David Graeber, in his book "Debt: the first 5,000 years", tells a compelling story about currency, taxation, markets, and Madagascar. An army arrives on the shores of an island (specifically, Madagascar), and tells everyone: "You must provide us these tokens (money) when tax time comes around. You can get these tokens by bargaining with soldiers, particularly quartermasters, who will be exchanging them for goods like food and leather. If you do not provide us your ration of tokens at tax time, you will be put in prison."
The idea of this idea is to run the Multiplicative Update Weights Algorithm manually, using index cards, as a tool for thinking. You impose a rule that the idea needs to "pay its way", in order to stay "out of prison".
You create a stack of index cards, each of which has an idea on it, and a current weight, initially something like 100. You also keep, separately, a total weight figure; let's say its initially 52*100.
Each round, you write down your current situation, then roll dice or something to select a random number between zero and the total weight. Let's suppose you roll 245. Let's call that number the "selector". Then you shuffle the deck, and look at the top card.
If the top card's weight is less than the selector, then you decrease the selector by the top card's weight, and discard the top card.
If the top card's weight is greater than the selector, then you try to follow the advice on the card. After trying (for a timeboxed period, maybe), write down your current situation, and compare it to your previous situation. If the advice was applicable and helpful, then increase the card's weight by a factor, say 1.1. If the advice was inapplicable and/or unhelpful, then decrease the card's weight by a similar factor. Remember to update the total weight counter as well.
After a large number of rounds (1000?), sort the deck by weight, throw out the 5 cards with least weight, find or invent some new ideas, add them to the deck (with the initial 100 weight), again updating the total weight counter.
I have not followed this procedure, but I would like to, possibly to create a "Software Design Tarot".