Civ Duel Zone  

Go Back   Civ Duel Zone > Site Stuff > Site Feedback
Home

Closed Thread
 
Thread Tools Display Modes
Old 29-04-2004, 22:13   #1
Kemal
 
Join Date: Mar 2003
Location: Netherlands.
Posts: 3,108
Default Analysis of Skyfish's saves

KEMAL – SKYFISH, A CLOSER LOOK AND EVALUATION OF PBEM IRREGULARITIES.


1. INTRODUCTION

As you are probably aware, in the past few months a PBEM game has been started and recently finished between me and Skyfish, agreed to be played as a “ladder match” on the Civ3 PBEM site of CDZ, meaning that the result of this game would count for the ranking system as used on this site for competitive 1x1 PBEM games. Spoiler pages and an open thread were also hosted by CDZ, and can be found at the following pages:

Open thread: http://www.civ3duelzone.com/forum/to...?TOPIC_ID=1070


Skyfish’s spoilers:http://www.civ3duelzone.com/forum/to...?TOPIC_ID=1080


Kemal’s spoilers: http://www.civ3duelzone.com/forum/to...?TOPIC_ID=1331


As is posted at the end of Kemal’s spoiler thread, the game has ended with me conceding the game and thus resulting in a win for Skyfish in this particular game.

Once the game ended, as is normal and accepted behaviour on this site, the players are allowed to access each other’s spoiler threads, to get a better insight on how the game has progressed, and see how the other player was doing and feeling about certain game situations.
Based on what I learned from the spoiler thread as well as what I already knew from Skyfish’s progress in the game when we were still playing, I got interested in how he managed to reach this situation where he was in total control of the game, as I had the feeling I had played very well for the entire length of the game, and was hoping to get to know some of the decisions he made and tactics he used, and to try and learn from these decisions to generally improve my abilities to play the game, and maybe put them to use myself later in future, other PBEMs.

To this end, along with his comments made by Skyfish in his spoilers, I, using the sent and recieved savegames still present on my PC, I replayed some of the IMO more important key moments in the game, using the admin-password to access Skyfish’s turns, to see how they affected the way the game progressed, and see if I could learn things from the way Skyfish handled these situations, and how I should have reacted when the game was played to have been able to be in a better situation than was the case in the original game.

In the process of evaluating some of these key moments, I on several occasions noticed that certain in-game situations severely deviated from how those situations would normally be expected to function under normal civ3 game rules, as well as a few moments where a situation originated in such a way that I have not been able to figure out how to reproduce it using the standard civ3 rules set as prescribed by the game itself, when applied to the situation of the game just prior to these events. Also, I noticed more than once inconsistencies and irregularities in how Skyfish described the game progressed at that time in his spoilers, and how the game was progressing for him at that time in reality.

Over the next few paragraphs, I will try to describe in more detail those situations that I’ve not been able to give a reasonable explanation for that have occured in the above mentioned game, as well as the effect these situations have had on the game itself. I will try as much as possible to use a chronological order for these events, except for situations that are closely linked together, which will be evaluated together in a single paragraph. I’ve provided savegames for each of these situations as well as as much info as possible to recreate certain event, for people to load, evaluate and replay these turns for themselves, something which I would strongly encourage anyone to do, to see and check where any of my conclusions and observations might have flaws in them.


2. LISTING OF UNEXPLAINABLE IN-GAME EVENTS.

2.1. GAME PARAMETERS

This game was played between Kemal and Skyfish as a PBEM game using version 1.27f of Civilization 3: Play the World.
The game was played on a large map, continents with maximum amount of land, normal/temperate/old game settings, deity difficulty, 6 AI opponents, sedentary barbarians.

Kemal was playing as the Vikings, Skyfish was playing as the Ottomans.

Map made and provided by anarres. Admin password: "admin"

2.2. CHRONOLOGICAL ORDER OF UNEXPLAINED EVENTS

2.2.1. TURN 1 AND 2

Summary of events: Irregularity between possibilities granted by the game to recieve certain bonuses from certain events and actual bonuses recieved by the Ottomans during these turns, as well as inconsistencies between the actual events taking place and the description of these events given by Skyfish in his spoiler thread.

As can be seen in the open thread as well as in Skyfish’s spoilers, at the start of the game a hut is visible one tile east of the starting position of the Ottomans, which according to game rules is opened when falling within a civs cultural boundary, resulting in a possible one-time bonus for the civilization that opens it. These results can give a variety of bonuses, but a single hut can only grant one bonus at a time, and only once.

