Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[zakastra]

Its a fairly well known fact that one of the hardest parts of developing a good game is balancing, especially with a small dev team, there is only so much time you can spend play-testing every possible permutation of game mechanics and player interactions, especially with a game that is complex and has a large degree of configuration options.

The most useful thing that we, the community can do to help the development of the game is to give lots of feedback, and test lots of different things, and say what works, what fails, and what is ludicrously overpowered (and then see how that compares with the devs intent ^^)

When the game moves away form the Corvus sandbox incarnation and into the real campaign, it would be helpful to know what kind of fleets the average player can handle. what AI compositions are weak and easily defeated, and what makes a very well rounded, balanced and deadly fleet, capable of handling any threat.

so you are probably thinking, well what are you trying to involve the community in? Basically, this, build fleets, large fleets, small fleets, Wierd fleets, and see how they do, make up odd variants, and them GIVE THEM TO THE AI. Sit back and watch the fireworks, and then report in. The grand goal if you like, is to find the perfect AI fleet composition, the fleet that performs the best despite the limitations of the software, its almost certainly going to be different to a given players perfect fleet, but then humans play in ways designed to use the AI's flaws against itself, and when an AI is designed to exploit humans for being human... we tend to sulk and call it a broken, stupid, cheating AI ^^. Why would we want to find this ultimate fleet? So we have a "last boss" or achievement challenge for players to work up to that doesn't involve a hard-coded AI advantage or cheese.

Mass testing this kind of thing will probably get a lot of useful data on game balance and if lots of people watch the AI's duke it out might help spot weird AI buggyness and fruitloopery that can be fixed.

On the other hand if you think this may just be a colossal waste of time, let me know ^^

[Jazwana]

A good (and possibly fun, if there was a 2x speed option for tactical map) suggestion.

I haven't done any of this on a large scale but many of us have experience with refitting ships for improved AI performance.  It's also led to some bugfixes/AI improvements.

For example, a one-sided Conquest with minimal to no PD on the other side does quite a bit better than the default Conquest.

[phyrex]

A good (and possibly fun, if there was a 2x speed option for tactical map) suggestion.

I haven't done any of this on a large scale but many of us have experience with refitting ships for improved AI performance.  It's also led to some bugfixes/AI improvements.

For example, a one-sided Conquest with minimal to no PD on the other side does quite a bit better than the default Conquest.

huh ? how come ?

[K-64]

Presumably because it doesn't keep trying to flip broadsides because of both being equally powerful

[Alrenous]

The Conquest is currently affected by a bug in the broadside detection code - tried to search up the post, sadly my search-fu was weak. But basically, it favours the left(?) side, unless you entirely strip that side of non-PD weapons. Fixed for next patch. Of course, stripping the right side means its favouritism is moot.




I think the OP is a great idea. Sadly, I have no skill in modding. However, I'll happily test the crap out of any fleets anyone wants to give me.
I assume by 'give them to the AI' you mean make the player fight them? As you can't automate your side's admiral?

I would love to fight a broken, cheating fleet. See what it takes to defeat it, as long as I'm not being scored or limited in my own fleet composition. (Hate being scored - I make my own victory conditions, thank you very much.)

[Nooblies]

The broadside AI checks between front and sides for the most firepower, but as it doesn't compare left and right sides currently it will tend to favour the left side fairly outright (as that is the first side it checks). That was the explanation Alex gave anywho.