Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[mikekchar]

I'm intentionally keeping the scope on this suggestion small with the hopes that it has a better chance of being implemented :-)

Problem: Food preferences are random with unobtainable items being just as likely to be preferred as common items.  Similarly, fortresses have no sense of culture because food preferences are all over the place.  There is no incentive to try to specialise with different types of ingredients.

Solution: Have food preferences change based on experience.  Every time a dwarf eats a cooked meal, there is a chance (very small and dependent upon the quality of the meal) that their food preference will change to one of the ingredients in the meal. When I say very small, I mean something on the order of 2% for a masterful meal, and maybe 0.1% for normal meal.  If a dwarf eats masterful meals every month, then their tastes will change on average once every 50 months (just over 5 years).  If they eat a normal meal every month, then their tastes will change on average every 1000 months (about 83 years).  Which doesn't seem like a lot, but if you have a fortress with 100 dwarfs, then after 1 year of eating nothing but masterful meals, about 20 of them will change their preference to an ingredient used in one of those meals.  For normal meals, you will get about 1 dwarf every year converting their preference to an ingredient in the meal.

Dwarfs actually eat meals fairly rarely unless you jump through hoops, so the numbers will be a bit smaller than that and it would need to be tweaked, but the idea is that a fortress will eventually have the majority of its members preferring a food that is actually available in the fortress.  It would also help to provide a kind of culture in the fortress, while still allowing some dwarfs to be outliers with crazy preferences.  Also, it makes young dwarfs and migrants dwarfs more likely to have crazy food preferences.

While I think this technique would be usable for all preferences, I think it would make sense to try it with food first since the mechanism is straight forward (meals have quality, there is a defined point at which a meal is consumed and the rate of preference change is predictable).  I also hope that it would be fairly easy to implement.  If it were successful, then maybe other things could be tried.

[StagnantSoul]

Meh, this would make food management easier, but I haven't had Gouda cheese in nine years and it's still my favourite cheese.

[PatrikLundell]

The general topic of food preferences has been brought up earlier in http://www.bay12forums.com/smf/index.php?topic=155850.0.
However, to adhere to the OP wishes to keep it (relatively) simple I'd suggest acquired tastes would be added rather than replace something (at least other than other acquired tastes), and also that there ought to be a "disgust" factor if the same thing is served too often, pushing the acquired tastes value down (DF must track past meals already, as dorfs get tired of eating the same thing all the time, so it shouldn't be a new thing to track). I'd also suggest the taste adoption rate should be increased somewhat to compensate for negative repeat factors. I'd find it desirable to have a system that encourages managers to provide food variety (without penalizing lazy ones too hard), so it shouldn't be optimal to produce just a small number of things.
Another possible way to tackle taste acquisition would be to make an RNG determination the first time a given type of food is encountered to determine whether it's an adoption candidate or not (if so, the adoption rate needs to be increased further). This would encourage providing food diversity.
(And, by the way, there is a typo in the OP: it should say 4 years).

[FantasticDorf]

Reading the OP, i like the idea that someone's likes could change once or twice in their lifetimes similarly to how their values might change after reading influencial books without going into detail into other posts on the thread.
However it'd be nice also if rather than a dependency for players to condition or micromanage dwarves into liking a food by force feeding them absurd amounts of plump helmets for 50 years, eating a prepared meal with a favourite ingredient would simply raise the decent meal need satisfaction much faster. As opposed to the system we have now where the decent meal need is dependentant on the ingredients, even though other prepared meals may be of a higher quality.

And likewise if you are eating something you dislike, like a rat from a cage (remembering that dwarves consider trapper caught animals delicacies) it will have reverse effect and lower + intensify the decent meal need.

[GoblinCookie]

My position on this was that food preferences should be culturally determined based upon what food items are available in the home civilizations biomes (and with the economy those they trade with).  So new dwarves would have the food preferences that reflect the foods available back home, but we could have blank slots which we can fill up with whatever food items are available in the new environment, along the lines of the OPs thread.

