Strange Results

Strange Results (Full Version)

All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945

Message

yubari -> Strange Results (9/27/2008 7:00:33 AM)
Regular visitors to the AAR forum may be familiar with my game with Jim Burns. We have reached September 1943, and are on the verge of a carrier battle between the Japanese and Royal Navies. However, we have run into a problem. The situation is this. I received Jims turn last night about half an hour before I had to go out. As it was a crucial turn, I decided to run it to see the results of what I thought would be a carrier battle. This first running of the turn gave a very violent carrier battle where both the Indomitable and the Zuikaku were sunk. Feeling somewhat pleased with myself, I saved the turn in a separate slot and then went out, with the intention of running the turn again the following morning and then sending it back to Jim. Unfortunately, running the turn again the following morning gave a completely different set of results. The Zuikaku wasnt hit at all, it was the Shokaku that took three torpedoes. None of the British carriers were sunk either. To reiterate the problem; running the same turn on the same machine has given two completely different sets of results. Has anyone else seen this?

Nemo121 -> RE: Strange Results (9/27/2008 8:34:09 AM)
Yes, I've seen this once or twice when I ran a turn, saved it and then went back and ran it again. I couldn't swear to that being the only situation where it occurred but Im fairly sure. What I did and what seemed to work was to just download the turn file again and run the original file again. That usually resulted in consistent turn replays ( so long as no more saving occurred of course ).

Local Yokel -> RE: Strange Results (9/27/2008 10:21:01 AM)
I assume that by 'running the turn again' you mean that you generated a fresh combat resolution file in slot 001. Your first save into a separate slot presumably preserved the turn file as you received it from Jim. Might be interesting to get a fresh copy of that file from Jim and compare it with the original copy he sent to you, using a file comparison utility. If they are identical in content (as they should be), then it suggests to me that the seed value for the random number generation involved in combat resolution is affected by some external factor on the Japanese player's machine, e.g. system time/date.

yubari -> RE: Strange Results (9/27/2008 3:02:52 PM)
Thanks for the replies guys. You are correct Yokel, that is what I did. I saved the turn after watching the replay but without giving any orders the first time in slot one hundred and something, and then generated a new and different file 001 when I watched the turn the second time. As per Nemo`s suggestion, I went back and downloaded the savegame file and then ran it again, for a third time, and it gave identical results to the results from the second time that I ran it. These are of course different from the results from the first watching. Given that they are identical files that I am running, and I have of course had no chance to change any orders, I have no idea what has caused this. Yokel, your theory about the random seed somehow being changed seems plausible. Given this, does this mean that the Japanese player could rerun the turn multiple times until they get a suitably positive result? That opens a whole nasty can of worms for PBEM players.

jwilkerson -> RE: Strange Results (9/27/2008 4:58:30 PM)
I'd suggest rerunning the turn without pressing any buttons while animations are actually running. Not, the "done" key ... not the "esc" key. Once a given animation finishes, then pressing these keys is ok. But interrupting the animations can cause different results. That is the only way to generate different results to my knowledge.

pompack -> RE: Strange Results (9/27/2008 8:16:51 PM)
quote: ORIGINAL: jwilkerson I'd suggest rerunning the turn without pressing any buttons while animations are actually running. Not, the "done" key ... not the "esc" key. Once a given animation finishes, then pressing these keys is ok. But interrupting the animations can cause different results. That is the only way to generate different results to my knowledge. Joe: Are you saying that the JAPANESE player can achieve different results when GENERATING the master combat by manipulating buttons or anything else? [X(]

