goody hut probabilities

[etj4Eagle]

Well there goes the simple idea that the game essentially rerolls the dice if a barb comes up for an expansionist civ.

In looking at Japan without building that first city, my data did not conform to that of a true expansionist civ. Instead it seems to support the hypothesis that the die is in effect re-rolled.

Here is the data (the meaning of the colored precentages is given at the bottom)

Tech: 15 (7.4%) 10.4% 15.2%
Gold: 25 (12.3%) 13.2% 10.2%
Unit: 28 (13.8%) 6.4% 7.8%
Deserted: 110 (54.2%) 58.8% 58.6%
Map: 25 (12.3%) 11.2% 8.2%

Percentages in blue are results from a map that refuses to pop barbarians
Percentages in Red are a subset of results from after the first city is founded, removing the barbarian component

In examining this data you can see that for the maps where barbarians are never poppped, the percentages act as if you have never built a city. In particular, the tech percentage remains ~10% verses the 25% for an expansionist civ.

Before that first city is built the ratio of goodies appears to be:
10:10:10:10:55 (Tech:Gold:Unit:Map: Deserted)

However once that first city is founded there does seem to be a slightly higher chance of getting a tech relative to the other non-barbarian options than before. Giving the ratio to stand as 15:10:10:10:55 on a 45:55 ratio that you will get something other than a hoard of barbarians.

For those uninterested in the beauty of the underlying rules here is a summery with everything you need to know for diety level:
black = non-expansionist after first city
blue = non-expansionist before first city
red = expansionist civ

Tech: 5+% 10+% 25%
Gold: 5% 10+% 5%
Unit: 5% 10+% 15%
Map: 5% 10+% 5%
Deserted: 25% 55% 50%
Barbarian: 55% 0% 0%

(the plus signs indicate that the actual precentage is a little higher, but I am only reporting data here in 5% increments. I would need a whole lot more data before I felt more comfortable than that. This is also why not all the precentages add to 1)

Well it looks like diety is licked, and now to turn my attention to regent.

 

[etj4Eagle]

Here are the results for Reagent:

Tech: 128 ~30%
Gold: 38 ~10%
Unit: 38 ~10%
Map: 49 ~10%
Barbarians: 149 ~35%
Deserted: 22 ~5%

Well my prediction for the barbarian precentage did not hold up (had predicted that it would be 25%), but that is not too surprising. I am a bit surprised though that the deserted camps only occur 5% of the time (as there is not much room left for it to decrease in going down to chieftan).

 

[sumthinelse]

Originally posted by etj4Eagle
Well it looks like diety is licked, and now to turn my attention to regent.

It's a good thing we finished because the Latin scholars cringe every time we write "diety" instead of the correct "deity" from "deus." We both sinned here. I wish we had a phonetic language. English spelling is so crazy

 

[sumthinelse]

It may be that there are 2 sets of probabilities (at least), scenario probabilities and "real game" probabilities. All my results are from real games because I do not trust scenario results for goody huts, at least not at this time.

It's kind of subjective, though. One way of looking at it is that "real games" have buggy goody hut results and the scenario results are the correct ones.

 

[etj4Eagle]

Well at least at diety your "real game" precentages and my scenario ones are very close. Especially, since I am only looking for precentages to the nearest 5%. In that case the values you posted on page 2 of this thread match those I have on page 3.

Of course there is still that issue of certain scenarios where the no-barbarians in goody huts never gets reset after building your city.

 

[Grey Fox]

I don't know about you, but I don't consider a Settler the same thing as a Warrior, so why don't you separate those two from the unit part?

 

[sumthinelse]

Originally posted by etj4Eagle
Well at least at diety your "real game" precentages and my scenario ones are very close. Especially, since I am only looking for precentages to the nearest 5%. In that case the values you posted on page 2 of this thread match those I have on page 3.

Of course there is still that issue of certain scenarios where the no-barbarians in goody huts never gets reset after building your city.

Yes, I believe the numbers you are posting here. I thought you were seeing an expansionist civ/deity that got 10% tech in one test (maybe I misunderstood), but I don't see those numbers here, so we agree.

I'm glad you worked on this with me. I think it helps to get closer to the truth when we have more than one head thinking about it.

 

[sumthinelse]

Originally posted by Grey Fox
I don't know about you, but I don't consider a Settler the same thing as a Warrior, so why don't you separate those two from the unit part?

I guess it was an attempt to get the complexity under control. Also, settlers do not seem to appear when you already have a settler. I think.

 

[etj4Eagle]

The other thing is that settlers appear so infrequently that it is hard to get good numbers on them. Also if there are various restrictions on when they will be produce, that can skew numbers. Along the same lines, we should also break out horsemen from the group as well.

However, back at the start of this project I did get a decent amount of settler data from a regent map that failed to produce barbarians. I actually originally had settlers as a separate option. But in examining this data, adding settlers to warriors came to a nice round 5% and at the same time settlers seemed to be about 5% of the unit production.

I would love to be able to examine the unit production ratios and rules and see if they also change via levels and the expansionist trait. Unfortunately this would requiring popping thousands of goody huts (at least on diety) to get a sufficient number of units produced. If I wanted a sample size of 100 units produced (a low min), that would require around 2000 goody huts on diety with a non-expansionist civ. I could cut that number to 1000 by not building a city, but if there is a rule that you can't get a settler if you already have one, well then we would be in trouble.

