Little Tiny Frogs

with Brad Wardell

Search

Forum Home | Login

Thomasaurusrex

Join Date 05/2005

0

Received 100 clicks from shared links to a single post

Active Commenter - Replied over 100 times

Novice Poster - Created 5 Posts

Weeks instead of months?

In one of the screen shots I noticed that the production time for a Defender (If I remember correctly) was displayed as week Instead of months? Does this mean that the game will progress by weeks instead of months or what? Im pretty confused so if some one could elaborate on this it would be wonderful. Thank you

Locked Post

Advanced Search

Watch this post

Do not email me updates for this post Email me updates for this post

Successfully updated karma reason!

GC is one turn = one week.

Successfully updated karma reason!

In GC2, the turns are equal to 1 week, instead of 1 month as they were in GC1/AP.

Successfully updated karma reason!

Ohhh.. so the game will be a lot longer then..

Successfully updated karma reason!

Not really. A turn is a turn, whether you call it 1000 years per turn or 1 day per turn.

What will make the game longer or shorter, in number of turns played, will be determined by the final play balance and the AI. Right now, it is my ignorant, beta playing opinion, that we will end with a game that is slightly longer in number of turns to end game (as compared to GC1), but that is due to the start taking longer in GC2. However, it does seem that once you are well established in the game, that you critical mass much more quickly then in GC1. So the actual play time of a game from start to end game (as compared to GC1) may well be shorter.

We will have to see what happens with the new features and smart AI, but it seems to me that basic framework is going to stay in place. Longer "starting" phase, and shorter "I know I've won, now to show the computer I've won" phase then GC1. That's just a function of the new planet tiling system. Larger to get started, but once you are, it will be easier to steam roll the AI. No matter how good Brad and crew make the map oriented AI, a human is going to be do better at the basic empire building, and that's why we customers will more easily kick its butt. Add in giving us the ability to custom design our ships just how we want them (so we can maximize our ships for our needs and play style), and while I expect GC2 to be much more fun then GC1, I think the AI will be more easily beaten by more players. You'll just outgrow it, and better utilize your tools of empire. (And that's coming from a rabid fan boy that thinks Brad and his people makes the best AI and game in the industry.)

Of course, it's only beta 1. Lots could change. But the new format is really going to emphasis world building and empire building so much more then GC1.

Successfully updated karma reason!

We are probably going to be tweaking a lot of the numbers in terms of costs of techs, ship components, etc, and population growth, etc. So don't assume that the start of Gal Civ 2 is going to play slower than in GC1/AP.

Also, regarding your statement that humans will always be better than computers at empire building, that's not necessarily the case. It's still a math problem, and computers are great at solving math problems. Remember how everyone complained at first in GC1 because the computer could get constructors built faster than the human and that it must be cheating? It was simply calculating the exact turn to purchase a constructor outright when the cost was optimal. So I wouldn't count on the AI being a pushover.

Successfully updated karma reason!

Hmm I dont remember the AI of GC1 being a pushover. Granted with a favorable start and the right mods I can win 100 out of 100 games, on gigantic maso, and I'm no stategic genius, but that is only because I can repetedly surprise the AI with a aliance win. I hope Stardock will make that impossible in GC2, having an aliance with 7 races all fighting each other is just not realistic. Aliance win is OK but it has to be a true aliance.

I do agree with Star Pilot that GC1 has the same flaw I have seen in all 3X games, from the time I know I have won to the AI realizes this takes to long, but GC1 is much less affected by it than most.

Successfully updated karma reason!

CariElf,

The computer is better at math. But it is still just an expert system code base. It isn't a learning system, so that means that it will always play the same way, given the same inputs. What's the most common inputs in a 4x game? Empire building. The AI will never be able to out build a human, because it will never be able to change its strategy on the fly, and flip from being a "bribe/flatter with gifts until I get SuperTech X" to "Focused build up my one super gem industry world so I can just horde death my enemies". We humans will have game after game after game to learn how the AI will play (ie, Drengin will always bully the minors, Altarians will always build for Research Win, etc etc etc), take that into consideration, and kick its butt.

The AI can calculate out the math for when its better to rush buy that little star fighter versus building it. But it cannot crunch the math that 2 big industry worlds, 3 medium research, and 4 tiny bank worlds is more effective then 1 big balanced world, 1 big research world, and 3 medium balanced worlds, and 2 farm/pop worlds and 2 "balanced" worlds. Unless the AI goes *back* and reputes its strategy and properties every few turns, it will lock itself in to whatever the conditions were at the time it needed to decide what to do. We humans will be refactoring our properties as we need, changing our short term and mid-term and sometimes long-term strategies on the fly.

