Topic: Constructing a trainingscamp
teppo Topic Opener |
Posted at: 2014-03-10, 09:59
There are some maps where barbarians are really strong already. With those particular maps, barbarians would benefit.
I'll invite you, if I happen to see you in the internet gaming lobby. You pick barbarians, I pick the map to be played Top Quote |
fk |
Posted at: 2014-03-10, 11:59
It is too early to expect that there is going to be any change in the game, but if so, I will be glad to accept the challenge, provided that you will not pick the barbarians, and that I can veto the map. Top Quote |
einstein13 |
Posted at: 2014-03-10, 15:09
Settlers II (gold edition) is possible to download from polish site: http://redakcja.pccentre.pl/bigm/download/Settlers-2-Gold-Edition.zip (Probably in Poland the law is not so strict :P) In my opinion the economy in the Widelands is not so far from the Settlers. Version II or I. Only some resources are added, but main game is the same: roads, planning, map shape. The major difference is for me that the game is 32-bit and has more possibilities (more units can be deployed, bigger maps) And to the toppic: You should calculate, how long the Bootcamp and Trainingscamp will build. I think, that it will not help the Barbarians to win. They will have more problems. Longer time and more resources to build one 10/10 soldier. Edited: 2014-03-10, 15:20
einstein13 |
fk |
Posted at: 2014-03-10, 16:17
S1 is also 32 bit (and large resolution, press v), but enough said.
That doesn't solve the problem, i.e. that the Barbarians would be stronger before they have built a trainingscamp and are the only ones with attack/defence points. Top Quote |
SirVer |
Posted at: 2014-03-10, 21:22
Isn't it interesting? We have a proposed change and some people believe it makes barbarians stronger (because they can train sooner), some believe it makes them weaker (because they need to enhance a trainingsbuilding before taking full advantage of it) and some believe the change is huge while others believe it will not make a big change. And all we have are arguments, since we have no idea how to come by data. We could implement the change and play a game or two, but this will still not make a big difference though, because other factors could dominate. We could let the AI play several games and even compare a barbarians tribe with the change against one without the change, but again, the other factors are probably a too big an influence to be really sure about what actually made the difference. How can we solve this problem in a more principled manner? Top Quote |
einstein13 |
Posted at: 2014-03-10, 23:30
Probably most of us can't tell anything. For example my opinion is not so valuable- I usually don't play barbarians. We need a "voice of experts": some people playing only barbarians. They know if current game gives equal chances or not. Another thing is to find a map and players who are almost equal: one barbarian and second empire or atleanths. Sometimes one will win, and other time- the second player. Then we can give them a change for barbarians and see what happened then. But this experiment is a bit hard to make (long time of testing and how can we find people for that?) Also we can take a team of some barbarians and maybe empires ones The test will be quicker and more reliable. There are some maps, where the fight can be done soon and there are possibilities to expand a lot. einstein13 |
teppo Topic Opener |
Posted at: 2014-03-11, 03:00
That would be straightforward, I guess. With b19 still far enough, there would be time to roll back. Has testing like that ever resulted into a rollback?
A bit braindead, unless there is a way to loop AI-only games without human intervention, go get reasonable statistics. Top Quote |
SirVer |
Posted at: 2014-03-11, 06:08
no, never. And this is no surprise, we are usually none the wiser after implementing a change like this - there are too many other factors influencing how a game plays outs.
sure, there is a way. But it will not help much: running a full game takes at least 10 minutes and we ideally want thousand of samples to help against all the other influencing factors. But even then - how can you be sure that the AI has no systematic problems with a new change (for example figuring out when to upgrade a buildings is much harder then just building one). Top Quote |
teppo Topic Opener |
Posted at: 2014-03-11, 14:28
We seemingly have mixed opinions. Somebody should count the votes. I do not mind being in the opposition this time. Top Quote |
einstein13 |
Posted at: 2014-03-11, 21:36
According to the discussion above: Against (2): fk (gold issue) einstein13 (time used to build trainingscamp) For (2): teppo (his idea) SirVer (would like to check it) Am I understanding it wrong? einstein13 |