in the open since October 2024

Decided,
not guessed.

Taumorrow builds systems whose every step is a proof anyone can check: a chess engine that cannot bluff, multiplayer games that are fair by construction, and a game of rules whose core no rule change can break.

01

Three works, one method

Each is a different game. Each runs on the same idea: write the rules as formulas a machine can decide, let the machine decide them for every case, and keep a record that anyone can recompute.

Tau Chess — a chess engine that cannot bluff

A chess engine whose every move is decided by logic, not estimated by a network. Shortcuts that make the search faster are admitted only with a decided proof that they never change the result. It plays frontier language models live, in public, and shows its own view of the position as it thinks.

What it learns, it learns from its own games, under a gate that is decided before any change is accepted. Games against language models are a measure of strength, never training material.

The Round as a Formula — multiplayer games fair by construction

Lotteries, shared pots, bets against a bank, a hand of cards, a move on a board: games where several parties act at once and one step decides the round. Usually that step is a program run by an operator you have to trust. Here the round is a formula, every party can recompute it from the record, and the record cannot be rewritten.

The method began as a whitepaper in December 2025 and is published as a paper with the executions a third party can repeat.

A Playable Nomic — the game where changing the rules is the game

Nomic has been played since 1982. Where it has been computerized, a machine executes the rules the players write. Here the machine also decides whether a proposed rule change is legal with respect to the constitution, before it takes effect, and an entrenched core is proven to survive every sequence of legal steps.

Two levels of self-amendment on a decidable substrate: the players change the rules, the rules cannot change what was proven.

02

How it works, in four steps

The same four steps carry all three works. Nothing here is a parameter; the parameters stay in the workshop.

1

Small formulas, each decided

The logic is split by task into small specifications. The engine decides each one for every possible input, so there is no “probably”.

2

Dependencies become wiring

The output of one piece is the input of another. A transport layer passes values along and decides nothing itself.

3

Shortcuts need a proof

Anything that makes a search faster is admitted only with a decided statement that it never changes the result.

4

Every step leaves a record

Each game is a hash-chained record you can recompute. If the machine is wrong, you can show exactly where.

Tau Language, its engine and its logics are the work of Ohad Asor and IDNI AG. The architecture built on it, the three works and the papers are Taumorrow’s own work. Parts of the method are patent pending. The materials are published under the Taumorrow Demos License: open to read, study, verify and learn from.

04

Learn and follow

Everything runs as it is. Take a specification, run it, change a rule, watch the verdict change.

Tau-Nomic tutorial series

Twelve worked specifications, from the first rule to a self-amending game, with expected results you can check.

github.com/taumorrow/tau-lang-demos →

Fair gaming, from the start

The December 2025 whitepaper and the public framework for trust-free games.

github.com/taumorrow/provably_fair_gaming →

Updates

Results, live dates and short write-ups as they happen.

x.com/tau_l0g1x →

Join the team

Taumorrow is becoming a team around one method. Engineers, logicians, game designers, chess players: if you want to build things that can be checked, write to us.

taumorrow@gmail.com →