As is stated in Skyfish’s spoilers here: http://www.civ3duelzone.com/forum/to...REPLY_ID=13335

Skyfish settled on the starting location with his settler, as we can verify by looking at the turn 2 of this game, and reveals that this hut at that time gave him a technological advance as bonus, that tech being “pottery”.
However, when replaying this specific turn as well as closer examination of some of the pictures provided by Skyfish in that given post, we can see that the hut, instead of giving pottery as a bonus, actually provides maps of the environment to the player as its bonus. This bonus is confirmed when opening turn 2 of the game, where the surroundings around Istanbul have been cleared of the fog, as well as by looking at the picture provided by Skyfish in his spoilers showing the american scout near his lands. On this picture, we can see 2 tiles of land being explored, those tiles being the ones located N-N-N-NE and N-N-NE-NE-NE of Istanbul, which reveal a pattern of exploration that can not be achieved in this game by using the (regular) method of unit-exploration, but only via map info granted by goody huts.

However, when looking at the map as well as the technological progress made by the Ottomans so far at that date, we can see that not only the surroundings of the map have been revealed, as would have been the expected result from the hut, but also that the Ottomans at this point have knowledge of the technological advance “pottery”.
There is only one opportunity at this moment in the game so far for the Ottomans to have been able to learn pottery in this game, that possibility being the hut east of the start granting it as its one-time bonus. This is also the way Skyfish describes in his spoiler how he recieved that knowledge. Since however, it can clearly be seen by playing the subsequent turn, as well as looking at the spoiler pics, that he has recieved map information from that hut instead of pottery, I have been unable to give an explanation on how this specific could have been obtained by the Ottomans using the standard civ3 rules.


2.2.2. Turn 5/15

Summary: The Ottomans follow an erratic path of exploration towards a nearby goody hut, and in turn 15 have obtained a new settler in their empire. In no way can the emergence of this settler at that time be explained using standard civ3 rules.


After building their first warrior, the Ottomans follow a path of exploration leading to a goody hut after a short time located E-E-NE-NE of Istanbul. At that time, the warrior decides to move in a southward direction from that point on to explore other lands, not opening the hut for understandable and logical reasons this early in the game. However, a few turns after turning southwards the only ottoman explorer suddenly moves back north again for no apparent reason, moving through lands already explored, towards the aforementioned goody hut. Then, after first passing it in turn 13, on turn 14 it apparently opens it, as turn 15 reveals the hut to be gone, and a new Ottoman settler located under the warrior that is then standing at the location of the hut.

However, when trying to replay this chain of events, after opening turn 14 (in which the hut was apparently opened, considering the situation in turn 15) there is only one action that the ottomans can undertake at that time, that being moving their only warrior. If this warrior opens the hut during this turn, it instead of a settler reveals a free warrior as its bonus. Since no other action can be undertaken by the Ottomans during this turn that could cause the RNG, that decides the outcome of huts, to change, the presence of an Ottoman settler in turn 15 instead of a warrior is a situation that I have not been able to find an explanation for using standard civ3 rules.

2.2.3 TURN 82

summary: Skyfish can spot Arab borders with his galley. Borders of Arab city of Shiraz have expanded due to an increase of culture that can not be achieved by using normal civ3 rules.

City of Shiraz has created a temple and expanded borders in a single turn, without loss of shields of their current building project.

City of Shiraz’s founding date from turn 83 onwards does not match the founding date given by the game to this city in earlier turns, as well as this (new) founding date not matching the events as they’ve unfolded in earlier turns.


As stated by skyfish in his spoiler thread here:
http://www.civ3duelzone.com/forum/to...REPLY_ID=16910

he mentions that his galley near the Korean city of Ulsan can spot Arab borders that it previously has not been able to see. This because of the fact that the city of Shiraz has, compared to previous turns, a larger sphere of cultural influence than before, allowing its borders to reach further, and thus able to be spotted from Skyfish’s current galley location.

Investigation of the city of Shiraz in turn 82b, so during Kemal’s turn, just before Skyfish’s turn in which he claims to be able to see their cultural borders, reveals the following situation in that city:


75.76*KB


note that the city has no cultural expansion at this time, nor any culture points accumulated yet, and is building a worker. Also notice the city’s founding date, currently at 1150 BC.

Investigation of the Arab city of Shiraz in turn 83b, one turn after the pevious pic, reveals this situation in the city of Shiraz:


80.6*KB


note that the city’s cultural borders have expanded, and that 10 culture points have been accumulated compared to the previous turn’s situation. Also, notice that the city now also has a temple completed that was not there the previous turn, while still producing a worker, and not having lost any shields in the city’s production bar.

Also, from turn 83 onwards, the founding date of the city of Shiraz has shifted from 1150 BC to 1250 BC. This date does not actually reflect the situation as it has eveolved in this game, since, when loading the situation of the game in 1200 BC, the city has not been founded yet (that happens in 1150 BC, as it orginally stated before border expansion).

