Announcement

Collapse
No announcement yet.

A study of Hut Outcomes

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • A study of Hut Outcomes

    This study of hut outcomes was conducted on version 2.4.2, mostly on Deity/raging hordes at various map sizes. It used both reload testing and continous play testing. Both types of tests gave similar statistical results.

    The study was done in two parts. The first part was concerned with determining what factors affect the frequency of the five types of outcome: Tribes, Gold, Mercenaries, Scrolls and Barbs. The second part studied the specific results available from each outcome type.

    Much of the information reported here is already well known but some conclusions, hopefully, will be new and able to help resolve long standing disputes.


    Rules governing the five types of outcomes

    There are five possible outcomes from tipping a hut: Tribes (including nomads), Gold, Mercenaries, Scrolls, and Barbarians. The base chance of each of these outcomes is equal (20%), stated as a ratio this is: 1:1:1:1:1. However, certain game circumstances modify this ratio in predictable ways. These can be expressed in six rules.

    No Cities Rule

    When a hut is tipped by a Civ which has no existing cities and the Turn# is less than 50 (i.e. <1500BC on Diety), then both Tribes and Barbarians are suppressed and their chances are added to Mercenaries. Thus the outcome ratio becomes 0:1:3:1:0 in TGMSB order.

    If the Turn# is greater than or equal to 50, then this rule does not apply and the ratio is normal, 1:1:1:1:1. Under these circumstances, it is possible to tip a hut for a Tribes outcome and, on suitable terrain, have it become your capital - complete with a Palace.

    Near City Rule

    If the distance from a hut to the nearest city is less than 4 (as measured by the rectangle system or the formula for distance given in "Calculating Caravan Delivery Payments") then both Tribes and Barbarians are suppressed and their chances are added to Mercenaries. The outcome ratio becomes 0:1:3:1:0 in TGMSB order.

    Nomads Rule

    If a Civ has more than one existing nomad (NONE settler) per 8 cities, the nomads-result of the Tribes outcome is suppressed and the chance is added to Scrolls.

    With less than 8 cities a single NONE settler suppresses the nomad result. There must be at least 8 cities to support one existing nomad, 16 cities to support two, 24 cities to support three, etc. If not, then Nomads are suppressed and the outcome ratio on nomad-suited terrain becomes: 0:1:1:2:1.

    Continent Rule

    If a Human Civ tips a hut on a continent on which it has no existing cities (but does have cities elsewhere) and the distance from the hut to the nearest city is greater than 12, then the outcome ratio is: 7:2:2:2:2. In other words, there is a 47% chance of a Tribes outcome under these circumstances.

    If the Civ is an AI, then the chance of a Tribes outcome is 100%.

    Note, all rules are applied together. For example, under these circumstances if the terrain is nomad-suited and the nomad-result is being suppressed by the Nomads Rule, the Tribes chance is added to Scrolls and the outcome ratio becomes 0:2:2:9:2.

    Invention Rule

    Once a Civ acquires Invention, the Scrolls outcome is suppressed and its chance is added to Gold. The outcome ratio becomes 1:2:1:0:1 in TGMSB order.

    Explosives Rule

    After any Civ acquires Explosives, the Nomad result is suppressed and its chance is added to Gold. Thus, on Nomad-suited terrain, the outcome ratio becomes 0:2:1:1:1.


    These are, I believe, the only factors affecting the type of outcome from huts. More specifically, the following factors were tested and found to have no statistical effect on hut outcomes:

    - The direction of the tile from which the hut was tipped.
    - The turn# on which the hut was tipped.
    - The type of unit which tipped the hut.
    - The terrain of the hut (other than the Tribes/Nomad terrain dependency.)
    - The location or continent# of the hut.


    Part Two - Results of outcome types

    Tribes

    The Tribes outcome has two possible results: Advanced Tribes and Nomads. Terrain is the determining factor in which is selected.. Advanced Tribes occur only on Grasslands or Plains. All other terrain types produce nomads. Nomads can also occur on Plains or Grasslands when the surrounding terrain makes the site poorly suited for a city.

    The Nomads result is subject to the Nomads Rule and the Explosives Rule (see above).

