Save/Load Mechanism: Perhaps the Greatest Flaw of GalCiv

from Stardock Forums
The current save/load mechanism of GalCiv is perhaps its greatest flaw. Not only does it add a layer of confusion and frustration to what is otherwise a fantastic game, it also limits strategic choices of the player and narrows his options considerably. This phenomenon stems from several bugs and questionable game designs. Some issues I would like to point out here are:

* The research progress indicator bug forces players to save and reload their game every turn if they wish to minimize waste.

* Sector influence is reset upon game loading, making it nearly impossible to pursue cultural conquest if you do not have hours of time to spend on each game session.

* Load times can increase to excruciating lengths towards the end game on larger maps. Although this may seem to be a result of more units and colonies, I suspect it is also a consequence of AI routines being processed upon loading a game. In other words, instead of just loading the ships and colonies to display them on the map, the game also re-calculates the strategy and tactics of each AI. The amount of time consumed by this task then increases exponentially as more ships and colonies must be taken into consideration.

* Diplomatic actions are more likely to occur upon game loading since the AI routines seem to be processed at this time. In fact, they are almost guaranteed to occur later in the game. This puts the player at a severe disadvantage since the AIs are very likely to trade behind his back every time the game loads. By simply saving and loading a game repeatedly and examining the “debug.err” file each time, I have been able to trigger over a dozen diplomatic transactions within one turn. The only reason I stopped counting at a dozen was because I had already spent half an hour of my time just saving and loading the game and I could not tolerate it anymore. If I had persevered, I have no doubt the technology gap between the AIs would have completely vanished, and the treasury of minor races would be cleaned out by extortions. The free tech trading between the AIs without regard to the value of the techs involved only serves to exacerbate this problem. How would you like to receive my “Majesty” tech for your “Overlord Technology”? Riiiight...

* To further aggravate the diplomatic problems arising from loading, the game actually saves some of the AI decisions triggered by loading without your consent upon exiting the game. If game loading prompted the Drengins to extort you, you will be stuck with that extortion attempt no matter how many times you reload the game. While this may seem like a neat feature to prevent cheesy players from reloading the game to avoid unpleasant diplomatic encounters, it also has the effect of penalizing casual players without large chunks of play time because these compulsory saves seem to ignore good events and only recall the bad ones. Therefore the more often you load, the more often you experience bad events. If it is a design choice to save AI decisions upon exiting a game in order to reduce future calculation time, we must have some consistency so the AI will remember to do both good and bad towards or against the player.

To summarize this post, the bulk of my discontent with GalCiv stems from:

Bug: Research progress indicator
Bug: Sector influence reset upon game loading
Design Flaw: Long load times that may be a result of AI routines being processed upon game loading
Design Flaw: Free trading between AI to a degree that resembles cheating
Design Flaw: Diplomatic actions being triggered by loading games
Design Flaw: Inconsistency in what information the game is saving without player consent
1,783 views 36 replies
Reply #2 from Stardock Forums Top
I haven't tried doing any analysis myself, but I have also noticed the diplomatic initiations after saving/loading.

Good post!
Sins of a Solar Empire II
Reply #3 from Stardock Forums Top
wow. I never noticed any of those. I'll have to pay close attention going forward. Hopefully, these will be investigated and if true, corrected. Good skullduggery.
Having said that, however, I can say with certainty that my enjoyment of the game has not been affected. I have played and enjoyed (to lesser and greater degrees) many other games with far more unsettling bugs and game flaws.

Cheers.
Reply #4 from Stardock Forums Top
Frogboy has said that saving sector influence would make the game save/reload slower, so fixing one problem would make the other worse. My casual observation has been that it doesn't take more than 5-10 turns for influence to get back up to speed, so it's not a big thing for me. It cetainly doesn't take hours of play. Then again, I play on medium maps.

I suspect that in a game like this where you have tons of information and complex AI calculations, it is inevitable that gigantic games are going to load slow.