Using normal civ3 game rules, I’ve been unable to find an explanation for any of the above mentioned events.

2.2.4. TURN 84

Summary: Skyfish contacts the Arabs as mentioned in his spoilers, via a galley that can be seen when he moves towards the arab borders. Replaying the turn and following the exact moves as skyfish has clearly described in his spoilers reveals no Arab galley in sight. Reloading earlier turns and investigating Arab coastal cities reveal no Arab galley nearby, or under construction at that given date in the game.

As Skyfish states in his spoilers here: http://www.civ3duelzone.com/forum/to...REPLY_ID=16910

He decided to risk his galley moving towards the Arab borders, and describes how he, based on his statements in these spoilers in his opinion fortunately, contacts the Arabs by spotting an Arabian galley in coastal waters there, allowing him to contact all civilizations in that turn, and get alliances and other favourable trades in place.

Replaying this turn, however, and following the exact moves as described by Skyfish in his spoilers, does not reveal any Arabian galley located at the position as in skyfish’s picture in his spoiler thread. Replaying using the Viking empire, who had a galley at that positon just 3 turns earlier, also indicates that no Arab galley will be near that location for a large amount of turns before or after the turn in which Skyfish spotted it, nor any of those galleys under construction in nearby coastal cities.

Using normal civ3 rules, I have not been able to find an explanation of skyfish’s statements and picture of the Arab galley at the location as shown in the picture in his spoilers, that can not be located anywhere when replaying these turns.

2.2.5. TURN 98 and 101

Summary: Skyfish twice obtains a leader in a war against the Americans, both in situations that I’ve so far not been able to reproduce, even though getting the exact same combat results. In one of the two situations, a leader emerges from a battle that under normal civ3 rules can never generate a leader.

As stated by skyfish in his spoilers here: http://www.civ3duelzone.com/forum/to...REPLY_ID=19200

And here: http://www.civ3duelzone.com/forum/to...REPLY_ID=19251

He twice gets a leader in this game, one used to rush the FP, obtained on turn 98, and one to rush Leonardo’s workshop, obtained on turn 101.

As for the first leader, though Skyfish has not provided any information about how he has gotten him, looking at the situation at turn 99 compared to turn 98, it can be deducted how and where he has recieved that leader in the game. Also, the exact results of all combat in the previous turn can be seen by looking at the status and location of his armies in turn 99 compared to 98.

Lets start by taking a look at the situation at the start of turn 99 for the Ottomans as it happened in the original game:

Eastern part:

55.08*KB

Western part:


60.21*KB

Clearly, by looking at the position of the leader and the way the marked elite units is positioned, it can be seen where and when the leader has been created.

Now, lets take a look at the Ottoman situation at the start of turn 98, the turn in which the leader is created:

Western part:


64.22*KB

Eastern part:


50.49*KB

After quite some time of testing I’ve been able to exactly reproduce the situation as it stands at the start of turn 99 in the original game, getting the exact same combat results for all 6 different battles that have taken place in turn 98, except for the fact that I’ve never been able to get the elite horseman that should have generated the leader, to actually generate one when replaying the turn. By testing some of my own older games, I’ve found that both the combat results as well as the generation of leaders depend on the same rng value, i.e. when doing the exact same combat you always generate your leaders at the same moment in the game.
This means that with the rng value that produces the set of combat results that can be observed at the start of turn 99, the elite horseman should always generate a leader in its battle.


As for how to reproduce these battle results:

- Wake the two horsemen north of Denver
- attack with the unwounded veteran horseman
- attack with the wounded veteran horseman
- wake the two horsemen north of St. Louis
- attack with one of these veteran horseman north of St. Louis
- attack with the other veteran horseman north of St. Louis
- attack with either vet horseman NE of Cincinatti
- attack with the elite horseman N-NW of Cincinatti
-
Of course, there might be the extremely small possibility that another rng value could produce the exact same combat values for all 6 battles that turn, and not generate a leader for the elite, however the even more unexplainable way the 2nd leader is created is not at all supportive for this implausable theory.

This 2nd leader has been documented by Skyfish a bit better, with his spoilers giving a picture of the situation where the leader is created, also containing the scores of all civs at that time. Looking at these scores, this turn appears to be turn 101, where apparently an elite horsemen takes out a Korean unit near Suwon, resulting in leader, a situation that turn 102 in the original game verifies, based on the position of the leader and the elite, renamed horseman.