jwilkerson -> RE: Strange Results (9/27/2008 8:34:38 PM)
quote: ORIGINAL: pompack quote: ORIGINAL: jwilkerson I'd suggest rerunning the turn without pressing any buttons while animations are actually running. Not, the "done" key ... not the "esc" key. Once a given animation finishes, then pressing these keys is ok. But interrupting the animations can cause different results. That is the only way to generate different results to my knowledge. Joe: Are you saying that the JAPANESE player can achieve different results when GENERATING the master combat by manipulating buttons or anything else? [X(] Sure - yes I am - most people call this the "synch bug". But whatever you call it - interrupting the animations can cause different results. I always try to remember to avoid hitting buttons during the animations. I guess the good news is by comparing the combat report to the replay the Allied player can tell if they are different. If they are then the "synch bug" has struck - and the Allied player could conceivably ask for a replay (with no synch bug) depending on the house rules between the players. The other case where synch bug can happen is if various files are different between the players. This usually happens during the upgrade process. If the files remain different then you can see a synch bug every turn. This is not caused by the Japanese player hitting the buttons, this is caused by the files not being the same between the installs. The solution for this is for both players to make sure their files are the same. Worst case, both can reinstall everything. If most turns do not have the synch bug - then the files being are not the culprit. An occasional synch bug in this case would indicate the Japanese player hitting the buttons (either esc or done) while an animation is actually running. This can change the random number sequence and produce different results. If this happens and I was playing Allies, I would probably ask for a replay, though I would make it a house rule before the game started that the Japanese player should not hit the buttons during the replay so the expectation would be there from the beginning. I can say that I have played thousands of PBEM turns as the Japanese and to my knowledge every turn which had a synch bug was caused by one of these two issues. I can also say that very few turns I have played have had a synch bug. So it can be managed - just don't interrupt the animations. An easier solution is to turn the animations off - though most players do not opt for that solution.

yubari -> RE: Strange Results (9/28/2008 1:36:08 AM)
I am not sure if this is the synch bug in action. I am playing as the Japanese side, and the same savegame file has generated two completely different sets of results. For the record I didnt press any buttons during the animations, but surely pressing buttons during animations wont change the results of the combat in the Japanese players turn? As I understood, there was one random number seed generator that was produced at the end of the allied players turn when they saved.

Jim D Burns -> RE: Strange Results (9/28/2008 2:10:03 AM)
quote: ORIGINAL: jwilkerson Sure - yes I am - most people call this the "synch bug". This is why I wish you guys would change it in AE so turns are executed and saved after the allies hit the save button. That way they could email the file to Japan without anyone having seen the turn results first. Jim

Yamato hugger -> RE: Strange Results (9/28/2008 3:11:56 AM)
quote: ORIGINAL: Jim D Burns quote: ORIGINAL: jwilkerson Sure - yes I am - most people call this the "synch bug". This is why I wish you guys would change it in AE so turns are executed and saved after the allies hit the save button. That way they could email the file to Japan without anyone having seen the turn results first. Jim ??? Thats the way it is and always has been.

Jim D Burns -> RE: Strange Results (9/28/2008 12:31:53 PM)
quote: ORIGINAL: Yamato hugger ??? Thats the way it is and always has been. No, Japan runs the turn and results are saved to his PC from that first running of the turn, the only thing he gets from the allies is a seed number I think. The problem is, if you over-write the file with the allies emailed file again, you reset that first running of the file results generation and get to generate new results, as has occurred here with Yubari's and my game. Were the results created and saved to a password protected file at the end of the allied turn instead, then the results would be created on the allied PC and first seen on the Japanese PC. The allies wouldn't be able to watch those results until the Japanese player returned the file to him. So no chance of someone re-downloading and re-running the turn over and over to try and get a better result anymore. Yubari's mistake was an accident, but it does confirm the current system is exploitable in its current form. Jim

Yamato hugger -> RE: Strange Results (9/28/2008 12:46:20 PM)
Ahh, you want the allies to be able to do their turn, run the results and if they dont like what they see, re-load the save the Jap sent, change their turn, re-run, ect ect ect, right? I see a problem with this. Do you? < Yubari's mistake was an accident, but it does confirm the current system is exploitable in its current form. > No, there is only 1 result. The first one. Any re-run is just showing what happened, and if its different than the first, then chalk it up to sync bug.

