Bay 12 Games Forum

a.) Working military (somewhat difficult in the newer versions).

What does that mean? It's far too subjective to mean anything at all. To one person, just recruiting a bunch of peasants into a squad counts as a "Working Military," -- even though they will all most likely die in any engagement. But to another person it isn't a "Working Military" until the units are all skill level 10+ in armor/shields/dodging/weapons and have equipped at least full iron armor and a steel weapon. Or what about one outrageously well trained dwarf in full steel gear, who stomps all over the invaders?

Need a better definition that "Working" military.

I agree that there's an unnecessarily large number of button presses required to complete a trade encounter. The worst offender to me is searching for food in the barrels menu. The next worst is re-specifying my desired imports, even if they don't change.

It would be an improvement to move Food/Drink to their own submenu for Move Good to Trade Depot action.

It would also be good to have trade agreements saved from year to year, and only need to make modifications to it when there are changes desired for next year.

You are assuming that the soil values are completely random and mutually exclusive; they are not. The chances are, that if some of the variables are right for a plant to grow, then the others will be too, maybe one or two variables would need to be fixed for it to grow well, but most plants would be able to grow in most soil types.

What I'm assuming is that they are independent variables, not mutually exclusive. That is to say, it is possible to have 1% in one variable and 99% in another variable, and furthermore that there isn't a dependency that says if one measurement is at 50% another will never be at 5%. The assertion that "Chances are if some of the variables are right, then others will be too" has only now been stated. If there are rules of dependency, they will need considerably more clarity before they can be programmed and analysed.

I've explored a couple possible scenarios, and given calculatable solutions for each of them. If you have another set of values you wish to try, then just plug them into the formula: spanPercent ^ numVariables = growPercent, where spanPercent is the average width of growable conditions in a typical variable, and growPercent is the percentage of all plants that will grow there, on average.

Note that these are average values, and they are part of the plant specifications. It is possible to have one measurement where everything above 10% is acceptable and another where everything below 90% is acceptable, and they average out to two variables with a 90% viability range. The design of the framework and the design of the plants that will live in it are interdependent, not independent.

In addition, a grow/not grow dichotomy is to be avoided. There will be plants that do decently over a wide range, but especially well in a small range.

Let's say that any given plant at least grows over 90% of the range of values for a given metric; that the bottom 5% is too little and the top 5% is too much of a given soil metric for a given plant. That means that if there was only a single variable, then there's a 90% chance that a given plant and a given soil combination would work.

Now let's look at what happens when we make it ten variables, and all of them have the same 90% coverage with only 5% on the ends that doesn't grow. In order for any given plant to grow, it must be in the 90% span of all ten metrics. That's 0.9^10 = 34% coverage. Another way to look at it is to say that even in this extreme of a 90% span of allowable values in all variables, about 2/3 of all plants cannot grow in any given soil type.

And this scenario is the exact opposite of "vast differences" in plant soil preferences. If we narrow the ranges to create more variety in plant preferences, the number of plants that can grow in a soil quickly drops off.

Engineering an acceptable solution could be done like this: Let's say we decide that the idea of having only 1/4 of all plant types grow on any given soil is a reasonable distribution. Furthermore, we wish to have a variety in plant soil preferences, say only 50% of the range of any given variable is viable. Then the math becomes 0.5^x = 1/4, where x represents the number of variables needed to achieve this goal. Solving for x: x = log(1/4) / log(0.5) = 2 variables.

Another scenario: have only 1/20 of all plant types grow in any randomly selected soil sample, and have plant preferences set to 50% of each variable's range. The result: x = 4.3 variables. Since variables can only be an integer, we can fudge around with it by setting x=5, and then letting the percentage range or number of plant types change. Let's let the percent range change, so that the formula is: x^5=1/20. Solving for x, the span of viability on each of five variables becomes about 55% of the range.

The difference between directed and sandbox games is in the level of definition of the finish of the game. Dwarf Fortress is a sandbox game because it does not explicitly define a goal.

But there is a fuzzy goal. That goal is to build a thriving dwarf fortress. This goal can be adequately achieved within the first three years of playing, after which the game is basically over.

There are objectives which the player may choose to obtain: Reaching maximum population, conquering hell, and getting a king/queen. These are what seem like achievements to me; unnecessary endeavors just to say we've done them.

Wiki is maintained by people, and people are fallible. You could always go change it.

You mean people like for instance you? No? Me neither. Just not a big fan of cleaning up someone else's mess.

Hey thanks for the link, that's an interesting way to further projects and help focus effort where the audience would most like it. Great idea, and I hope the fundry really takes off. I like the idea of pledged financial incentive to complete a feature, but that's a personal, psychological makeup.

I'm pretty sure that Toady wants to direct the development effort himself, and as such it would represent a conflict to him, and may even be perceived as a loss of control over the project. Why give up a financial stream to be replaced with a smattering of project requests that don't get finished? In case you hadn't noticed, there's a huge level of unfinished business in Dwarf Fortress, so the signed off Completed Feature thing isn't happening like it should in order to work well with the Fundry.

To get an idea of exactly what people would vote for, check out Eternal Suggestions