[SixOfSpades]

My position on this was that food preferences should be culturally determined based upon what food items are available in the home civilizations biomes (and with the economy those they trade with).

That works for the starting 7 and migrants, but children born in the fort should start with no preferences at all, only developing them by exposure. It always annoys me when I embark in a Temperate Coniferous Forest and see newborns that like rhinoceros meat and jaguar teeth.

If I had my druthers, dwarf preferences would be more randomized than they are now: not simply ONE stone and ONE gem and ONE metal and ONE wood and ONE leather and ONE bone and ONE finished item, etc., but instead a random number [0 to 4] from each category (probably capped at the same overall total, though). Also, multiple preferences in the same category would add a positive modifier to the experience gained from reactions working with that category . . . so a dwarf that likes 3 different kinds of booze would learn the Brewer trade more rapidly than a dwarf with only 1 alcohol preference. Now, that conflicts somewhat with

I'd suggest acquired tastes would be added rather than replace something

because ALL dwarves eat & drink, far more often than they do most other tasks, and making every dwarf like 13 kinds of food and 6 kinds of booze is the opposite of making one's dwarves individually distinctive . . . and if those artificially inflated preferences can then turn around & bump up the learning modifier for related skills, then that sets up a feedback loop where EVERY dwarf in the fort wants to become a Cook or Brewer, other professions would largely fall by the wayside. So I support having new preferences replace old ones, and only of the same type. So a dwarf who likes claro opal, pyrite, and turquoise may decide that she likes moonstone, but that shouldn't affect her overall commitment to being a Jeweler, so she'll have to give up one of her existing 3 preferences in order to make way for moonstone.

[mikekchar]

I just thought I'd explain a bit of rationale for the suggestion.  It's really more of a pragmatic change to improve the situation rather than an overhaul that will fix the situation.  The reason I suggested replacing a food preference is because the scope for things going wrong (and hence making the change harder to implement) is much lower.  If you added preferences, then you need to make sure that dwarfs don't get too many, etc.  Although, after making this suggestion, I've discovered dwarfs that like no foods and dwarfs that like 2 foods, so perhaps it is already more complex than I think.

One thing I've also noticed in my current game is that children are often satisfied by "masterful meals" despite having nothing in the fortress that meets their preferences.  So perhaps there is a bug somewhere.  It seems that it *does* try to take food quality into account.  One thing I'm going to experiment with is intentionally running out of raw ingredients so that all dwarfs are forced to eat meals and see what it says.

Finally, I like the idea that there are outlier tastes in the game.  I also don't mind if dwarfs are born with preferences for things they have never eaten.  I never ate natto (google it ;-) ) until I was nearly 40, but the first time I tried it I loved it (even though it was not masterfully prepared ;-) ).  But I do think that the population should tend towards a cultural medium that depends upon idiosyncracies of the individuals as well as availability of the ingredients.  So even though Korean and Japan have very similar climates and availability of food, their cultures evolved to have very different cuisines (incredibly spicy in the case of Korea and very simple flavours in the case of Japan) .  I think my suggestion will go in that direction.

But there will still be problems, and I think all of the suggestions in this thread (and others) are very valid.  I'm really just trying to think what could be done in a couple of hours of work without having to overhaul the code, rebalance the game, etc, etc.  I imagine that this is such a low priority (compared to the other things that could be done), that if it's not a simple tweak it's just not going to get looked at for a few years.

[PatrikLundell]

To clarify (as SixOfSpades misunderstood it): I did not suggest every dorf should adopt every kind of food they eat immediately, but rather use a slow adoption process (in the same vein as mikekchar's initial suggestion), so a dorf may acquire one or two things over a life time. If replacement is easier to do that addition I'd rather take that now than an overhaul during the late 20:ies.

And yes, different dorfs have different numbers of preferences of different categories (I had one amusing case of dorf who liked to consume water. Unfortunately I scrubbed that fortress as it was just a test embark, so I didn't see how that panned out).

