HN Explorer

3 comments

[ 4.2 ms ] story [ 13.1 ms ] thread
dekhn
Most of the quality of a dish comes not from the selection of ingredients, but having a good understand of mouth physics and chemistry. That's mentioned in the article, but given the same ingredients, a great chef can make a far more appealing dish. This would involve steps like browning, careful heat control to obtain Maillard reactions, heat control to get crunchy dried texture surrounding a soft interior (crunch followed by burst of flavor is always a winner).

basically: fry bacon til crisp, dry, pour a little melted chocolate on it, add crispy bacon (fried until brown), decorate with thinly sliced mint. The different between an unskilled chef and a skilled chef making this dish will greatly outweight any dish the IBM supercomputer invents by combining incredients. Why? bacon is high in umami, it's crisp, the chocolate provides the sweet and bitter flavors, and the mint contains tons of aromatic molecules.

dsizzle
Doesn't much of the "mouth physics and chemistry" depend on the ingredient selection?

>"bacon is high in umami, it's crisp, the chocolate provides the sweet and bitter flavors, and the mint contains tons of aromatic molecules."

Bacon is always high in umami, chocolate is always sweet and bitter, etc.

I do wonder about your point regarding technique... would a recipe be rated differently if the bacon wasn't crispy?

dekhn
Sorry, I meant to come back and clarify: the ingredient part of what I listed could be discovered easily. Knowing how to balance all the levels of crispy, the sweet/bitter/fat of the chocolate, etc, at runtime is much harder. You can include those things in the cost function for the heuristic search, but they're harder to achieve that naive assembly of the ingredients.