| block | mints | issued | used of budget | pressure after |
|---|
every block has a budget: the headroom under K shifted right by seven. that is one part in 128 of whatever has not been issued yet, so the budget shrinks as the curve fills and the chain can never issue the last token in a rush. the shift is chosen so that a full two eth pat fits in one block at every position: at the opening it fills a block almost exactly, and deeper in the curve two eth buys fewer tokens than the block allows. half of the budget is the target. a block that issues more than target raises pressure for the next block by a share of the excess over eight targets. a block that issues less lowers it the same way. an empty block multiplies pressure by seven eighths.
pressure is a multiplier, and a multiplier that starts at zero stays at zero. so the first block that crosses target from a resting state seeds pressure at one sixty fourth before applying its multiplier. from there thirty six consecutive full blocks carry pressure past one, which at ten blocks a second is under four seconds of crowding. five hundred and twelve idle blocks, about fifty one seconds, return it to exactly zero and the next crowded block seeds again.
pressure splits the eth a minter sends. the share that walks the curve is the sent amount divided by one plus pressure. the rest is surcharge. both go into the cistern. only the curve share creates tokens. at pressure one, half of a mint buys tokens and half buys nothing but a higher book for everyone. the surcharge is not a fee, because nobody receives it. it is backing.
it does not reject a mint. only two things reject a mint: a curve entry above two eth, and a block whose budget is already spent. it does not depend on the clock. it depends only on what the previous blocks issued, which is a quantity every node agrees on and no operator can edit. it does not expire on a schedule. it decays by use, and by the absence of use.