My biggest problem with the save/reload process is the bug that apparently causes some saved games to become corrupted and unplayable. I've had to abandon a game entirely because it will not load (CTD every time). All other concerns are trivial in comparison, IMO.
Reply #5 from Stardock Forums Top
What's the research progress indicator bug? I've noticed some odd behaviour, but I've not followed the boards closely enough to have heard about this one before.
Reply #6 from Stardock Forums Top
"My biggest problem with the save/reload process is the bug that apparently causes some saved games to become corrupted and unplayable. I've had to abandon a game entirely because it will not load (CTD every time). All other concerns are trivial in comparison, IMO."

Vanir, I've had this problem too. I sent my turn to CariElf. She could load it without any problems. She played one turn, saved and sent it back to me and voila, I could continue my game.
Maybe this works for you too. Just send it to a fellow player who might be able to open your save, play a turn and save it again.
Reply #7 from Stardock Forums Top
The research indicator bug is the one where if you make a change to the Research slider on the Domestic Policy screen, your changes may not be reflected accurately--or not be reflected at all--on the Research screen. Reloading the game updates the Research screen accurately, but also has other negative effects.

I have always liked the Civ3 solution of saving the random seed(s) in the savegame file so that reloading the game doesn't change the random events that will happen to you during the course of the entire game. That way you can't change the cards that are dealt to you, but you can change the way you respond to them.

GalCiv's 'solution' is to penalize you for reloading the game. The main reason I don't like this solution is because of the things ZaH mentioned, but also because it penalizes players for discovering a bug and trying to work around it. In a game with bugs, penalizing the players for finding them is a 'bad idea', IMHO.
Reply #8 from Stardock Forums Top
I agree that this is a big problem with GalCiv... the whole point of saving a game, and then loading it is to return to the same point - if I play contineously, or save and load a few times in the middle should not make any difference! Not to my influence, not to what the AI chooses to do, not to anything.

Saving and loading should be external and not effect things inside the game... if I load a game, it should be exactly as it was when I saved it... continuing a loaded game, should be just as if I had never stopped to take a break and saved it...

This annoys me about GalCiv (which is otherwise a superb game).

/unic
Reply #9 from Stardock Forums Top
to: Vanir, LordTheRon

I've had similar crashes/corrupted SG. If your problem is similar to mine, these events ALWAYS occurred immediately after having used a terror star (ie any save game in which I had used a terror star then caused a CTD when loaded). I managed to fix this via a codec update (http://www.stardock.com/files/ms_mp3codec.exe).

The crash this supposedly deals with seems totally unrelated (as referenced in the GalCiv support FAQ) and I saw no reason why this should work... but it did.
Reply #10 from Stardock Forums Top
It takes about 3 turns for cultural influence to be recalced correctly, it takes about 12 turns after that to win if you have dominating culture, so I can't see how this issue prevents a culture win 'unless you have hours to play'. Before I turned off the culture win I won several times this way, once directly off a reload as described above.

As for all the things mentioned above being a design flaw, I disagree: they are a design tradeoff. In order to have as much AI tracking and data interaction as the game does it either had to calculate its data on the fly or it had to have save files so huge that the current load times would be a blink of an eye in comparison. Or features would have had to be cut. Instead the Stardock team split the difference and came up with an engine that does a massive amount of dynamic calculations on the fly (i.e. on loading and on 'next turn').

It sounds to me like this is a case of some people doing massive amounts of save/loads, and then not happy with the results of that. I load a game when I start it, I don't load it again until the next session or unless the game crashes (which it hasn't for me for a good while, nice work CariElf & Co.). If you want to save/load compulsively to try to achieve certain results (whatever results) then you have to realize that you aren't playing the game in a normal manner, and thus you aren't in a position to criticize the game based on that manner. You certainly don't -have- to save/reload like this, it's a choice the player makes.

It should be noted too that there is a big difference between load times, and reload times. Starting up a game of GalCiv and loading a late-stage gigantic galaxy is middlin' quick. Reloading the same galaxy after that is much slower, probably because of all the deallocation routines that have to be run to clear out the older game before the new one is loaded.

