Testbed

From Stars!wiki
Jump to: navigation, search

Most experienced Stars! players spend a significant amount of time testing ideas and designs before they use them in a real game.

Testing Race Designs

The 'normal' testbed for race designs is a small / normal universe with ACC BBS on. For a regular race design in a regular game, the race should be able to achieve around 25,000 resources by 2450 in this testbed. For games that have other settings(eg ACC BBS off or Slow Tech on), the testbed should usually have the same settings as the game it's being tested for. Most players testbed with no other races in the universe, though some prefer to have a few AIs mixed in to give a better idea of how crowding would affect the performance.

the "std BB test"

Standard Testbed Battleship

Std BB testbed: alone in a small, packed, maxmins, accbbs, nre (comets!).

Std BB design: BB-TGD-organic-bears-3bsc-3jam20-OT-20arm missiles (can be buildt by all PRT/LRT combos).

Tech required to build 'em: 10-24-12-13-11-7

Build 100 benchmark BBs, the year you manage to do that is your benchmark. For example, a decent (non-CA) -f race can do that ~2455. There's HGs that do it in the mid 40ies ofc, and I've seen people claiming they do it in the low 40ies with special QS races. (Iztok)[1]

Testing Battle Plans/Ship Designs

Before committing a large fleet to battle, experienced players will attempt to simulate the battle. The BattleSim game setup is generally used for this. The test will be run multiple times to see the effect of different possible orders and fleet compositions - and for the enemy fleets too. When appropriate, you should test the effect of the presence of a third(or fourth) race - even a single ship changes the battleboard starting positions.

BattleSim can also be used for testing different ship design variations.

Testbed records

Raw resources by 2450 of a race in a test environment(not necessarily small/packed/etc):

This record is disputed by some players, alleging that he cheated.[2]
  • 192k by Varn, 21 Nov 1999 [3]
  • 156k by Asimov, 16 Nov 1999 [4]
  • See also [5]

Record vs Real

(Mark Constantino on Testbed records vs real game peformance - 27 November 1999)

On RGCS, in reply to Michael Birke, Mark Constantino wrote:

>Well, I do not take any of those Resource records serious. In a real >game, most of these races would have problems even to survive.

That's quite true. In order to optimize for the testbed you often take all tech expensive, NAS, and only enough mining capability to build factories (10/3/7 + OBRM). This is a death knoll in a player to player game because you need to get your tech high enough early enough to defend yourself -- mine enough minerals to be able to build ships, and have the scanning ability to tell what's on a planet. Frequently, minerals are the bottleneck to building warfleets, and with OBRM and low mining capability you're dead in the water.

A more realistic testbed is x resources by 2450, plus y tech levels, plus z minerals. If you can build a given warfleet by 2450, then you're in good shape. This usually means about 50K resources, 60K minerals or 300 cruisers, 15-20 starbases, and 10-14-11-9-7-3 tech levels minimum (but better if you have 14-16-12-12-11-4) in a no max minerals setting.


If these monsters we've been developing can be assured to be left alone for the first 50 years and have enough space to expand, then 150K resources will get them the advantage in tech and probably war capability -- but it's a great gamble for sure.


A decent -f race will have a warfleet by 2430 that will be able to destroy any of these testbed monsters with a blitz attack (before the monster can research and then build a warfleet) because HG's lack the Germanium to build enough ships, and they have most of their economy in a few key planets near their homeworld, which, when knocked out, will seriously cripple them. The key to early and rapid economy building is to concentrate your pop on the juiciest worlds; decentralization to avoid making fat targets limits the rate of economic advance (both in breeding and building factories).


A factoryless race will want to decentralize very early of course, and so aren't susceptible to a blitzkrieg attack because they present a whole lot of little targets instead of a few big fat and juicy ones.


But actually, the key element in a player to player game is diplomacy. If you can convince your neighbors to leave you alone, then as an HG or HP, you have a great chance of doing well in the game. But if you're neighbors are HG's or they are -f races, then the chances of them leaving you alone are smaller because in the first case you're competing for the same planets, and in the 2nd case your neighbor is out for blood early (or should be, because that's the whole point of a -f race).


>And as Asimov said, the best race/testbed creators must not be the >best players, too


The best players don't waste their time in testing useless scenarios like no enemies and max minerals. Needless to say, I'm not one of these people! I like the exercise as a conundrum that needs to be solved.


>The IT has strategic advantages that make up for the lack of >ressources.


I've just shown that an IT can get 130K resources by 2450, which combined with their strategic advantage will be more than a match for that 250K CA.


>For testbed records, I think his prescanning was ok.


There's not much difference in prescanning than in building 5-6 Smaugarian Peeping Toms in the first turn and scanning the universe that way (which is what you do when you prescan). Although I prescan because I'm counting the number of starting greens in a given universe (not enough, I regenerate the universe) -- so it's pretty much automatic that you would do this for this kind of testbed.

Mark Constantino

Links