Money Does Grow on Trees: How to Convert Forests into Lots and Lots of Gold

This is a remarkably hijacked thread.

Yes, clearly failgold with trees is powerful. You should try it on arboreal...

well, i decided not to touch balance because that's a tricky subject, but some of the units are just stupid (mothership/carrier/ultralisk), and there have been periods of MASSIVE imba/stupidity (4gates in pvp every single game until recent patch, unstoppable rine/scv rushes TvP in the early days, 5rax reaper vs z, infestors >>> everything until recent patch, EMP >>> protoss until most recent patch, terran having overly high winrate since beta), but to be fair blizz does attempt to balance the game and is getting ever closer.

Truly, my biggest complaint is with bnet 2.0 which IMO sucks despite all the pre-release hype. So yeah, you're right about that.


And I agree about civ4 having horrible UI / programming problems. I remember when I bought a new (beastly) computer and thought, "finally! I can play past 1000ad in civ4 without every click taking 2s to register". I also remember my feeling of disappointment when it still took just as long...

Keilah said:

I remember when I bought a new (beastly) computer and thought, "finally! I can play past 1000ad in civ4 without every click taking 2s to register". I also remember my feeling of disappointment when it still took just as long...

Click to expand...

Yep...and the bigger the maps, the more units, the more redonk lag gets.

So, sort of back on topic - MarigoldRan, what kind of lag do you experience in all of these huge marathon games that you play? Also, what's your processor and RAM?

This game is single threaded, no? Or am I confused with its predecessor?

Yeah, it's pretty sad that games with so much potential are wasted by their own developers so badly. Just imagine a Civ4 with a proper programmed interface, fast algorithms, slightly more clever AI (both AI leaders and gouverneur/automatization) and all the other stuff that could be fixed so simple and quick ... still, i love that game. It's one of those games that has a "soul", so i'm really deeply satisfied if i've wiped Shaka out Civ5 otoh has no soul at all. It's just plain stupid and makes me wanna kick the developers in the nuts.

ahcos said:

Civ5 otoh has no soul at all. It's just plain stupid and makes me wanna kick the developers in the nuts.

Click to expand...

Yeah, I heard that same song and dance about IV when I purchased IIIComplete circa autumn 2007. Now that I have purchased IVComplete earlier this year, I hear this about V. It's a shame that devs are using the "consumers" as their "beta" testing ground. This is fine and dandy when it comes to open source software. Heck, that's the whole point. But when it comes to "retail" versions, things like UI must work. Patches are supposed to be for polish only.

It'd be less of a problem if Civ5 was a good game under all the bad balance and poor AI. But it isn't. 1UPT failed bigtime for Civ, just doesn't fit the game at all, just like the bs they brought in the game instead of civics (keep forgetting what it's called... maybe because it's such a horrible game mechanic compared to civics) or the laughable Diplo. I had great fun with Civ4 vanilla when it just came out, because all the bugs and interface flaws there was good game underneath. Don't have that feeling for Civ5... my #1 game disappointment in the first decade of the new millenium. By far.

Regardless, I do understand that sc2 as a game works quite well in terms of gameplay and interface, and in any case does not have the terribad memory usage issues that Civ 4 and 5 have.

MarigoldRan said:

If Blizzard could make the AIs smart enough to beat most humans at Starcraft II, Firaxis should be able to make smarter AIs too.

Granted, Blizzard has A LOT more money to spend on development.

All other things is secondary to that.

Click to expand...

Actually I always thought Blizzard didn't give a damn about single player. The only reason the higher difficulty AI in sc2 challenges people is that they just give it rather massive bonuses, much like in Civ. In fact, dating back to sc1, I remember the battle.net page saying that the insane AI would school you in the art of playing Starcraft... until you found it it was just giving itself infinite resources. It actually can't even beat custom scripts that don't cheat sometimes.

That being said, the AIs do attempt to have some resemblance of trying to play the game and don't arbitrarily play by different rules.

Back to topic: To summarize, fail-gold from forests is really GOOD. Any objections?

Click to expand...

Yea, thanks for bringing this up. While I had thought of intentionally building for fail gold, this allows you to essentially exploit trees anywhere where you have a city. It's a neat micro trick, but my only worry is that someone has to actually finish the wonder. I guess that's not a problem on the higher levels.

ahcos said:

just like the bs they brought in the game instead of civics (keep forgetting what it's called... maybe because it's such a horrible game mechanic compared to civics)

Click to expand...

This was one of the chief reasons for the hate against IV. "What is this civic crap, [censorship] that!"

1UPT does blow. That's a totally different game a la Panzer General II.

I can't disagree with anything TMIT mentioned. I guess I just haven't examined Civ4 closely enough myself to notice the broken mechanics. Like I've never once cheesed an AP victory, just cus I haven't felt like it. I never automate my workers just cus I don't feel like it. You're right though, those things are badly broken. It just hasn't really affected my overall game experience. Smarter AI would, but then I'd lower the difficulty instead. They kind of make up for dumb AI by giving them advantages on harder levels. Not perfect but it works.