The only thing in ZaH's post I agreed with was the research slider thing, personally I don't find it a problem because I just eat the waste and call it a loss, but I understand his point. It would be nice if that particular calc could be done each time the research screen is brought up, that way the player would always see an accurate 'turns to go' number.
Reply #11 from Stardock Forums Top
I am one of those players who goes for an hour or so at a time then quits, so the reload resets are something I learn to deal with.

My experience with the culture win is that there is 3-4 turn wind up to get the influence spread out again. As for the 12 month clock, mine seems to resume where it left off after the power up is complete. If it showed 4 months before the quit, it would show 4 again after the reload and the influence built back up. Unless the aliens find a way during those 3-4 turns to knock you below the threshold to reset the counter by getting more influence improvements or culture starbases, things should be dandy there.

The research slider is inaccurate on the turn of a reload, true. But, it fixes itself right up on the next turn. Research will progress as it progresses. I know some try to count down to the last point because excess is wasted, but I also just eat the loss.

The most valid point I have seen so far is the diplomtic initiatives. This is a major one. Instead of a diplomatic strategy being saved as part of the file and being reloaded as is, the AI diplo strategy is recalculated from the ground up every reload. This can dramtically change the balance of power, especially if the poor human had to bribe everyone to start the wars in the first place. But I think that this cannot be fully handled until the overall frequency and value of AI diplo deals is straigthened out in the regular course of the game.
Reply #12 from Stardock Forums Top
Vorlin, the times I've reload were due to:

1) Research slider telling me I'm getting something next turn when in fact I am not. I had plenty of resources to spare to achieve the tech, so why should I 'eat the loss' as you say? I should be able to reload without penalty, because it was not my fault that the tech was not awarded when it should have been. My technology choices are not random; they are due to strategy, and when I schedule a certain tech to be completed by July 2184 that means I need it July 2184, not August 2184. I use just-in-time planning extensively, and I have production queues that are expecting a certain tech to be available at a certain time, not a month later due to some stupid interface glitch.

2) Various problems such as a UP issue that makes no sense in the current context. For instance, when the vote for a galactic cease-fire came up when no one was at war, I reloaded for a more reasonable issue. I felt that was a reasonable thing to do because if the game were 100% finished then that glitch would not have happened. I should not be penalized for that. I abided by the decision of the new issue even though it did not benefit me.

I'm not for reloading to make the game easier. The game is already easy enough. That's why I suggest saving the random seed in the game file as a solution that has worked for other games. I do think though that there are plenty of occasions where reloading shouldn't be punished, not the least of which is turning off the computer and going to bed. ;)
Reply #13 from Stardock Forums Top
Za H, The AI has several times made offers like Overlord for Majesty.

Always from an aligned trading partner, when I was at war with a mutual enemy.

No problem with the trade itself, but shouldn't happen every time you load.
Reply #14 from Stardock Forums Top
Salviati:

I downloaded that codec you linked to, but it won't install. I get a message saying it can't find the correct file. I downloaded it a second time thinking it may have been a corrupted download, but it made no difference. There were no terror stars in my game, so it may not have been the same problem anyway.
Reply #15 from Stardock Forums Top
"I do think though that there are plenty of occasions where reloading shouldn't be punished, not the least of which is turning off the computer and going to bed."

Amen to that. It shouldn´t make any difference, if you play through the entire game in one session, or 3 turns at a time. Those who do the latter currently get a worse experience. :(

My impression is this is what happens: I pay an enormous sum to the Ai for going to war against a particular enemy. They will stay at war for a long time, *UNLESS* I save and take a nap. In that case, chances are they will negotiate while the computer is turned off and I am sleeping, and when I load they will make peace immediately. :notsure:
Reply #16 from Stardock Forums Top
I should calrify that when I wrote "cultural conquest", I was thinking of planets flipping in addition to the 90% influence cultural game win. Due to how sector influence is calculated and accumulated over time, planets are far more likely to flip through influence if you don't reload the game and reset it.