    Before 1000AD, all cities founded from Advanced Tribes have a size of 1. After that date, the size of the city can be 1 to 4. It will always have a Temple. In addition, it has a 50% chance of having a Marketplace, 33% chance of a Granary, and a 25% chance of a Library.

    Gold

    The Gold outcome has three results: 25 gold, 50 gold, or 100 gold. These occur in a ratio of 4:10:1 on Deity difficulty level. Lower difficulty levels increase the chance of 100 gold and decrease the chance of 25 gold. After the year 1000AD, each of these amounts are doubled.

    Mercenaries

    There are two result classes of the Mercenary outcome. In general, these can be described as 1-move units and 2-move units. Each of these result classes has an equal chance of appearing from a hut. The specific types of units depend on the discovery of technologies. Once the governing technology has been discovered by any Civ, the unit begins appearing from huts.

    Initially, Horsemen and Chariots are available as 2-move units in a ratio of 2:1
    Polytheism introduces Elephants. The ratio of Elephant, Horse, and Chariot is 3:2:1.
    Chivalry replaces all 2-move units with Knights.
    Monotheism replaces all earlier 2-move units with Crusaders.
    Leadership replaces earlier 2-move units with Dragoons.
    Conscription replaces all 2-move units with Riflemen, a 1-move unit.

    Archers are the initial 1-move unit.
    Iron Working causes Legions to replace Archers as the 1-move unit.
    Gunpowder causes Musketeers to replace Legions as the 1-move unit.
    Guerrila Warfare causes Fanatics to replace Musketeers as the 1-move unit.

    Mercenary units are supported by the nearest city, unless the nearest city is foreign in which case they are NONE.

    Scrolls

    The result of a scroll outcome can be any unacquired Civilization Advance for which the Civ has all prerequisites.

    The chance of obtaining a specific Advance can be calculated by listing the position numbers of possible advances in an alphabetically-ordered list of all advances (as under @Civilize in Rules.txt). An advance's chance of selection is statistically equal to the distance between its position and the position of the advance above it (wrapping the list from bottom to top).

    For example, with no starting techs the list of possible advances and position numbers out of the total of 93 advances would be:

    Alphabet 2
    Bronze Working 9
    Ceremonial Burial 10
    Horseback Riding 37
    Masonry 48
    Pottery 66
    Warrior Code 88

    And the chances for each advance are:

    Alphabet 7/93
    Bronze Working 7/93
    Ceremonial Burial 1/93
    Horseback Riding 27/93
    Masonry 11/93
    Pottery 18/93
    Warrior Code 22/93

    Note that Ceremonial Burial, appearing immediately after Bronze Working, has only 1 chance in 93 while Horseback Riding benefits from the long gap between itself and CB.

    Barbarians

    (incomplete)

    General Conclusion.

    The assumption of randomness in the determination of hut outcomes appears to be valid. Although sequences of similar outcomes do occur, they are unpredictable and fall within bounds predicted by a random model modified by the rules and conditions described above.
    Last edited by samson; December 4, 2002, 17:24.

  • #2
    Nothing here seems to be contradicted by reported experiences. Good Job. How did you figure out the one nomad for each 8 cities? Is that per civ or all civs added together.

    RAH
    It's almost as if all his overconfident, absolutist assertions were spoonfed to him by a trusted website or subreddit. Sheeple
    RIP Tony Bogey & Baron O

    Comment


    • #3
      Another classic. Just added it to my "samson's threads". Yes, I searched all threads started by samson and put them in the aforementioned folder. I even think I searched threads where samson posted.

      I must be careful with the praise, like the proverb says:
      Flattery is nothing less
      than setting a trap.

      Must soon post an intelligent post in here, only been stupid and ignorant ones for a while. I'm not in it for the count.

      Comment


      • #4


        As rah says... seems like everything that has printed before except the 8 city and nomad stuff. I love how new stuff still comes up years later.

        Keep on Civin'
        RIP rah, Tony Bogey & Baron O

        Comment


        • #5
          Re: A study of Hut Outcomes

          Some questions/comments:

          Originally posted by samson

          Near City Rule

          If the distance from a hut to the nearest city is less than 4 (as measured by the rectangle system or the formula for distance given in "Calculating Caravan Delivery Payments") then both Tribes and Barbarians are suppressed and their chances are added to Mercenaries. The outcome ratio becomes 0:1:3:1:0 in TGMSB order.
          Is this to ANY city, or ONLY your own cities?