Jim D Burns -> RE: Strange Results (9/28/2008 1:40:14 PM)
quote: ORIGINAL: Yamato hugger Ahh, you want the allies to be able to do their turn, run the results and if they dont like what they see, re-load the save the Jap sent, change their turn, re-run, ect ect ect, right? I see a problem with this. Do you? < Yubari's mistake was an accident, but it does confirm the current system is exploitable in its current form. > No, there is only 1 result. The first one. Any re-run is just showing what happened, and if its different than the first, then chalk it up to sync bug. If the turn executes (behind the scenes, no replay at this point) and saves into a password protected file after the allies hit save, there is no way the allies can look at it or see the results at all until they send the turn to Japan, Japan does his turn, and then sends it back to the allies. So the results can't possibly be re-done, so I don't understand your objection. From what I understand, Yubari got two different results, and could not then duplicate the original results when he tried to several times. Jim Edit: I just checked my emails and it seems I'm privy to some critical info Yubari failed to post here. He had to go back into the original save (the one he saved into slot 100 something) to get me a file with the original outcome. The new turns results were different. Perhaps it has something to do with him not saving it to the save slot we use or something, I don't know.

Jim D Burns -> RE: Strange Results (9/28/2008 2:03:42 PM)
quote: ORIGINAL: Local Yokel some external factor on the Japanese player's machine, e.g. system time/date. Having seen both results, I'd say it is weather. It appears that weather prevented all the strikes from launching and only one Japanese strike was launched in the new/different results. Weather affected earlier strikes as well, so target choices and whatnot were all out of whack by the time the carrier raids flew. So perhaps weather doesn't use the hard coded seed number and uses something else generated on the Japanese machine. Jim

Yamato hugger -> RE: Strange Results (9/28/2008 9:14:23 PM)
Are the results REALLY different, or is the replay just different? Are the damages to ships / bases different? Numbers of planes shot down different? I suspect they arent. The way I understand it, what you are proposing is exactly the way it is right now. A random seed is generated during the allied phase and its passed to the Jap with the save. If I am wrong on that, Im sure JW will say so.

yubari -> RE: Strange Results (9/29/2008 5:05:50 AM)
Oh heavens, I feel like I have opened Pandora`s box with this one. Essentially, I have discovered a method which allows an unscrupulous Japanese player to run the turn essentially an infinite number of times, and to obtain an infinite set of different results. Of course I won`t explain this method on the public forums, but if any of the programmers or moderators want to know about it, then then can PM me. Given how simple this method is to do, I am astounded that it hasn`t been discovered in the four years that the game has been released. For Jwilkerson, I have spent a very tedious morning running the same turn multiple times, doing various different clicks and pressing escape at different times to skip the combat animations, and at no time has this changed the real combat results. Although I cannot prove it, I am quite sure that this is not the cause of the problem. For YamatoHugger, yes the actual results are completely different. For the turn in question, I have run it to produce three different sets of results. In the original running of the turn, the CVs Indomitable and Zuikaku were sunk. In the second running of the turn, no ships were sunk. In the third running of the turn, there was a disaster for the Royal Navy where they had all four carriers sunk, and torpedoes into two battleships. Although I havent done it, I am quite sure that I could use the same method to get an infinite number of different set of results. Note that I am the Japanese player, and so always see the correct combat replay, and the results are confirmed in the ships sunk screen. Jeez, I wish I hadnt found this bug, as it is something that any Japanese player can do very easily, and affects the integrity of all PBEM games. You need to fix this for the AE. Again, if any of the programmers or mods want to contact me, I can send them savefiles and combat reports etc. I am sure that the same method would work on all machines with all games.

yubari -> RE: Strange Results (9/29/2008 8:13:49 AM)
I have been doing some more testing and I get the suspicion that it may be the weather that has caused this problem. If this is generated randomly at the start of the Japanese players turn, and I have unintentionally found a way to regenerate the weather with subsequent runthroughs then this may be causing the different results. What prompted me to think this is that various different runnings of the turn have resulted in a CV battle in the morning, or afternoon phase, or in one case with no CV battle at all due to bad weather. Just an idea. Yamato hugger, I have sent you some files.