[SixOfSpades]

I also don't mind if dwarfs are born with preferences for things they have never eaten.  I never ate natto (google it ;-) ) until I was nearly 40, but the first time I tried it I loved it.

In that case, the best way would probably be to make the preferences there, but invisible to both the player and the dwarf until they're exposed to it, which could trigger a happy thought: "Urist was delighted to discover a new favorite food recently."

[mikekchar]

I like that idea.

[NW_Kohaku]

I'm intentionally keeping the scope on this suggestion small with the hopes that it has a better chance of being implemented :-)

Where is the evidence that a smaller-scope suggestion is more likely to be acted upon by Toady?  If anything, whenever Toady does anything, he goes whole-hog on it for a month, so more complex ideas are more likely to attract his attention.

In any event, I already threw my two cents in on that previously mentioned thread here, where I stated that it would make more sense to make dwarves prefer a general flavor rather than an extremely specific creature. (I.E. they like "salty foods" or "fishy flavor" instead of specifically "striped bass".)

The biggest problem I have with the current system is that dwarves can prefer things that they could never have even seen, or might not even exist, such as panda meat in a world where there aren't any bamboo forests to support pandas.

The system you suggest in the OP would technically solve that problem, but could easily defeat the purpose of preferences in the first place: If all you serve is plump helmet biscuits, then even with only a 2% chance of switching favorites, it's an eventuality that everyone switch to loving plump helmet, and there's no reason for any variety at all outside having just two food types to prevent the coded rejection of having the exact same meal for years.

In that case, the best way would probably be to make the preferences there, but invisible to both the player and the dwarf until they're exposed to it, which could trigger a happy thought: "Urist was delighted to discover a new favorite food recently."

The problem with this (which I also argued against in that previous thread) is that secret preferences functionally mean that players need to not just potentially have every food type in the game the way that we have it now, but that we now need to actually collect every possible food type in the game if we're going to take advantage of the preferences system at all. 

In the previous thread, I likened this to a Dungeons and Dragons game where, to get through a dungeon, you need to put a random item up to the door, and there are no hints, and it can be literally anything from the equipment list of any splatbook the dungeonmaster is using, so you wind up having to take at least one of every single item in the game with you everywhere just in case you need to use it as a key. 

That doesn't encourage clever puzzle-solving to figure out how to meet the demands, it just puts a burden on the player to horde absolutely every random thing "in case they need it one day".

[SixOfSpades]

I stated that it would make more sense to make dwarves prefer a general flavor rather than an extremely specific creature. (I.E. they like "salty foods" or "fishy flavor" instead of specifically "striped bass".)

Agreed. I assume that the specific-animal food preference was originally intended just as a temporary placeholder, until Toady got around to coding more realistic recipes for things like goulash, omelettes, and pizza (as well as other types of balanced meals that simply don't have names). Then again, people do exhibit firm likes/dislikes for some very specific foods, especially when those foods have strong flavors (e.g. durian, anchovies, jalapenos, etc.).

I suppose the best way would be to set up a system of variables, like Sweet / Salty / Spicy / Fruity / Meaty / Fishy / Bitter / Crunchy / Moist / Sharp / Fatty / Smoky, etc., and give each ingredient (and booze) its own value on each of those axes. For the time being, prepared meals should have have their flavor variables be the average of all their ingredients, until someone develops a better method. Meanwhile, each dwarf would also have his own value for how much he likes / dislikes each of those general tastes. Depending on how strong those likes & dislikes are (i.e., how particular the dwarf is about food), the dwarf may have one or more preferences for specific ingredients that fit his individual pattern.

The system you suggest in the OP . . . could easily defeat the purpose of preferences in the first place: If all you serve is plump helmet biscuits, then even with only a 2% chance of switching favorites, it's an eventuality that everyone switch to loving plump helmet.