And the performance issues, absolutely on point. Civ5 suffers from a lot too, despite not really looking better than Civ4.

And yes, it has been a downward spiral. Civ5 release was atrocious the bugs. Even a year later, bugs may be fixed, but the game still sucks from a design perspective. That's why I keep playing Civ4 really. It's the best (and pretty much only) 4x computer game I enjoy.

As far as blizzard, they really outdid themselves with starcraft and wc3, and then world of warcraft, but you can see their quality slipping as well. It's impossible to stay on top forever. Their issues however are generally in design, not in software bugs. Because they have gobs of money now and a huge fanbase who will wait forever for releases, they can spend lots on QA and take a lot of time to release stuff. Quite a nice advantage. Personally I feel like Bioware has surpassed blizzard in terms of well designed, exciting games, but I guess we'll wait and see how the old republic stacks up against wow and pair diablo 3 up against mass effect 3 (totally different games but I think they come out around the same time).

ecuwins said:

This was one of the chief reasons for the hate against IV. "What is this civic crap, [censorship] that!"

1UPT does blow. That's a totally different game a la Panzer General II.

Click to expand...

Policies. They reason they suck is you don't choose anything. Like you can't go freedom and then have a revolt and change to tradition. You're just stuck. It's like a crappy exp talent system for your civ.

Civ5 really just lacks depth. They tried to make up for it by cheesing combat with 1upt. Oh man you need to have so much strategy with 1upt and flanking bonuses and crap! Who cares if the rest of the game is about 1/5th as deep as Civ4, it's all about the combat now! And the AI still sucks at it.

@TMIT: is it possible the unit-move slowness has something to do with multiplayer? I can imagine some sort of display-after-each-move is needed there, but maybe they forgot to turn it off for single player?

I've been hijacked!!!!!! NO!!!!!!

Which is fine. As long as you talk about Starcraft. I like Starcraft and Starcraft II.

BTW, has anyone here played Thief?

yah i loved the first few levels of thief where you are hiding from human guards but i hated running away from dinosaurs and zombies who can smell you or something.

Thief 2 is very similar, except pretty much all your opponents are human.

For SC2, the meta-game is changing. There's a lot less 4-gates for toss, and players are starting to really figure out the game and its timings. Of course, Heart of the Swarm will change everything again.

For my computer, I got the cheapest one at Best Buy a few months ago. So it's pretty new.

Thief 3 was a wonderful game, which I have played multiple times. Shalebridge Cradle, anyone? Thief 1/2 I've never played unfortunately - discovered the series too late.

But regarding Civ4 AI vs. SC2 AI... Civ4 can't use higher APM to compete with humans, true. But at the same time, it doesn't have to dynamically update planning if something new comes up in the middle of it thinking things through. Its pathfinding can be far simpler (more discretized map, no collisions). There are fewer unit-roles it needs to worry about - you need to move and control marines very differently from medivacs, differently from thors; you're going to need to write a unique AI for handling every single unit in SC2. But for the most part in Civ4 it splits into 1-movers, 2-movers, and siege, with only minor tweaks for unit-specific stuff. I could go on, but bottom line is: Civ4 is an easier problem.
And SC2's AI does just fine - below Very Hard difficulty, it's a non-cheating APM-throttled AI which actually has to play pretty similarly to how a human can. On Very Hard, it's not APM throttled (although it still doesn't get an APM much better than a human pro can manage), but still doesn't cheat. On Insane, it gets significant resource cheats (7 minerals/gas at a time instead of 5 IIRC), but otherwise plays by the same rules as everyone else.

you have to realize that since it's mathematically impossible for AI to "playout" the map to the win through simulating turns, that every programmer tasked with programming AI here will have to do mostly heuristic algorithms maybe combined with min-max (but I think civ devs don't use min-max at all).

That means that AI in CIV can't be better then the human who programmed it.

Btw just for the lulz... when I was studying programming I had there a course about AI and programming AI (well it was mostly only part of the course) and I have to say that I loved it.
I had to make some program which should apply some of the knowledge gained so i made easy scenario.

board with 10x10 tiles, each tile has "height". Peon attacking from up to down gets damage advantage. There was fixed number of peons randomly thrown around.

The analysis was min-max with evaluation function, the foresee was 1 turn (it was mid 90's and I didn't want to make it too complicated for myself).
Then I turned it on... and what happened? Well all peons went to their nearest hills and then just stared at themself. Technically it is ok behavior and probably the best, but doesn't make interesting watch and surely doesn't lead to victory of any side.

So there is always to strike balance between good AI and fun AI.

vranasm said:

So there is always to strike balance between good AI and fun AI.

Click to expand...

THIS

That said, it wouldn't hurt for somebody to turn on the profiler and find out what the CPU is doing between turns.

vranasm said:

you have to realize that since it's mathematically impossible for AI to "playout" the map to the win through simulating turns, that every programmer tasked with programming AI here will have to do mostly heuristic algorithms maybe combined with min-max (but I think civ devs don't use min-max at all).

That means that AI in CIV can't be better then the human who programmed it.

Click to expand...