          Nomads Rule

          If a Civ has more than one existing nomad (NONE settler) per 8 cities, the nomads-result of the Tribes outcome is suppressed and the chance is added to Scrolls.

          With less than 8 cities a single NONE settler suppresses the nomad result. There must be at least 8 cities to support one existing nomad, 16 cities to support two, 24 cities to support three, etc. If not, then Nomads are suppressed and the outcome ratio on nomad-suited terrain becomes: 0:1:1:2:1.
          Comment: I think I'd change the wording here from "support" to something like "permit a new nomad".


          Continent Rule

          If a Human Civ tips a hut on a continent on which it has no existing cities (but does have cities elsewhere) and the distance from the hut to the nearest city is greater than 12, then the outcome ratio is: 7:2:2:2:2. In other words, there is a 47% chance of a Tribes outcome under these circumstances.
          Again, to ANY nearest city or ONLY to your own nearest city?


          These are, I believe, the only factors affecting the type of outcome from huts. More specifically, the following factors were tested and found to have no statistical effect on hut outcomes:

          - The direction of the tile from which the hut was tipped.
          - The turn# on which the hut was tipped.
          - The type of unit which tipped the hut.
          - The terrain of the hut (other than the Tribes/Nomad terrain dependency.)
          - The location or continent# of the hut.
          Did you test other difficulty levels, different sizes of maps, or 2.42 vs MGE?


          Nomads can also occur on Plains or Grasslands when the surrounding terrain makes the site poorly suited for a city.
          I was about to respond that I had at least once gotten a nomad on Plains (I think) when I saw your qualifier. Could you comment on what makes the surrounding terrain "poorly suited"?


          In addition, it has a 50% chance of having a Marketplace, 33% chance of a Granary, and a 25% chance of a Library.
          You might want to test to see if having Pyramids suppresses the chance of getting a Granary. In your spare time... Just for the fun of it...


          Polytheism introduces Elephants. The ratio of Elephant, Horse, and Chariot is 3:2:1.
          If Chariots is the "rarer" but more powerful unit, I am suprised that Elephants would be even more likely than Horse to show up once the tech is there. Might make going for PT early a good idea for early conquest games with lots of hut tipping.

          An advance's chance of selection is statistically equal to the distance between its position and the position of the advance above it (wrapping the list from bottom to top).
          I'd love to know how you figured this one out. You are the master!

          Comment


          • #6
            Rah -

            The 8 cities/nomad is per Civ. After the discussion of our different experiences with nomads in a recent thread, I figured it might have something to do with style of play. I play mostly OCC or perfectionist style games, whereas MP players are rapid expanders. It was just a matter of adding more and more cities until I started seeing nomads appear when I already had one. Then a little systematic testing to flush out the 8 cities rule.

            Lars, Ming - thanks.

            Elephant -

            Both of the rules that involve the distance to the nearest city mean the distance to ANY city, not just yours.

            I did test other difficulty levels. The difficulty level affects the amounts of gold, as I mentioned, and the number of barbarians. Map size appears to have no direct effect on outcomes other than by increasing the number of huts themselves. I confirmed some things on MGE, but original tests were all on 242.


            Could you comment on what makes the surrounding terrain "poorly suited"?
            I haven't been able to quantify it yet, but I believe it has something to do with the food production capacity of the city radius terrain. Perhaps the ability to support a size-N city. Don't know. Might be the same thing that makes the AI decide to found a city on a square or not.

            The relationship between the chance of a specific Advance and its distance from other advances on an alphabetical list was uncovered by logging the results of 100 scroll outcomes with no starting techs. Just like in the example, Ceremonial Burial turned up only once and tipped off the mechanism.

            Comment


            • #7
              Re: A study of Hut Outcomes

              Originally posted by samson
              Near City Rule
              If the distance from a hut to the nearest city is less than 4 (as measured by the rectangle system or the formula for distance given in "Calculating Caravan Delivery Payments") then both Tribes and Barbarians are suppressed and their chances are added to Mercenaries. The outcome ratio becomes 0:1:3:1:0 in TGMSB order.
              samson,
              In the past I have run a thorough test about the barbs popping from huts in the vincinity of cities. I think the 3-3 square (i.e. a square 3 steps diagonally from the city) is also barb free.
              But there are also different opinions. See my and the following posts in Barbarian activity question page 2.
              Civ2 "Great Library Index": direct download, Apolyton attachment

              Comment


              • #8
                ST -

                A hut 3 diagonals from a city would be at a distance of 4.5 which rounds down to 4 and thus is just outside of the barb-free zone.