Even more complexity, then. Suppose each dwarf has TWO sets of taste preferences: A "base", permanent one, and one that fluctuates with what he's eaten lately. When he eats an item, the game looks at the item's taste properties and decreases his perceived desire for those tastes by a proportional amount. Let's say he eats a raw plump helmet, which tastes 12% Crunchy, 16% Fruity, and 8% Bitter. His desire for those tastes at his next meal will drop by those same amounts--but over time, they will gradually creep back to their baseline values, so he would have to endure a monotonous diet for quite some time before he got really sick of it, and actual starvation would reset his preferences entirely, making him happy to eat anything.

In that case, the best way would probably be to make the preferences there, but invisible to both the player and the dwarf until they're exposed to it.

The problem with this is that secret preferences functionally mean that players need to . . . actually collect every possible food type in the game if we're going to take advantage of the preferences system at all.

Actually, no, for two reasons.
1) Say a dwarf has a hidden, "true" favorite preference. That shouldn't stop her from having a favorite-so-far preference, chosen from among the foods that she has eaten, and for all she knows it's her most favoritest ever. If the happiness boost from eating a favorite-so-far ingredient isn't as good as the boost from the "true" favorite, then it should be damned close. If/when she ever encounters her true favorite, her previous preference might slowly fall off the list . . . but she'll still likely enjoy eating it, because it matches her preferred tastes.
2) If we do switch to a taste-based system, then there will be NO need for specific ingredients, only for ingredients that have a broad variety of tastes. A dwarf who likes Sweet, Fishy, and Crunchy foods might logically have shrimp as a preference, and if you don't have shrimp you might never be able to make him truly ecstatic with the menu . . . but if you've got sweet pods, cave fish, and celery, you can at least keep him moderately pleased.

[NW_Kohaku]

I stated that it would make more sense to make dwarves prefer a general flavor rather than an extremely specific creature. (I.E. they like "salty foods" or "fishy flavor" instead of specifically "striped bass".)

Agreed. I assume that the specific-animal food preference was originally intended just as a temporary placeholder, until Toady got around to coding more realistic recipes for things like goulash, omelettes, and pizza (as well as other types of balanced meals that simply don't have names). Then again, people do exhibit firm likes/dislikes for some very specific foods, especially when those foods have strong flavors (e.g. durian, anchovies, jalapenos, etc.).

I suppose the best way would be to set up a system of variables, like Sweet / Salty / Spicy / Fruity / Meaty / Fishy / Bitter / Crunchy / Moist / Sharp / Fatty / Smoky, etc., and give each ingredient (and booze) its own value on each of those axes. For the time being, prepared meals should have have their flavor variables be the average of all their ingredients, until someone develops a better method. Meanwhile, each dwarf would also have his own value for how much he likes / dislikes each of those general tastes. Depending on how strong those likes & dislikes are (i.e., how particular the dwarf is about food), the dwarf may have one or more preferences for specific ingredients that fit his individual pattern.

The system you suggest in the OP . . . could easily defeat the purpose of preferences in the first place: If all you serve is plump helmet biscuits, then even with only a 2% chance of switching favorites, it's an eventuality that everyone switch to loving plump helmet.

Even more complexity, then. Suppose each dwarf has TWO sets of taste preferences: A "base", permanent one, and one that fluctuates with what he's eaten lately. When he eats an item, the game looks at the item's taste properties and decreases his perceived desire for those tastes by a proportional amount. Let's say he eats a raw plump helmet, which tastes 12% Crunchy, 16% Fruity, and 8% Bitter. His desire for those tastes at his next meal will drop by those same amounts--but over time, they will gradually creep back to their baseline values, so he would have to endure a monotonous diet for quite some time before he got really sick of it, and actual starvation would reset his preferences entirely, making him happy to eat anything.

In that case, the best way would probably be to make the preferences there, but invisible to both the player and the dwarf until they're exposed to it.

The problem with this is that secret preferences functionally mean that players need to . . . actually collect every possible food type in the game if we're going to take advantage of the preferences system at all.