jumper -> RE: Strange Results (9/29/2008 11:11:32 AM)
quote: ORIGINAL: jwilkerson Sure - yes I am - most people call this the "synch bug". But whatever you call it - interrupting the animations can cause different results. I always try to remember to avoid hitting buttons during the animations. I guess the good news is by comparing the combat report to the replay the Allied player can tell if they are different. If they are then the "synch bug" has struck - and the Allied player could conceivably ask for a replay (with no synch bug) depending on the house rules between the players. This is very important word..[;)] Actually I�m using ESC during animations all the time and it is not affecting the replay results at all so far. And because sometimes I need some more datas for my AAR I�m frequently running the same replay several times. No matter if I hit the ESC or not, it is ALWAYS the same. At least for me. The allied player sometimes informs about the out of synch issue, but I doubt it is cause by it, because we are out of sync even if spend the hour watching the endless bombing replay without "escaping". I understand that allied players are afraid that their japanese counterparts might run the turn several time and thanks to the magical ESC key got several different result from which to choose, but it is not possible IMO. Or at least it is not so easy. From time to time, when combination of unknown certain conditions are met (for example hitting ESC is probably one of them or another application running in the backgrount etc..), you may get 2 different results, but it is very very rare and you might spend the whole month trying this without any success. Only 100% sure way to get different results, what comes to my mind now, is to run the turn several times under different versions of certain files like PWHEX.DAT or WITP.EXE.

Yamato hugger -> RE: Strange Results (9/29/2008 12:59:03 PM)
quote: ORIGINAL: yubari I have been doing some more testing and I get the suspicion that it may be the weather that has caused this problem. If this is generated randomly at the start of the Japanese players turn, and I have unintentionally found a way to regenerate the weather with subsequent runthroughs then this may be causing the different results. What prompted me to think this is that various different runnings of the turn have resulted in a CV battle in the morning, or afternoon phase, or in one case with no CV battle at all due to bad weather. Just an idea. Yamato hugger, I have sent you some files. Its pretty much what I thought. Sync bug. The game files are identical (save 6 and 7). Only looked at 3 things to determine that: 1) Jap point total is 25,543 in both saves 2) Allied point total is 13,093 in both games 3) Yorktown only carrier sunk in both games. You havent found a way for the Jap to cheat.

yubari -> RE: Strange Results (9/29/2008 2:00:14 PM)
Sorry Yamato Hugger, I have just checked back at all of the save files I generated and I sent you two turns which were actually the same. I will send you another which actually is different, which is saved as turn 008. Maybe Jim Burns can give his side of the story because I have sent him some of the files as well.

Jim D Burns -> RE: Strange Results (9/29/2008 2:02:53 PM)
quote: ORIGINAL: yubari Maybe Jim Burns can give his side of the story because I have sent him some of the files as well. LOL my side... I already did my turn and sent it back. I even posted a lengthy AAR on the battle, so quit testing and get my turn back, I need a fix. [;)] Jim

Jim D Burns -> RE: Strange Results (9/29/2008 2:29:14 PM)
OK, just to show that Yubari has in fact proven that Japan can re-run turns, here are two separate results for the same turn. The top image is the results we are playing out, the bottom image is an alternate set of results he generated for the same turn. [image]local://upfiles/5815/0946185DB95B4F4E9B7D2666DDE2F49A.jpg[/image] And for those who simply do not want to believe their eyes, I gave orders to the Illustrious in the set we are playing out, so it isn't a ships sunk list delay or something that prevents it from showing up in the first set of results. Jim Edit: AE really needs to address this issue.

Yamato hugger -> RE: Strange Results (9/29/2008 3:28:27 PM)
The ones he sent me had the same points with different .txt files. I followed his directions trying to repeat what he is doing and it doesnt. For whatever reason his is an odd-ball that no one else seems to be able to do.

Yamato hugger -> RE: Strange Results (9/29/2008 4:42:43 PM)
Frankly it really doesnt matter if its a bug or not. It isnt going to be changed. They wouldnt re-write the replay routine for AE, I doubt seriously if its going to be done for WitP. Live with it.

madgamer2 -> RE: Strange Results (10/5/2008 1:48:14 AM)
Does this mean if playing the Jap AI that if I have the ANIMATIONS on but click them off while running the turn that it affects the combat results for that turn? Madgamer

Yamato hugger -> RE: Strange Results (10/5/2008 2:12:53 AM)
No. Sync bug only happens in a 2 player game.

Page: [1]

Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

0.84375