Geoscape Balance Discussion v19 Experimental Build 2

[Solarius Scorch]

DNK is right that the general difficulty should depend on the chosen difficulty and not the player's performance. Game too difficult? Try harder or start again at the lower level. Game too easy? You seem to have outgrown this particular difficulty, why don't you play insane.

However, I don't like the auto-resolve feature for ground missions. some of the reasons were explained by others, but I think the biggest one is that Xenonauts is a tactical simulation game. Removing or diminishing this element is simply not right; that would be like playing Sim City where only 30% of your time is spent building stuff and the remaining 70% is minigames with vehicles and such; these features add a lot, but it's a city tycoon dammit.

Having said that, I vote for an auto-resolve of air battles.

You shouldn't be penalized for doing more in the game at any point. If you progress the aliens quicker due to the player being 'better' you are treading a very thin line between challenging them and making sure only the true masochistic amongst us will play.

Precisely. I wouldn't like to sound too aggressive, but I haven't seen a single post that agrees with this or even considers it a somewhat viable option, while there were many protests. Which is a strong indication that most people probably don't like this idea; I certainly do not.

I think instead we should aim to implement a difficulty curve so that new players to the game and genre can win the game on easy most of the time without being beat into the dirt and then have the other difficulty options multiply the negative impacts that effectively speed up the 'ticker' that governs what weapons and ships/alien races the Alien commander has available to use. You are then effectively saying that on the harder difficulties there is less room for error. Once you get that nailed as a base level of difficulty then you can alter the tactics that the aliens use on the other difficulties and fine tune things for each difficulty. Such as working on the triggers at which the aliens will assault your base (they should be required to successfully scout it first!) or send out air superiority mission and the number of active UFO's that can be in play at a single time.

On a total tangent.. how hard would it be to let the player play as the aliens? Maybe you could track data from that and use that to help build the AI?

I'm not an expert, but I think enabling playing aliens in quick battles (as opposed to a real campaign) should be easy. However, I can't see how the gathered data can be effectively used to improve the AI. Analysing data from thousands (or millions) of battles would be an enormous task and Goldhawk has neither resources nor time to do so. I think it'd be a very interesting experiment though.

[Korgath]

I'm not an expert, but I think enabling playing aliens in quick battles (as opposed to a real campaign) should be easy. However, I can't see how the gathered data can be effectively used to improve the AI. Analysing data from thousands (or millions) of battles would be an enormous task and Goldhawk has neither resources nor time to do so. I think it'd be a very interesting experiment though.

My thinking there is that they track stats for other things relating to the human side of the war (Or does that button not really do anything?) and generally the player base itself is better at finding imbalances and exploiting them which could pave the way for the AI to use similar strategies.

In a game like this you don't want everything to be perfectly balanced, you want it to be perfectly imbalanced (See this video to see what I mean

) but the imbalance has to be something that the AI can react to and throw you off guard. So far every single mission I do is 4 rifleman + 4 snipers and that works for everything, period. But if aliens started hiding inside buildings or compartments within their ship more often it takes away my key advantage of having high damage, high accuracy, shots coming from outwith their reaction range. I'm willing to bet I could metagame the hell out of whatever the AI tries to do but right now it doesn't do anything to try combat my own strategy. Even if you can implement a rudimentary system that can monitor what the player does and respond with a varying degree of effectiveness depending on the difficulty setting it can change the game from being a dull rinse/repeat to something a bit more engaging; which leads back to my thought earlier.

If we end up with a game where the AI follows a distinct strategy and cannot adapt to what we as players do then games replay value dive-bombs significantly and players start handicapping themselves (TFTD Superhuman one base challenge, anyone? What about playing through Icewind Dale 2 on HoF mode with a single character?). Advancing the ticker based on what we do does help combat it but it's not a particularly elegant solution since you are setting up some very arbitrary variables that are still easy to metagame for the player once they learn the 'rules' of how and when it changes things up. Perhaps the ticker could also be linked to how 'smart' the AI is somehow?

If the game is set up with a slight imbalance between certain things and the AI can pick and choose from a selection of strategies that are then modified by the difficulty level selected it means that suddenly instead of whoring snipers I might need to throw 2 or 3 machine guns on my ground combat crew or stop the aerial combat from being a simple x counters y since the AI now chops and changes the UFO's deployed and combinations thereof.

I don't know anything about actually doing this within a game engine, I'm no developer, and I have no idea if Goldhawk have the resources to achieve an AI that can effectively monitor and respond to player input in any meaningful way but the more they do to prevent players using the same strategy over and over the better the game will be for any level of player skill. It would also help prevent the same tiring arguments between hardcore and casual gamers when it comes to game balance; the AI would be able to cope with both types of players as long as they pick the relevant difficulty that represents their own skill level.