Actually, no, for two reasons.
1) Say a dwarf has a hidden, "true" favorite preference. That shouldn't stop her from having a favorite-so-far preference, chosen from among the foods that she has eaten, and for all she knows it's her most favoritest ever. If the happiness boost from eating a favorite-so-far ingredient isn't as good as the boost from the "true" favorite, then it should be damned close. If/when she ever encounters her true favorite, her previous preference might slowly fall off the list . . . but she'll still likely enjoy eating it, because it matches her preferred tastes.
2) If we do switch to a taste-based system, then there will be NO need for specific ingredients, only for ingredients that have a broad variety of tastes. A dwarf who likes Sweet, Fishy, and Crunchy foods might logically have shrimp as a preference, and if you don't have shrimp you might never be able to make him truly ecstatic with the menu . . . but if you've got sweet pods, cave fish, and celery, you can at least keep him moderately pleased.

Well, you've pretty much adopted my suggestion as the first rebuttal to my problem with the previous suggestion.

If you have a two-favorites system, though, I have to wonder why you'd bother hunting for a "true favorite" when you already have a "good enough favorite"?  I mean, it doesn't really solve the problem that there's literally no way to be sure what that secret favorite is other than to perform the almost certainly impossible task of trying to have EVERY FOOD IN THE GAME just to be sure nothing falls through the cracks.  Mechanically speaking, the only interaction that the player has with this system is that they control the availability of food variety, and hiding what their target is from them just ensures they can't look for the most important dwarves and only break their backs bending over backwards to just supply the king with his favorite great white shark roasts by going out of your way to start a great white shark breeding program. If the solution to it is to make a second, easier target, why have the impossible-except-by-blind-luck target at all?

While some players will force a limited variety of the highest-efficiency foods (or just the simplest foods, such as plump helmets when quarry bushes are blatantly superior), I'd suspect most players (especially when mechanically incentivized to add variety) will have the largest variety of foods that their fortress can reasonably support in a sustainable manner without demanding overmuch micromanagement.  Asking the player for more than that is just asking the player to choose between doing unwanted chores that make the game less fun for them, or getting hit with whatever punishment is meeted out for failing to comply with this utterly bonkers random favorites.  (I'll point out that slade preferences in nobles were grounds for immediate "unfortunate accidents" and were patched out of the game for a reason. It's one thing if the mayor/duke/king really likes cups and demands lots of fancy goblets, because you can actually directly comply with that.  It's another if they demand white zircon, which isn't in your fort, but you can trade for it.  It's something else entirely if they're demanding something you can't give them, and they can't even tell you where to look.)

Keep in mind, most fortresses will have access to EITHER tropical OR temperate above-ground crops.  Most fortresses have access to a few breeds of livestock, but players are not going to build pastures to house every single possible variety of livestock plus every wild animal.  Players pretty much stop at either having geese or turkeys for eggs, and maybe switch to crocs if they can catch and tame some, they're not going to individually breed guinea fowl, ducks, chickens, geese, turkeys, plus all the others. 

This is generally why I push for a "flavor" style of approach - it's a matter of what you're asking the player to do.  If you want the player to cover every base, you need to make it possible to cover every base, and the best way to do so is to only demand a reasonable amount of set-up before it just becomes a Sisyphean chore.   

You need to start out with asking what you actually want to get players to do, and build your rules around incentivizing that kind of behavior.  The current "dwarves revolt if fed nothing but the same food over and over" mechanic exists just to prevent a nothing-but-plump-helmets fortress among players who favor economy of their own actions over any form of simulationism. 

Asking players to have, say, 3-5 types of meat/seafood, at least one type of egg and/or one type of milk, and 6-18 varieties of vegetables, herbs, and spices in general would be a reasonable goal, in my estimation.  From there, you reverse-engineer the mechanics, such that you have tastes that follow specific categories.  If you make, for example, most creatures "taste like chicken", but then have some dwarves really love (cow beef) steak, you're functionally demanding that players raise cows, and therefore, have pastures even though chickens (or geese or turkeys) are easier to care for.  (And that might be worthwhile, if your goal is to make players have to use pastures instead of relying upon the easier-to-manage poultry, or just having an all-vegetarian fortress.)