                In the attached capture, an English warrior has tipped a hut exactly three diagonals to the east of Nottingham for a very nasty outcome.
                Attached Files

                Comment


                • #9
                  samson,
                  I am sorry, I have read your post too fast. My results are exact to yours.
                  Also I didn't realize that the barb-free area is defined by the distance from the city.
                  Civ2 "Great Library Index": direct download, Apolyton attachment

                  Comment


                  • #10
                    Originally posted by samson
                    I haven't been able to quantify it yet, but I believe it has something to do with the food production capacity of the city radius terrain. Perhaps the ability to support a size-N city. Don't know. Might be the same thing that makes the AI decide to found a city on a square or not.

                    The relationship between the chance of a specific Advance and its distance from other advances on an alphabetical list was uncovered by logging the results of 100 scroll outcomes with no starting techs. Just like in the example, Ceremonial Burial turned up only once and tipped off the mechanism.
                    First off, nice work. I started looking at huts, discovered a few things (the safe radius outside a city that produces no barbs, for example) and then got bogged down in all the potential variables and gave up. So for putting in the time and figuring it out.

                    I agree that terrain makes a difference - I've seen nomads on plains and grass with rough terrain. And I'm convinced that figuring out how the program decides nomad/tribe will also help us find out how city sites are chosen. I'm sure after a few more huts, you'll have that too.

                    As for the scroll thing... just brilliant Never in a million years would I have thought of that.
                    "I'm a guy - I take everything seriously except other people's emotions"

                    "Never play cards with any man named 'Doc'. Never eat at any place called 'Mom's'. And never, ever...sleep with anyone whose troubles are worse than your own." - Nelson Algren
                    "A single death is a tragedy, a million deaths is a statistic." - Joseph Stalin (attr.)

                    Comment


                    • #11
                      Once again, congratulations

                      SG[1]
                      "Our words are backed by empty wine bottles! - SG(2)
                      "One of our Scouse Gits is missing." - -Jrabbit

                      Comment


                      • #12
                        Brilliant.
                        The tech thingy explains why I so often get horseback riding where I only crave for ceremonial burial... I am sure that it could be used by some modders to give "interesting" outcomes of hut discoveries.
                        As for the terrain stuff, did anyone try to see if the chieftain-level advice has anything in common with the ai's placement of cities? In particular, does the ai look at all the city radius before determining whether it can settle or only the 8 surrounding squares, and do only the 8 squares affect the tribe/nomad outcome or the whole city radius?
                        Clash of Civilization team member
                        (a civ-like game whose goal is low micromanagement and good AI)
                        web site http://clash.apolyton.net/frame/index.shtml and forum here on apolyton)

                        Comment


                        • #13
                          ST - No problem, I appreciate your confirmation of the results.

                          SG - Thanks.

                          STYOM - Yeah, huts has been like the blindmen and the elephant ... everyone seems to have got hold of a different piece of it. There may yet be a few surprises left in there too. I'd like to figure out what kind of random number generator is at work.

                          As for the terrain thing, I agree with you: AI city site selection and the nomads/advanced tribe result is probably using the same algorithm. It seems to be food-production related.

                          LDiCesare - I have a number of plains/grassland huts that pop for nomads. And I've counted up the terrain in everyway I can think of, nothing concrete. Just a general feeling that low-food supply usually means nomads. I think you may be right that the Chieftain advisor is dispensing AI advice on the matter.

                          Comment


                          • #14
                            Yeah, I'll also bet it's the same threashold for building cities.
                            A minimum food/production in the city radius.
                            It's almost as if all his overconfident, absolutist assertions were spoonfed to him by a trusted website or subreddit. Sheeple
                            RIP Tony Bogey & Baron O

                            Comment


                            • #15
                              This is top notch......I've seen most of it before, but never so concisely and lucidly explained.

                              Comment

                              Working...
                              X