Nomads

[Blaupanzer]

The possibility that the player influences the probabilities related to huts by personal actions is Awesome! That 'keeping NONE settlers reduces the odds for getting another,' or 'researching in a certain pattern increases/decreases the odds of getting a particular tech,' or 'mercenary and money result possibilities wax or wane depending on the player actions related to other probabilities' would be stunning findings if confirmed. It would say that the designers and developers intervened with complex coding to specifically reward some play patterns and discourage others in ways far more subtle than previously realized. For example, the "riot factor" seem to be a design ploy to discourage too rapid expansion. However, the nomad factor in this line of research shows a reward to the player who immediately converts the nomads to cities, thus increasing the chances of finding and converting another (ICS). May be the law of unintended consequences at work. The idea may actually have been to increase the chance of getting another NONE settler if your first was killed. Very interesting.

Anyone actually in contact with the programmers of Civ II? Perhaps we could just ask them.

[Caesar the Great]

Just a random thought..... perhaps a nomad is an advanced tribe on a non grassland/plains square... I can never remember getting a nomad on a reasonable city square.... I'll test it out tonight....

[Caesar the Great]

Hmmm... apparently it didn't work, probably my test was flawed... entered a hut with a warrior, was a city, went back to a save, changed terrain to mountain and still gave me a city...

Ok... heres something... I opened 20 huts on a large map, only capital city, 10 advanced tribes, 6 units, 3 techs, 2 money, disbanded second settler, went to a new continent, three advanced tribes, then a tech, then a nomad, then tech, advanced tribe, barbs, tech. Another continent, advanced tribe, tech, unit, unit, barbs. Another continent, nomads.

Opened all huts on continent, 6 barbs, 4 nomads, 22 advanced tribes, 10 units, 575 gold total, 12 techs.

[Scouse Gits]

Originally posted by Caesar the Great
... Opened all huts on continent, 6 barbs, 4 nomads, 22 advanced tribes, 10 units, 575 gold total, 12 techs.

This we have to see!! Could you please either attach a save file to a post or e-mail this one to us?
_________________
The SGs well into the second

[War4ever]

Samson..... i think your odds are a bit high...... i don't question your results....i trust you to be an honorable civer and highly knowledgable. However in my over 10000 civ games (probably way higher) there is no way that 25% of my huts popped without a nomad produced a nomad.

However i do believe the odds of getting a nomad increase dramatically without a NON settler on the board. This is why i always use them as new city sites as quickly as possible.

I also think that different continents reset the hut seed. Therefore the chance of getting a nomad on any island is probably greater than getting two nomads on the same island.

I know for a fact they are unlimited on the poles...... which may help prove the "continent" theory?

If i think of something clever i will post it , until then i am stumped on my own testing

[samson]

War,

I agree with you than 25% is too high for the occurance of nomads in general. In my test the next hut was always popped right after disbanding the nomad. If using up the nomad is a trigger of some kind, then perhaps waiting to found a city with one until just before popping another hut might raise the odds of getting a nomad.

Also, I was popping huts in a specific order in the hut pattern -- lower left, upper left, lower right, upper right. Sometimes every hut in a pattern was nomad. The next pattern might yield none at all.

I really don't know if there's something here or not. It's perplexing.

[Scouse Gits]

The major problem with hut testing is controlling the variables in the game, to enable a reasonable deduction about which conditions affect the outcome. For example, we all know that a comparison between a hut popped at 4000BC with one visited in 1AD will be meaningless as so many things have changed during the first 100 turns. On a much more particular level is it reasonable to compare the results of two huts popped at 3850BC if different units did the opening? For a while I thought that better results came when a horse/archer was used instead of a settler. The hypothesis failed after 100s of tests!

On the subject of Nomads - my best score was 39% in a trial of 100 tests - using the same hut and reloading. (3850BC/2.42/Deity/Hordes)

------------------

SG(2)

[shade]

I think you ppl are making it all to complicated
do you really think that any programmer would write an algorithm of a few hundered lines for a simple thing like that if it could be done in 2 or 3 lines?
eg:
Gift:=random(10);//gives a random value from 0 to 9
case Gift of
0:give barbs;
1:give nomad;
....
9:give tech;
end;
If i were to make it dependend of other variables then at least it would be fixed variables(place of hut,direction of entry,year but not what you got from your previous hut and what you did with it)

=>here's another testing possibility
build a scenario and place your units next to each single hut,so you can pop them in 1 turn and see what happens when you reload and do exactly the same thing(even same order of opening the huts.

last but not least
1234567890
0879634152
6475809132
which is most random???

shade

[samson]

Shade,

There are six possible outcomes from a hut: units, barbs, techs, tribes, nomads, and gold. And there is certainly a random selection factor at work. But the probabilities of each of those of outcomes are not equal. If you think they are, you haven't been paying attention.

Furthermore, there are clearly conditions under which some of those outcomes are unavailable:

1) before you found your capitol, barbs, nomads, and tribes almost never appear.
2) after Invention techs disappear as an outcome.
3) on rough terrain, you get nomads but never tribes.
4) if you already have a NONE settler, nomads are much more rare than if you don't.

Obviously the hut-algorythm is nowhere near as simplistic as you would have it. Certain conditions are being checked in order to filter the results. If a "trigger" event occurs which alters the probability of an outcome, perhaps the predictability of the outcome immediately after that event is higher. This is the kind of thing we are investigating.

SG,

I agree about variable control. This is why I have been using CHEAT MODE in very artificial circumstances to test ideas. If any of the ideas seem valid under strictly controlled conditions then they can be tried in game play.

