My S&L design kept the saving and loans as separate organized by rarity graphs, the color being set to risk that cash deposits are dinged, automatically, to balance across nodes, making interest payments.
A somewhat related issue, what happens when a rare liquidity addition happens in the pit? The existing deposits are, relatively less rare, they get pushed down the tree. But, they lose an interest income, they have lost a partial payment for their prior rarity. Down the tree, bit error will be smaller. probability * size of bit error has to be within bound of each other. More probable event, the smaller bit error. So, the borrower get a break, someone took a huge loan and pushed him down the tree, the skips a partial expense for being rare prior.
The when and how of dinging for payments is a function of work load as bots clamor to spot the openings, in lending or depositing. The ex post becomes even mote ex post when the pit boss is locked out. So we see, more and more, a stable set of contract parameters, mostly related to queueing imbalances on the graph and how they are dispersed.
Anf, the smoothness requirement on the pit boss. It can be behind, it might be managing bin sizes and node matching. The breaking points, on and off the tree, have to leave most of the tree mostly in the same level of disorder, proportional to cycles on the graph. Just,. no sharp noticable breaks in smoothness.
No comments:
Post a Comment