However, when opening turn 101, this is the situation as it stands at that moment in the game:


92.01*KB


Notice that all details (score, build/growth progress of cities, location of units) on the pic are the same as in the pic Skyfish provides in his spoilers, except for the fact that no elite horsemen is in a position to attack the Korean warrior at the location that Sky provides in his spoilers, and that the position of the leader in turn 102 indicates, as the one where the leader is generated. Only a veteran horseman is able to attack this given warrior, which also in this situation does turn elite after attacking the warrior, and even matching the exact same amount of hitpoints after the battle as the horsemen that has created a leader according to Sky’s spoiler and turn 102 in the original game. However, since this horseman has attacked already this turn, it can therefore never generate a leader again by attacking another unit that turn. It can also not have gotten one defending against an enemy attack, since we can see that at the start of turn 102 the leader has already moved 3 tiles, showing that it was created during the Ottoman turn, and not during an interturn.


69.61*KB


Once again, considering the above and using the standard civ3 game rules, I am not able to find an explanation for both these situations at this moment.


Kemal






__________________
<b>\"In the Game of Thrones, you win or you die\" </b>
Kemal is offline  
Old 07-05-2004, 19:16   #2
ProPain
Customized Admin :)
 
ProPain's Avatar
 
Join Date: Mar 2003
Location: sailing the seas of cheese.
Posts: 5,852
Default

Skyfish,

Checking of the saves from this game has shown that these saves are edited. Same goes for another game you played that we checked.

We'd like to know if you have an explanation for these irregularities and if there are other games that might reveal similar fenomena.

We'd appreciate it if you would discuss our finding in an MSN chat with us.

PP and anarres.
__________________
I fed my Dog the American Dream
Well, he rolled over and he started to scream
He said, I dig the taste of salt but it don't keep me alive yeah, yeah
ProPain is offline  
Old 09-05-2004, 15:12   #3
anarres
anarchist butcher
 
Join Date: Mar 2003
Location: United States of Whatever.
Posts: 4,677
Default

Sky, we have checked with a hex editor and there is definite editing of the save files.

We now need your response on these issues before any decisions are taken.
__________________
<b>Calculate the probability of culture flips: Flip Calc</b>
anarres is offline  
Old 09-05-2004, 19:23   #4
ProPain
Customized Admin :)
 
ProPain's Avatar
 
Join Date: Mar 2003
Location: sailing the seas of cheese.
Posts: 5,852
Default

Sky here's an irregularity that we can't explain and that you can't have missed and we would like to see cleared up:

Turn 15b Sky vs PP:





71.07KB

No huts in view, no contacts with other civs. F6 looks like this:


121.91KB

Next turn pottery is discovered:


143.6KB

BUt also alpha has appeared somehow:


131.69KB

But no contact available:


127.25KB

Turn 17 research on maths is started:


124.6KB
__________________
I fed my Dog the American Dream
Well, he rolled over and he started to scream
He said, I dig the taste of salt but it don't keep me alive yeah, yeah
ProPain is offline  
Old 09-05-2004, 20:05   #5
anarres
anarchist butcher
 
Join Date: Mar 2003
Location: United States of Whatever.
Posts: 4,677
Default

Unfortunately my first game with Sky shows signs of external save modification too.

On turn 16 (3250BC) Sky is 10 beakers from Pottery (70 out of 80 beakers):


94.36*KB

You can see from this pic that the forrests are 3 turns from being chopped. This means that in 2 turns the forrest should be chopped:


55.34*KB

From Thebes you can see there is a maximum of 9 possible beakers, and that involves a riot next turn. Note that Sky mentions using a scientist worth 3 beakers to get there in time, but that still wouldn't give 10 beakers without rioting:


97.22*KB


In turn 17 (3200BC) Sky already has pottery when the turn is loaded, and is now only 24 beakers away from researching Mysticism:


93.36*KB


105.13*KB


Checking Thebes you can see the forrests are already chopped, which simply shouldn't be possible this turn:


99.13*KB


39.88*KB
__________________
<b>Calculate the probability of culture flips: Flip Calc</b>
anarres is offline  
Old 09-05-2004, 20:15   #6
anarres
anarchist butcher
 
Join Date: Mar 2003
Location: United States of Whatever.
Posts: 4,677
Default

We have contacted Sky about the irregularities in Kemals game today, although he left after a few minutes of conversation and did not return. We exchanged a couple of emails asking Sky to come and talk to us about this today, but Sky declined saying he would contact us soon to talk about it.

Sky has not so far had the chance to respond to the posts concerning PP's or my games with him.

We can see no possible explanation for these irregularities except that Sky has edited the saves either by hand, or by using another applicaiton such as C3MT.

If Sky wishes to discuss these items at all we will give him the chance to do so, privately if he wishes.

This thread is now closed. Please refer to the announcement thread in site feedback if you with to make a comment on this issue.
__________________
<b>Calculate the probability of culture flips: Flip Calc</b>
anarres is offline  
Closed Thread

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT +2. The time now is 10:15.


Powered by vBulletin® Version 3.7.3
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.