Latest test:

Playing as the white civ, I tried a bunch of starts until I got one with a hut that could be popped immediately in 4000BC in one move by the first settler, then saved the game. Then I loaded Civ2 onto two computers where it had never been installed. I copied the save to each machine, loaded it and popped the hut. I did 12 reload/pops on each machine, plus another dozen pops on the original machine.

The sequence of results was different on each machine. Since initial conditions were identical (even the history of civ-play on 2 of them), there seems to be a true random element involved in the selection of the outcome. However, each machine gave statistically similar results: 8 or 9 mercenaries and a couple of gold or techs.

[solo]

samson,

yes, the underlying probabilities become even more evident after a 1000 or more tips of the same hut, reloading before each tip. The percentages come out neatly divided. In your 4000 bc example, I forget the actual percentages, but it was something like 1/6 techs, 1/6 gold and 2/3 units for the results. (maybe it was 1/4, 1/4,1/2, Ribannah can tell you that I'm not to good at splitting percentages!) Even sub-groups like gold and techs had their own probabilities. For example, I believe it was 25g 1/6, 100g 1/6 and 50g 2/3 of the time. For the techs coming out of that same hut (starting with no free ones) I got the following percentages after 304 tech results in about 2000 tips (This was a while ago and I'm not positive about the 2000, as I was only interested in tech results at that time):

Pottery 18%
Alphabet 16%
Warrior Code 22%
Horseback Riding 26%
Bronze Working 7%
Masonry 10%
Ceremonial Burial 1%

This certainly is NOT random! After this I tried another hut on a different kind of terrain. The probabilities were roughly the same, although this time Warrior Code beat out Horseback Riding at about the same rate as above, with just minor changes for the other techs. Once a tech is discovered, a new set of probabilities comes into effect the next time there is a tech result. To those who say I get CB from a hut a lot, I'm sure they do, when the probabilities change as the list of eligible candidates changes.

In the same way, the presence or absence of NON settlers and/or nomads has a big effect on the probability of a nomad result, as does the terrain on which the hut is found. I don't believe other things like previous results, the game year, the unit doing the tipping, position in the pattern or direction from which the tip comes from are factors in the equation, and have to agree with shade that programmers like to keep things relatively simple, because I am one myself.

[shade]

Samson:don't worry I have been paying attention.and until now I haven't seen anything that could really prove any predictabillity.
I know the example I gave was very simplistic,but some of the ideas i have heared here are practically impossible to implement.I would not make it that difficult.
But now seeing your results(and those of others)i rather get the idea it might be something like this:
when you start your game(new game or load or reload)the computer calculates the number of huts(X),then ussing some standard info about your game it calculates the statistics(A% tech,B% mercenary,...)it could use things like mapsize,temperature,age,landmass,climate and year of start(this last one certainly included).Then you make a list with such a distribution
eg 1-> Tribe/nomad
2-> Tech
3-> Tech
4-> Gold
5-> Tribe/nomad
...
now you have a list with X gifts.
you can put them random in the list or go to a randomnumber between 1 and X when a hut is tipped.
then you get your gifts randomly but with the same statistics for the same startingconditions.

if you have this you can add some extra conditions => no units when they can't be supported(so no settlers without capital,...)
(I'm just guessing like you ppl)

[samson]

Solo,

I notice the seven techs you found at 4000BC with a no-tech start are the seven first researchable techs. I wonder if the techs possible from a hut at any time are always the same as those offered for research. And if so, is it the last offering or the next offering that is available from a hut?

If this was known, there might be better times to tip huts for techs. You might know, for example, that Warrior Code was unavailble and therefore needn't worry about getting it from a hut. This should be testable.

Also, what civ was used for the 2000 tip test? I wonder if the percentages of those techs match up with the civ-specific preferences that the SGs found for starting techs.

Another thing.
Is the distribution of results from single hut reload tests the same as the distribution from multiple hut one-tip tests? The latter, of course, would be what counts in game play.

[samson]

Shade,

Just because something is not perfectly predictable doesn't mean that the probabilities behind it can't be determined.

Look, we know that the program tries to compensate for poor starting locations by giving free starting techs. And that it attempts to handicap or help civs that are ahead or behind in the tech race by adjusting their research costs. Clearly there is play-balancing at work there.

Is it inconceivable to you that hut results might be used for play-balancing also? That the probabilities might be skewed towards what it thinks you need or don't need? Or in line with what a civ's style of play is?

Has anyone tested hut results using different Power rankings?
Or different levels? Are techs more plentiful at Chieftain than at Deity? There are lots of intelligent ways in which huts could have been used to vary the game. Complete randomness is only one of them.

[Scouse Gits]

Originally posted by samson

I notice the seven techs you found at 4000BC with a no-tech start are the seven first researchable techs. I wonder if the techs possible from a hut at any time are always the same as those offered for research.

Sorry, no - been there, done that!

I think there is a way to enhance the prospects of getting Advanced Tribes and Nomads. During a number of tests the odds were increased if you popped the huts in the same turn you had built your first city - or for nomads - two cities, so having no NON Settlers.

A hit rate of 64% for Advanced Tribes was once gained from 100 tests. For this you don't have to use both settlers building cities.
a) Found Capital
b) Find a grasslands/plains square at least 5 tiles away with a hut - I think it helps if there is a handy food special like wheat, and an Advanced Tribe is usually a good bet! (This can work well without any special) We have always thought of hut tipping as before or after the capital is built - but in the same turn seems to pay dividends. The awkward thing is that it works very well on some maps but not on others when the chances can fall to around 20%. As most of us agree, there are some other unknown forces at work.

--------------

SG(2)

[samson]

SG2,

That's consistent with my nomad testing. Up to a 50% chance of a new nomad if the next hut is tipped on the same turn as the last nomad is used up.

"Sorry, no - been there, done that"

You mean any tech is possible at any time as long as you have the pre-requisites?