Chess, but time is a resource.

The economy, deep dive

We will be diving into the time economy in detail, explaining how the three main economies price the same events differently, and how that changes the way you play.

The price of victory: mate costs money

Delivering checkmate takes moves, and moves cost time. Approximate time costs for the basic mates:

Practical move counts vary with piece placement and defender resistance; bills are the toll cost of that many moves at base rates, not a hard guarantee.

Because tolls never scale, the conversion bill is the same in every rated pool, one number to learn, valid from Zeitnot to Mainspring. So while playing that means you can be completely winning on the board and unable to afford the win. If you reach K+R vs K with less than 70 seconds on your clock, you will flag mid-conversion if your opponent doesn't slip up. Being winning isn't the same as being able to pay for the win.

1. Budget your endgame before you enter it. Before liquidating into a winning ending, check whether your clock can cover the conversion bill.
2. The insufficient-material draw rule is your opponent's best shield. If they're down to a bare king when your flag falls, you get a draw, not a loss, a desperate defender's best hope is often to shed material fast and race your clock.
3. Mate in the middlegame is cheaper than mate in the endgame. A sharp, clean mating attack costs a handful of tolls; a technical grind costs a fortune. KiriChess rewards players who finish efficiently.
4. Know the technique, don't improvise it. The cost per mate above assume the textbook method, driving the king to the edge with no wasted moves. In KiriChess endgame study isn't optional polish, you need to know how to convert efficiently or you'll run out of time ending in a draw or loss.

Trade math per economy

The three main economies price the same knight trade completely differently:

Under Burn, an equal trade drains 10 seconds from both players, trading becomes a weapon, and "can I even afford to trade?" becomes a question. Under Bounty, trades inflate both clocks and games run longer the more material is captured. But your opponent might be low on time, making an equal trade disadvantageous. Under Steal, only genuine material wins matter.

Why the queen costs 4 and not 5

By raw material logic, the queen (worth a rook + bishop) " should" cost 5s to move if the rook costs 3 and the bishop 2. She's deliberately discounted, for two reasons. First, tolls are paid per move, and the queen is the most-moved piece in most attacks, linear pricing would tax attacking chess into the ground; a ten-move queen attack at Q=5 costs 50 seconds on top of the double price sacrifices already pay. The discount is a pro-aggression subsidy. Second, the check table above. At Q=4, queen checks are exactly self-financing. At Q=5 they'd be gap-negative, attacking with your queen would lose you the clock war even while giving checks. The 4 is load-bearing.

The six rated pools

Each pool locks in a time control, an economy, and a capture multiplier (C) . The multiplier scales capture transfers/burns only, tolls and chips stay at base in every pool, and promotion is free everywhere. The time control decides how much you get to think; the multiplier decides how hard the material war hits. The two dials never fight each other.

Custom configurations are always available for unrated games, every dial in this manual can be turned, including full economy scaling (F-mode), where tolls and chips scale too.

The thinking budget

The clock number on the pool label is not how long you get to think. Tolls drain it, Burn drains it further, Bounty refills it. Subtract the economy from the pool and divide by a typical game (~40 moves per player) and you get each pool's real thinking budget, here is how I currently calculate it:

• The toll column is a constant (~81s per player), because tolls never scale, the only variable drain is Burn.
• The ladder is monotonic: 1.4 → 2.1 → 5.5 → 7 → 16.7 → 20.5 seconds per move. Thinking time rises with the time control, always.
• Averages hide divergence in high-multiplier Steal pools, Pendulum's "5.5s" and Sundial's "20.5s" describe the game, not either player. Once material falls unevenly, the winner thinks in luxury while the loser scrambles.
• Burn drains are estimates of typical behavior, good Burn players trade less because trades cost them, so real Ratchet and Mainspring thinking budgets drift slightly above the table.

Real-world game duration

How long a game actually lasts depends on the total time in the system, and on the level of play, beginners usually play shorter games (~25–35 moves) with heavier material flow and rarely use their full time; advanced players (~40–50 moves) push games toward the flag or a paid-for conversion.

• Steal is zero-sum, so multipliers don't shorten the system clock, Pendulum's ceiling equals Flagfall's. What C2 changes is decisiveness: one blunder produces an unrecoverable gap and a fast flag.
• Burn pools always run well under their nominal ceiling, and the beginner–advanced spread is widest there: beginners trade recklessly and collapse both clocks early, while advanced players who hoard material can stretch Mainspring near twenty real minutes.
• Zeitnot is the only pool where better play makes longer games, strong players farming bounty push past the nominal ceiling. The casual pool rewards staying alive, by design.
• Sundial games are as long as the players make them, with almost no system drain, the ceiling is nearly the full half hour, though in practice a single C4 swing usually decides matters well before the clocks run down.

Why capture-only scaling (Second Edition change)

The First Edition scaled the entire economy (F-mode: captures, tolls, and chips all multiplied) in the longer pools. Playtesting exposed a flaw: the toll multiplier silently taxes thinking time. An F2 pool didn't only double the stakes, it also doubled the toll cost of every move, draining minutes of pure system time. First Edition Pendulum, marked as the careful-calculation pool, actually left less thinking time per move than Flagfall; First Edition Sundial destroyed over five minutes of its own pool in tolls and priced a basic rook mate at four and a half minutes, so winning players routinely flagged mid-conversion.

Capture-only scaling (C-mode) separates the two dials: the time control sets how much you think; the capture multiplier sets how hard the material war hits. Two elegant properties fall out for free: the check profitability table is universal (tolls and chips at base everywhere means "pawn check = +3, queen check = self-financing" is true in every rated game), and the conversion bill is universal (K+Q ≈ 48s, K+R ≈ 70s, in every pool: "budget the win" is one lesson, not four).

F-mode still exists as a custom-game option for players who want the full-tax experience.

Why promotion is free (Second Edition change)

The First Edition charged time-value × 2 for the promoted piece. It was removed on a triple-taxation argument: the pawn already pays its way to the eighth rank in march tolls, and the new queen is instantly priced by Rule 1, she's the biggest capture bounty on the board, standing on the most dangerous square she'll ever occupy. A fee at the finish line taxed the same achievement a third time, and in low-clock endgames it punished the player for winning, the worst feeling an economy can produce. Removing it also collapses the priced-event list to exactly three things: moves, captures, checks. The fee survives as a custom-game dial for players who miss promoting to a knight because it's all they could afford.

Open dials (known trade-offs, watched in playtesting)

None of these are settled. Every dial above is exposed in custom games: different piece values, different tolls, F-mode, alternate capture personalities, so if you play one and it feels off, or you try a custom config that fixes something, that's exactly the kind of feedback that's wanted. Better to hear it now than lock it in.