In GC1, I need a moderately average start (compared to the AI in reach planets) to always beat the AI. In GC2, I will know starting up the game, I've already won for at the first 5 major passes of the game, after the beta ends. Why? Because my world building and repurposing will be better integrated with my map needs and strategic focus. Add in any little "cheese" I find that I like (so I don't think its "cheesy" to do so), and GC2 will most likely be easier to win then SMAC. And that's saying something.

Again, it's not my lack of respect for Brad and you and everyone else at StarDock. I think GC1 has one of the best AI fitted to its game universe. But it cannot learn from what I do in game, and therefore the code can never play a better game on its own. I, on the other hand, can see what the AI is doing to kick my butt in the games "we" play together, and therefore I can play a better game by imitating it and/or adjusting how I play my games in the future with it.

It's the basic problem with expert systems. An expert system is only as good as the code algorithms and its database of what to do. You guys end up making a better AI then most, but part of that is because you guys revisit your AI opponent system, and update its code with new algorithms and better tuning for its environment. Now, look at this... in GC2, you have added 2 new layers for the AI to have to be tuned to meet. Tile management and ship design.

Tile management has to be done at several levels to maximize its effectiveness. Just maximizing a world for money or growth or industry won't be enough. It will have to be integrated with the the short, middle, and long term strategy needs, as well as local, regional, and global main map needs! That's a lot of layering. Given enough time, I am sure that you people, being the brilliant and highly gifted code wizards that you all are, will get there, if you stay with it. But how long would that be? 5 years? It takes a long time to find the correct weights for the various inputs which feed into your expert system, and there's just a certain point that as a company, you'll pull off your "AI people" from improving the AI of GC2. You'll need them for other projects.

On top of that, you have a serious problem with ship design. It's easy to crunch the math to build a ship that maximizes for one area (ie, missile, drivers, beam), but it isn't easy to decide when to mix and match. And those are the tools that the AI will use to fight on the main map (ie, ship on ship). And those tools will further be mixed and maxed to make new attack kits (ie, fleets). That's a lot of direct inter-dependance between those three aspects, and that interaction will make it harder to tune each piece.

If GC2 had some sort of actual learning system in it, so that it could see the patterns of previous play and compare if the current patterns are favorable or infavorable, then I wouldn't be so confident that even someone as dense as myself will easily beat the computer without favorable placement or events. But since it won't, and we will... it's only a matter of time for the common GC2 consumer to being able to beat the AI easily. Any game that gives the player that much freedom is makes the game that much easier to beat the AI. Because the AI cannot learn, and it isn't perfectly tuned to its environment.

Look at SMAC. It's just Civ2 with custom units. That's all it is. I know people that cannot be Chieftain on Civ2 that easily crush SMAC at librian. My girlfriend, for instance. Why? Because the AI in SMAC is still the AI from Civ2. It isn't tuned for the ease of getting Air Doctrine (flying units). It isn't tuned to build a mass of cheap units. The AI in SMAC is better then Civ2, in terms of its actual behavior (it masses before invasions, makes better use of spies, etc). But that one added complexity, humans designing their own units, allows humans to easily and completely destroy the AI in SMAC. (The AI in SMAC uses preset unit designs, and doesn't try to build its own ships because that was too time consuming to meet their project design times).

Again, I'm not out to pick a fight. But there are so many new layers added to the game for the AI to deal with, that is going to split the AI Dev Team's time between all those layers, trying to make it get some decent performance/behavior out of those areas. That leaves less time to concentrate on figuring out how to make it play smarter on the main map. I don't recall hearing you guys hiring a whole team of AI coders... so that means that your current AI team, which is Brad (and who else?) is going to be spread around even thinner. And it will take a lot longer because so much of a player's strength in the game will be an emergence of his strengths across the various layers of strategy in the game. From basic tile management to Long Term Goals, that's a lot more then in GC1. The AI is going to be putting on some serious weight over the next year to year and a half. But after that? I know you guys really support your products, but with a GC3 possibility in just, say, 5 years from now, would there be anything other then Star Dockian professional pride to drive a continued effort in improving the AI as it becomes clearer and clearer what we customers are doing to more effectively and more effeciantly win while playing GC2? Or would that effort be better spent planning GC3? You guys really did jump from doing 1 to 2 quickly. From my simple outsider point of view, it would seem that you will likely repeat that... starting up on GC3 if GC2 proves at least as popular as GC1 was. Although you might take a longer break between games, I suppose. Doing the same thing, over and over, gets boring.

Just my meandering thoughts.

Successfully updated karma reason!

Welcome Guest! Please take the time to register with us.

There are many great features available to you once you register, including:

Richer content, access to many features that are disabled for guests like commenting and posting on the forums.
Access to a great community, with a massive database of many, many areas of interest.
Access to contests & subscription offers like exclusive emails.
It's simple, and FREE!

Sign in or Create Account

View all recent posts

Mark all posts as read Delete cookies created by the forum Return to Top

Stardock Forums v Server Load Time: Page Render Time: