Guilds > Cooking Research Coordination > What We Do

One of our aims is to find combinations of cooking ingredients that increase the duration of a meal.

The best way of increasing the duration is to find ingredients that can be combined in equal amounts: The contribution of an ingredient towards a meal's duration is partially dependent on the largest amount of any single ingredient in a meal.

Let us use an example to explain this:

First recipe (all ingredients contribute positive durations):

• 4 Ingredient A, 3 Ingredient B, 2 Ingredient C, 1 Ingredient D
• Total ingredients: 10
• Highest amount of any single ingredient: 4

Second recipe (all ingredients contribute positive durations):

• 2 Ingredient A, 2 Ingredient B, 2 Ingredient C, 1 Ingredient D
• Total ingredients: 7
• Highest amount of any single ingredient: 2

The second recipe will result in a higher duration than the first, even though there are fewer ingredients in total.

If you could cook a meal with only one of each ingredient, you would have an optimised recipe for duration. Also, the more different ingredients are added to a recipe, the more chances there are of increasing the total duration of the meal. To be able to combine ingredients in an effective way, we need to know the stats that the ingredients can provide (this has largely been done), which ingredients interact with each other, and the duration contributions of all ingredient pairs in the recipe.

In the first recipe, there are 3 pairs of interactions: Ingredient B interacts with Ingredient A and Ingredient C, and Ingredient C also reacts with Ingredient D. Ingredient A does not interact with Ingredient C or D, as there are other ingredients in between (just look at the amount of each ingredient to determine the order).

In the second recipe, you cannot see which ingredient interacts with each other, as there are three ingredients with equal amounts. So it is possible that the interaction works as the written order suggests, or it could be, for example, that Ingredient A is between Ingredients B and C.

The order in which ingredients interact with each other in a meal when they are present in equal amounts never changes. The kitchen always assigns the same order to ingredients, regardless at what time they are added to a recipe, which other ingredients are put in, etc. We have devised methods of finding out what the order is, and are working on placing all ingredients in the Ingredient Order list.

To illustrate what this means, let us consider this portion of the ingredient order:

• Cabbage
• Pale Dhamasa
• Dates
• Honey
• Spinach
• Mountain Mint
• Carp
• Garlic

If we cooked a meal that consists of one each of these 8 ingredients, the ingredients would interact as follows with regard to duration:
Cabbage interacts with Pale Dhamasa
Pale Dhamasa interacts with Cabbage and Dates
Dates interact with Pale Dhamasa and Honey
...

In other words, you could say that Cabbage is the most dominant of these ingredients - the recipe behaves as if Cabbage were in it in the highest amount, and Garlic in the lowest.

To predict the total duration of this recipe, we need to know the duration that each pair of ingredients contributes. The pairs that interest us are:

• Cabbage + Pale Dhamasa
• Pale Dhamasa + Dates
• Dates + Honey
• Honey + Spinach
• Spinach + Mountain Mint
• Mountain Mint + Carp
• Carp + Garlic

No further interactions happen in this recipe (at least as far as duration is concerned).

Most interactions between ingredients result in negative durations. The trick is to find strings of ingredient pairs that provide positive durations.

Let's assume the following (this is an example; it's entirely made up):

• Cabbage + Pale Dhamasa => positive duration
• Pale Dhamasa + Dates => negative duration
• Dates + Honey => positive duration
• Honey + Spinach => negative duration
• Spinach + Mountain Mint => negative
• Mountain Mint + Carp => positive
• Carp + Garlic => negative

To build a good recipe from this, we would need further interactions, so let's further assume:
Pale Dhamasa + Dates => positive
Honey + Mountain Mint => negative

A good string to use could be one each of:

• Cabbage
• Pale Dhamasa
• Dates
• Honey

It might be a good idea to also add Mountain Mint and Carp to the recipe, but we don't know as long as we have not quantified the positive duration between Mountain Mint + Carp, and the negative duration between Honey + Mountain Mint. If the positive duration outweighs the negative one, we would further increase the duration of the recipe if we added Mountain Mint and Carp.

We are currently working on the following:

• Monitor ingredient stats on the BaseEffects page
• Place all known cooking ingredients in the Ingredient Order
• Populate the Timer Length Table to identify useful pairs of ingredients
• Identify strings of ingredients that can be used in the same Layer and provide desirable stats
• Develop and maintain Tools that help chefs design their own recipes

For those who are interested, here's broadly how we do this:

• Monitor ingredient stats:
  • Cook a meal with 7 of one ingredient (Cooking level 3), measure the stats. As stats are changing over time, we do this on a more or less regular basis.
• Place ingredients in the Ingredient Order:
  • People with high Gastronomy (at least 60?) receive certain messages when eating a meal with positive duration. The message says something along the lines of "The combination of ingredient A and Ingredient B was particularly good, whereas the combination of Ingredient C and Ingredient D was pretty bad". Ingredients that are listed first in these combinations (Ingredient A, Ingredient C) are always further up in the ingredient order list than those listed second (Ingredient B, Ingredient D). We can narrow down the place of an ingredient in the ingredient order by repeatedly cooking and tasting recipes that contain this target ingredient plus other ingredients that are close to it.
• Populate the Timer Length Table:
  • We have two spreadsheets that help us with this. One is designed to test one ingredient against the 38 primary ingredients, the other lets you fill in which ingredient pairs you are testing against each other and suggests a filler ingredient. The second type of test can only be done for ingredients whose relative positions in the sort order are known. It uses 1 of each specified ingredient, plus 5 of the filler ingredient. We use a tool that can automatically update the timer length table from these spreadsheets.
• Identify strings of ingredients that can be used in the same layer:
  • Work your way down the ingredient order list, and check which ingredients make sense to combine. Useful combinations are those of ingredients that provide positive target stats, result in a high duration, and are close together in the sort order. The interesting part of this research is to find combinations that are less obvious - e.g. where you sacrifice a little duration to gain more duration or stats further down the line.
• Develop and maintain cooking tools:
  • There are currently two tools for designing cooking recipes: Cooking Assistant and Cordon Bleu. One part of maintaining them consists of regularly generating new data files from the timer length table. Another part is to fix bugs and add new features to these tools.

If you are interested in helping, there are many ways in which you can help. We can always use donations of mushrooms and herbs, volunteers who help test ingredients, people who are interested in programming cooking tools, and people who teach newbies how to cook. Chat any elder of the Cooking Research Coordination guild if you are interested in joining the guild - though of course it is not necessary to join the guild to contribute to cooking research!