[mikekchar]

How many suggestions have very limited scope?  From what I can see, none of them.  Every time someone makes a suggestion, the thread is inundated with, "No, what we should really do is completely redesign the way it works from the ground up".  Which I completely understand, but I'm just saying that the frequency of a non-occurrence is necessarily zero :-)

Small tweaks have happened in the past when a release was winding down, or immediately after a release during bug fixing.  The timing of the suggestion is not by accident.

To be honest, even *I* wouldn't prioritise this up the list unless I thought it could be done in an hour or so.  I mean, if we want to talk about food -- well, it's completely broken and the other threads on the topic have wonderfully good suggestions.  If we want to talk about needs that are impossible to meet, then fixing the problem with dwarfs unable to have anything other than casual relations (and the occasional miracle wedding) is dramatically higher than this, IMHO.  I'm just saying, "Hey, if you happen to be in the code and have an hour, this is a small tweak that will improve the situation while keeping the risk of breakage very, very low".

I can't really think of any other suggestion that is specific, small, and will have a predictable result.  Which is not to say that they aren't great suggestions.  It's just that they will need the old code to be investigated and ripped out, the new code to be implemented, a testing phase and almost certainly a balancing phase to deal with unanticipated results.  And, yes, Toady will do that, but only at the cost of other functionality so it becomes a rather dramatic prioritisation call.

[NW_Kohaku]

How many suggestions have very limited scope?  From what I can see, none of them.  Every time someone makes a suggestion, the thread is inundated with, "No, what we should really do is completely redesign the way it works from the ground up".  Which I completely understand, but I'm just saying that the frequency of a non-occurrence is necessarily zero :-)

There are plenty, but they're also the ones nobody remembers, because they didn't engender any real debate. 

As I said before, I contest the very notion that a "small suggestion" has any inherent superiority to begin with.  The only real remaining purpose of the suggestion forum, if it can be said to have one at all, is so that contributors can hash out amongst themselves the ramifications of different ideas and come to an agreement over what the goals of any changes or systems may be, as well as suggested implementations that Toady may not have considered (such as pathfinding optimization threads). Beyond that, generally speaking, with a decade of discussion and very slow progress getting to any of the Eternal Suggestions, there hasn't been much new under the sun in terms of actual topics (as pointed out before, this topic has come up several times before), so the value to be gained in discussion of the topic is in refinement of the idea, not coming up with something short and simple that discourages discussion on the grounds that discussion makes things less simple.  (Those "wonderful" suggestions are "wonderful" because they have been refined after taking into account plenty of differing arguments and points of view to not only determine the best problems to solve, but also handle all the complex side-issues and edge cases brought up by dissenters.)

And for as much as it's taken time to get around to the big asks like minecarts and wheelbarrows, there are tremendous loads of "small, simple things" that have been neglected for just as long, even when they were easy. 

Or, to make a short and simple twist on an idea, don't let the good be the enemy of the perfect.  One of the major reasons Toady has given for his opposition to GUI updates, for example (although I should point out I strongly disagree with the reasoning in the case of GUIs, since that develops a feedback loop on the actual player experience, altering actual design goals), is that he doesn't want to make a change he later has to change again, since he finds it counterproductive.  "Stopgap" solutions that would then have to be ripped back out and redone again later are therefore actually far less likely to be implemented than coming up with the complicated perfect solution, towards which Toady might be able to work in small pieces.  Toady, for example, created things like grazing eating grass in spite of not having carnivores eat meat, nor browsers eating leaves, nor scavengers going hungry because it was an interative step towards the perfect, even though it suddenly ruined game balance and simulationism by making ranching crocodiles vastly more economical than ranching cows.

If you're going to suggest something "small", it therefore has to be a small step towards something that is big, anyway.  And that requires first finding out what that big thing is you're building towards.  That means you need to create that large, complicated perfect solution, then build back to it, arriving at something like "let's make cows eat grass" in order to build up to the giant, complicated thing that is the goal.