If we take 500 influence difference as the minimium value where a saving throw for rebellion will be made, and look at a sector with 100 base human influence and 50 base Torian influence, it'll take 10 turns before you even get a chance to convert a Torian planet in that sector. More likely than not, the Torian would build some influence resistance improvements or what not and further delay the rebellion. If you happen to load the game before the planet if converted, then all your progress is lost, and you have to wait for the influence to re-accumulate the next time around.

Now maybe I have completely misunderstood how influence triggers planet rebellions and perhaps only base influence is a factor, but my experience in game tells me otherwise. The longer I play, even by just clicking the turn button and doing nothing else, the more likely an alien planet will join my empire voluntarily on any turn. For anyone trying to cultural assimilate an alien neighbor, the wait time involved in influence re-accumulations can be fatal to his strategy as the aliens will have more time to react if this player ever reloads his game.

As far as the programming difficulties with regards to influence calculations are considered, I really don't think they can consume so much time to affect game performance. We might have a recursive formula for sector influence calculations, but I'm quite certain it can be converted closely, if not fully, to a state function if we examine the math in detail.
Reply #17 from Stardock Forums Top
Political state information should be practical to save (I would think) from game session to game session. But I can't say that definitively because I don't know the code routines involved (i.e. I don't know how much of the diplomatic part of the game is also dynamically calculated). As noted above, I don't reload much so some of this stuff I can't comment on well because I've never seen it. If the AI's radically switch political stance (war->peace, peace->extortion, etc.) during a reload then I'd consider that a problem.

I wonder if the diplomatic issue are occurring because the AIs are working with bad data? We've all see that it takes one turn or more (the first turn is the most important) before the game's data 'settles down', but if the AIs are doing diplo stuff on reload then they are working with bad info (wrong influence values, wrong economic values, etc.), which would then explain bad or erratic choices on their part.

If diplomatic relationship info can't be stored in the save file for whatever reason, perhaps a viable solution would be to delay the AI's diplomatic recalculating until the 2nd turn of a newly loaded game? This might not be possible either, but if it is possible it might make the reloading of a game result in a situation more closely resembling the one you had when you saved it.
Reply #18 from Stardock Forums Top
Only noticed the sector influence one, BUMP though.

~SDC~
Reply #19 from Stardock Forums Top
My problem with research is the other way around (if I understood you folks right). When I reload, my techs take longer to be researched. Quite frustrating. But I'm confident in the future patches...
Reply #20 from Stardock Forums Top
Your techs are taking longer to research because their progress is now being calculated correctly. Had you hit the turn button without a reload, you may end up not finishing your research even though you were told it would be done "Next Month".
Reply #21 from Stardock Forums Top
There's also a bug whereby research that is supposed to be completed in "1 Month" (i.e. the turn after the next) will actually be completed "Next Month". This appears most often when you have some astronomical research bonus from resources. The game doesn't seem to handle the calculations correctly and may underestimate your research investment.
Reply #22 from Stardock Forums Top
Hmm, I haven't run into that one yet.

The sad part about it is that save/load issues are a fundamental part of the game that could easily be explained away as a 'design issue' and therefore not likely to be changed even if prone to problems. Most companies would not even think about touching something like that after release. Stardock might be different though...
Reply #24 from Stardock Forums Top
This becomes IMO more of a problem with score being tied into play time.

eg. Two theoretical identical games except for reloads, the one that doesn't reload will have the higher score.

That's not right.
Reply #25 from Stardock Forums Top
I can't see how reloading a game could possibly cause a significant difference in scoring. A few points at the most, if even that. You'll see far more difference just from the effects of random events. Get a nice economic boom and have that double-research event happen and watch the game fly towards a tech victory. Get a 10 year recession and watch it crawl. Losing a couple of turns to a reload recalcing the games previous values is a drop in the bucket compared to what random events do.