Hi!

Im making a empire builder (single-player, turn-based) similar to total war. How to avoid too much save-scamming in events such as spying, diplomacy etc?

I know i cannot avoid it fully, and that's fine i guess. But since I allow saves, should these events be fairly deterministic to avoid too much benefit of reloading until satisfied?

Lets say a spy can steal technology. So maybe not:
Higher level spy = higher % to suceed (since 20 % or 45 % spy sucessrate is boring if you can just reload until sucess anyway).

Maybe you always suceed but higher level equals more turns to complete it, or less anger from the empire you stole from?

Another idea is to have a delay: you start spying, and the sucess/ fail state is saved. But the spying isn't completed until 5 turns later. So you CAN reload if it failed but is it worth it to replay the last 5 turns? It deters some cheating but you are still able to do it.

(same with diplomacy, threats / haggling becomes boring if you can easily try anything without consequence)
(also, load or save takes about 1 second so no "help" from the inconvinience of slow load/save:) )

Remember, how it's done on Master of Orion. The spy works in the field for some time and at the end of it you get the result, however the roll for it has been done at the beginning. At the end you already have some 10 turns done and generally wouldn't be very interested in redoing it again just for one spy..

Other notion is that save-scamming is just a derivative of poorly developed feature. If mechanics for some event is so simple, then perhaps it is just minor feature not worth the attention. Or you can develop the spying gameplay further, without leaving it to random number generator.

Hi, 

I'd agree with what @Fandanguero said, however, i'd like to add that maybe save-scamming is a something that is totally up to the user. Many strategy games allow for users to save at any point and those who want a challenge tend not to scam saves, but disallowing users to do it, might not be prudent. 

Instead, perhaps offer an "Iron Man" mode, similar to Stellaris or CK2. That would disable saves  per say, and instead save at the end of each turn or choice, or just have it save automatically save at the end of each turn or choice, regardless. 

I think that having the ability (at least at first) to rollback your save because of a bad decision is great for learning the game without being penalised for learning.

Just my thoughts

I generally fall into the camp of letting players enjoy single player gameplay however they prefer to do so.  If that involves save scumming that's the player's decision and if they're that way inclined they will probably enjoy the game more as a result of being able to do so.

If you do want to discourage scumming however, personally I think the best solution is to make random events deterministic.  You can still have random chance of outcomes such as a spy that has 30% to succeed, but determine that result based on a seed that will remain the same if the game is reloaded.  The player may still save scum to undo their own decision they're unhappy with, but there wouldn't be random events they can reload until they get a favourable result.

For what it's worth, I used to play Freeciv quite a lot and it never even occurred to me to do that even though it's possible. And even now that you've mentioned it, I still wouldn't do it. On the other hand, Gish annoyed me so much by constantly sending me back to previous levels that I literally went into its source code to add infinite lives.

Point being, players only cheat when they find playing the game properly to be more tedious than going out of their way to cheat. So if a lot of players are getting through the (single-player) game by cheating, what you should look into isn't trying to prevent that cheating, it's fixing the gameplay so that you stop encouraging it.

And here's another way of putting it: if someone finds playing a game tedious enough to cheat, and you stop the cheating, what that's going to do is encourage the player to just stop playing the game. Fix the gameplay so that they don't want to cheat anymore, though, and you'll both keep them playing the game and discourage cheating.

FWIW I always leave cheats available in the games I develop. Usually these are put in for testing purposes and hidden away behind some sort of special key combination or command-line argument (or both). In fact my current (albeit still-on-hiatus) project literally lets you teleport all around the map if you want to with a very simple cheat code and some knowledge of file names. I don't know for sure because none of these games are super popular, but I don't believe the people who play them have a tendency to cheat much if at all.

I agree with your general points of savescamming and cheating in single player games.
But on topic: as I stated I still want cheating to be slightly less of an obvious option.

So it seems concerning my specific question, best options:

1. As I proposed, delay the outcome a couple of turns (but decide outcome from the start)
2. As jbadams proposes, make the randomness deterministic: e.g. store many dice rolls of a spy when that spy is created, and use these rolls in order, no matter when or if that spy levels up (which grants better success-rates). Is this what you meant?

Any other actual ways of doing it?

I think it would be best not to have randomized pass/fail events in the game at all.  Winning due to a coin flip gives no sense of accomplishment; losing due to a coin flip rightfully causes the player to blame the game and not themselves.  Gambling addiction notwithstanding, there is nothing fun or compelling in being at the mercy of the random number god.

Delay for the outcome (waiting 10 turns before you know the result makes savescumming less appealing).

Still, I would reconsider the % chance in the first place...

Another idea, you are  guaranteed to succeed but you lenght is random, so each turn you have a 5% chance to succeed (so there is no failure per se, only time till success).

22 hours ago, suliman said:

1. As I proposed, delay the outcome a couple of turns (but decide outcome from the start)
2. As jbadams proposes, make the randomness deterministic: e.g. store many dice rolls of a spy when that spy is created, and use these rolls in order, no matter when or if that spy levels up (which grants better success-rates). Is this what you meant?

Any other actual ways of doing it?

Regarding 2., as @jbadams noted, if the player knows that the outcome is deterministic (and they will know when they try savescumming ), they will then do another kind of savescumming: save the game, try to spy, if it fails reload and don't try to spy at all.

There is an alternative approach to this that goes in the other direction: increase the randomness (insert the insanity wolf meme here )
If spying is not the only random thing in the turn, but also resource production, battle outcomes, random events and everything else, then reloading a save will make the player lose all the good rolls that they got, just to save one bad roll (the spy roll). If you make enough rolls per turn, and make some of them be good and some of them bad, the player could get discouraged from trying to roll until they get all of the good rolls.

And another solution, one that is very user-unfriendly: make saving and loading as tedious as possible. Don't give the player any quicksave/quickload options, and make the loading be long and slow, and load the whole game state from scratch without any caching. If the player has to waste minutes loading the game multiple times just to save one spy, they will just give up

Well no I hate slow load/saves so that's not an option Also it seems too devious to add a "sleep" function just to waste the players time!

Increased randomness will be hard with my game, so I think i'll need to use one of the other solutions.

Thanks guys! (more input is always welcomed)