Though on some of the lower difficulty levels it may be possible to investigate this. And at least get an upper end limit to the chance that a unit will be a settler on higher difficulty levels.

 

[Grey Fox]

Originally posted by etj4Eagle
The other thing is that settlers appear so infrequently that it is hard to get good numbers on them. Also if there are various restrictions on when they will be produce, that can skew numbers. Along the same lines, we should also break out horsemen from the group as well.

Horsemen? You can only get Settlers and Warriors, no other unit. Unless you have that faulty patch...

 

[Bamspeedy]

I cannot say that I never get more than 1 settler because I have two recent games in V1.21 with two settlers.

Cracker - I've been thinking about this. I know, in one of those games it was GOTM8 and you used a worker. In the 1 rare game I got more than 1 settler, I may have used the worker, I don't remember. Perhaps there is a flag, that once a scout gets a settler, you can't get anymore, but if you use a worker, there isn't a limit?

Horsemen? You can only get Settlers and Warriors, no other unit. Unless you have that faulty patch...

I'm almost positive I have gotten conscript horsemen from goody huts with v1.21f. It's usually not until I have most, if not all the ancient age techs, though. (obviously I need to have the horseback riding tech). Nothing more advanced than horseman, though.

 

[kring]

Originally posted by Grey Fox
Horsemen? You can only get Settlers and Warriors, no other unit. Unless you have that faulty patch...

Ah, the first day 1.17 patch. I didn't consider it faulty; other than downloading the editor separately. I miss it.

 

[Ozymandius]

I have also had a horseman late in the game with 1.21 patch at Regent level

 

[cracker]

Bamspeedy,

I am sure that it takes lots of data to sort these probabilities out by experimentation.

For the record, the most recent GOTM event had a worker pop a settler out of hut1, then a warrior popped a warrior out of hut2, and then the conscript warrior from hut2 popped a second settler out of hut3.

So that data provides no chances to see if a single unit doing the popping might have an installed flag.

I do know that things would probably average out in the long run because in GOTM7 with expansionist scouts, I got virtually nothing from about 12 goodie huts. The yield was just maps and deserted for the most part.

I for one will be glad when this randomness is better understood, Sorry I have not been able to contribute more data.

 

[sumthinelse]

Yes, I am going to rename this thread "goofy huts!"

I don't think I have ever produced a settler with an expansionist civ, but I am not sure. I do know that I just tried 230 goody huts with Zulu/deity and got 15 warriors but no other units. This does not mean the probability is zero, but I think it's likely that in this case the probability of a settler is less than 1/50. Since expansionist civs at deity get a really big tech %, maybe it's fair for them to never get settlers. Or maybe these %s are "just the way it is."

I do get settlers from huts when I hit them with Aztec (militarist+religious) jaguars.

The trouble is the Aztecs were on a tiny world and the Zulu on a huge one. On a huge world it may be the case that there is more chance of complicating factors somewhere?

Has anyone ever gotten a settler from expansionist?

 

[cracker]

I have popped numerous settlers from huts while playing as the Irocs on all levels from Regent up to Deity. Also there were numerous settlers popped by various people in GOTM7 which was expansionist on Deity.

 

[Qitai]

Check Tournament elite 1-1, Russia, expansionist. You can get a settler if you settle the city and pop the hut east of the start loation in 3950BC.

I have a theory that you always gets settler if you settle city and then pop a hut on 3950 using units. Never failed me so far.

 

[sumthinelse]

Originally posted by Qitai
Check Tournament elite 1-1, Russia, expansionist. You can get a settler if you settle the city and pop the hut east of the start loation in 3950BC.

I have a theory that you always gets settler if you settle city and then pop a hut on 3950 using units. Never failed me so far.

Qitai you're right about it being possible, and I just got a settler with Zulu. I believe that in some situations, the probability is very low. Goofy hut probabilites for deserted and barbs seem pretty stable, but everything else seems to vary according to unknown conditions, especially settlers.

 

[etj4Eagle]

I have some initial results for the Chieften difficulty level:
Tech: 180 ~50%
Gold: 55 ~15%
Unit: 37 ~10%
Barbarians: 25 ~10%
Nothing: 17 ~5%
Map: 39 ~10%

I also modded the requirements for horsemen, so that they have no tech prerequisites and also added riflemen with no tech prerequisites. The idea being to investigate the unit breakdown.

Based on these results I found that only Warriors and Horsemen are produced (I am not sure why I never found a settler). And these are the rough chieften percentages between the two (when you can get build both):
Warrior: 6 ~15%
Horseman: 31 ~85%
Note that these numbers are very rough, as the sample set should probably be at least trippled before I say I have good values. And also with the small number of occurances so far, a settler with a 5% probability should only have shown up 1.85 times, of whice a deviation to zero is very reasonable.

I other thing, all these trials were done with Japan after waiting a turn from the founding of the first city. All goodyhuts were popped with MA.


Well that is the last of the data I will be able to post for two and a half weeks. Now its off to some job interviews and then volunteer as a scout leader for a week. And knowing my luck, the new patch will probably come out next week, continuing my trend of always being out of town when one is released (though at least this time I won't be at my alma matar, that was getting freaky with a 3 for 3 record).