Not true at all. An AI doesn't have to completely solve a game to just consider more options, or look further ahead than the programmer can (or does; some stuff is just too boring for humans). Deep Blue may not have solved chess, but it certainly could beat it's programmers. The problem is Civ4 AI is extremely greedy - it doesn't really do long-term planning at all. It just picks a tech it wants and goes for it, picks units it wants and builds them, picks a wonder it wants and tries for it... no coherent game plan for preparing to maximize the value of that tech, those units, or that wonder. No idea of where it's going to proceed after that point. It's not that the AI doesn't look ahead far enough; it's that even on the most broad strategic level, it doesn't look ahead at all. This is a particularly glaring problem at the start and end of the game - at the start because the AI's starts are full of wasted techs, builds, and moves; at the end, because the AI's endgame is basically "hope to stumble into victory." It might win SS, but only because it happened to get the right mix of cities to produce the parts, and happened to get the right set of techs to start it heading towards a SS victory. It might win UN, but not because it's been consciously trying to build up the support for that vote during the game and not because it beelined Mass Media specifically to get a Diplo. win; it can win because it's been a generally friendly sort to the other AIs out of natural inclination and map quirks, and happened to become a UN candidate.

there is always to strike balance between good AI and fun AI.

Click to expand...

Truer words were never spoken. It's been trivial for years to write a FPS AI that is better than any human - just always bunny hopping, spinning at max speed in circles while running from point to point, instantly headshotting anyone who comes into sight (maybe with some small map-specific knowledge to let it know where to worry about campers). Fairly nauseating to watch and absolutely no fun to play against, of course.
BWAPI led to an entertaining competition a couple years ago over who could write the strongest SC1 AI; the winning entry used muta micro. to micro every muta individually, as well as a bunch of other similar high-APM tricks that would simply not be fun to play against.
Any of you who have played with a group of players who know you're the strongest player in the group know one potential problem with a smart Civ4 AI - the more often you win against it, the less it should be willing to trade with you in future games, and the more ready to dogpile you, because you've demonstrated that you're a bigger threat than the other AIs.

Long story short, hearing that horn blow against you as you finish that final unit for your SOD is a magical thing

dalamb said:

@TMIT: is it possible the unit-move slowness has something to do with multiplayer? I can imagine some sort of display-after-each-move is needed there, but maybe they forgot to turn it off for single player?

Click to expand...

I don't know the logic behind it. It's especially bad in blazing time MP though, because you actually have to wait for the UI to catch up to you while the timer runs. In this sense nukes are particularly bad, should you reach that stage. It is quite literally impossible to launch 40 nukes in a single turn if you don't launch them in stacks, because the game will make a big animation + screen shake and then run the damage calcs (EXTRA slowly, though this slowdown happens when you nuke nothing too so maybe it's not the damage calcs after all).

In other words, unit movement should be instant for MP in particular, if the players have opted for an "animations off" game (which most would I'd guess), because otherwise the UI is bleeding legit play time from the players.

you have to realize that since it's mathematically impossible for AI to "playout" the map to the win through simulating turns, that every programmer tasked with programming AI here will have to do mostly heuristic algorithms maybe combined with min-max (but I think civ devs don't use min-max at all).

Click to expand...

Reading this one would ALMOST think that the AI actually has any hueristics that take it to a specific victory condition. In vanilla it had none. In BTS, it has some rudimentary culture strategies devised to the best of Blake's knowledge at the time and that is worth credit as it's often the only true threat to losing at all in BTS aside from RNG early screw DoW...

Of course, jdog5000 and BBAI crew managed to put that in. Now we compare:

- Paid professionals with years in a development cycle

vs

- Volunteers on CFC

Normally, you'd go with paid professionals, but the pattern for civ is actually against them . BBAI will make rudimentary but nevertheless significantly more effective plays at each of the VC...

That means that AI in CIV can't be better then the human who programmed it.

Click to expand...

It can be, if said human can find a way to figure out which hueristics to use. There are methods to do that.

Btw just for the lulz... when I was studying programming I had there a course about AI and programming AI (well it was mostly only part of the course) and I have to say that I loved it.

Click to expand...

I wonder how many things in civ IV and V code run counter to what you learned in that course . Certainly, diplo turn scaling does.

So there is always to strike balance between good AI and fun AI.

Click to expand...

It would have been a HUGE help if the AI in IV or V were to at least play the same game as the human is playing, and not made up pretend rules fairy land game where only it and the programmers knows its own rules. That should lead to fun and better AI; after all you're playing civ IV to play civ IV, not fairy friendship magic CPU land. If you need to drop handicaps on it (and it's reasonable to do so) then do that, but at least make it play the game! Maybe then it would need LESS of these bonuses...

That said, it wouldn't hurt for somebody to turn on the profiler and find out what the CPU is doing between turns.

Click to expand...

I'd be interested in this too, though I wouldn't even know how to go about doing it in normally, let alone in linux. I can tell you what though: I have WITNESSED the AI moving units late-game. Judging by how cathy moved her 50+ units into that city, I'd guess that at LEAST 1/3 of turns at that point are spent on the physical act of moving units, possibly more.