Why do aircraft cause drones when in base?

[Avenoct]

I'm probably a benchmark for the imperialist strategy, at least facing the AI. The most unpalatable thing I can imagine is merely repulsing an enemy..probably time and again.

But I find it pretty easy to run FM as far as the SMAC dynamic goes. There are just too many ways around the pacifist drone. Often the only time I start a turn with my units out of my territory is when a Vassal state gifts me their units on the way to battle.

I like the litmus test for imperialism/isolationism..perhaps we could think of a few more. From the data you could make a sort of SMAC personality test and match up players to their best possible strategies and factions. For instance "Which do you find yourself doing more often: Making lots of military units in advance to be prepared, rush-building units only in an emergancy, or keeping a few garrisons on hand but not extras?" Such a test could point to whether a player leans toward a Yang/Police type empire versus a Morgan/FM type empire.

I think the main thing I can't grasp r/e isolationism is how it relates to a global free market. I see imperialist states and their modern PC substitutes: the global corporation, the World Bank, and the enforcement of Human Rights by occupation, etc....as almost the highest form of Free Market. SMAC seems to add weight to this idea because Economy is so intimately linked to commerce. I just can't see that policies of isolationism strengthen a market tied intimately to global trade. Far superior (to an economy, not morally) would be highly controlled and stable imperialist ventures around the globe combined with WTO type free trade agreements. What am I missing?

If I'm not missing anything, Free Market should flourish especially when a player has troops in their submissive Vassal states, offering 'protection'.

-Smack

[King Stone]

I consider it a bug for a few reasons:
1) Its not documented and contradicts documented behavior. The units are in base inside territory, so they shouldn't cause drones.

2) Nearly identical units (with SAM ability) exhibit the documented behavior.

3) Bad decisions are still bugs. They may have know about it in testing but left it in.

Its a classic case of trying to force the player into behavior patterns. Weapons are neither offensive of defensive, they are potential. How you use them is what matters. It is a bug that you cannot have non-SAM planes ready to fly defensive sorties against worms and such and run patrol patterns around your coasts without having drones. Military in your territory should not cause resentment. The fact that the problem is easy to bypass is not a mitigating factor. From a game balance perspective it actually makes it worse that it is so easy to overcome, you may as well have not done it at all.

BTW, to was SP, it would be pretty tough to get people to accept a change to the planes in MP!

[Blake]

I feel it is not a bug, in the sense it is carried over from Civ2 days.
Then again like others here I dont think it is realistic, planes are often a clean and efficent way of dealing with 'problems' (in terms of loss of life that is). I'm sure free marketeers like clean and efficent solutions.

Democratic (Civ2) and Free Market citisens would be much less happy about a base crawling with infantry garrisons.


Then again lots of other stuff was a carry over from Civ2 and makes about no sense in the SMAC world, the key example being Weapons and Armour.

If you want a game where the concepts make a great deal of sense play CTP2 (with MedMod ). Unfortunately CTP2 just isn't as fun to play as Civ (this is due to 'look and feel' rather than stupid concepts)

[Basil]

Originally posted by King Stone
The fact that the problem is easy to bypass is not a mitigating factor. From a game balance perspective it actually makes it worse that it is so easy to overcome, you may as well have not done it at all.

There's one thing that makes it even worse than King Stone says. It may be easy for *human players* to overcome, but the AI doesn't know how to deal with it. I've seen AI's totally stalled for game-decades as they suffer huge drone revolts from the needlejets they've built, then start dismantling them for lack of resources, then decide to switch to Planned, then after a few turns in Planned decide no, they really want to do FM...

Agreed that since Civ II worked the same way (if I recall correctly), it was probably an intended feature and not a bug. I think it's good that negative police gives problems even in peacetime, but it would probably have been better to do that by giving penalties for having too many troops, rather than penalizing bombers/missiles and only bombers/missiles.

[King Stone]

As I said, bad decisions are bugs. If it is a holdover form Civ2, that is even worse. Stifling innovation for the sake of continuity is a very bad tradeoff. I don't really worry about the AI, in general computer players are uniformly pathetic in all games and will remain so until AI programmers get enough proccessor cycles and memory resources to do a good job, and the buying public starts to demand a better competition. Don't hold your breath.

[Adalbertus]

King Stone wrote:
... in general computer players are uniformly pathetic in all games and will remain so until AI programmers get enough proccessor cycles and memory resources to do a good job, and the buying public starts to demand a better competition.

I don't think that computing resources are the main problem for the AI. IMHO, you gain most of your turn advantage in the first 100 turns in SMAC. Here, computing resources are not as important as the thing that is really lacking: Time to finish the product, partly because of greed of the share holders , and partly because most people wouldn't pay the $200 per game for the salary of really good programmers doing a really good job. Which doesn't justify CTP2 crashing reproducibly every 10 turns

[johndmuller]

If you've ever tried to program a game, you probably discovered that just getting it to play by the rules is hard enough and that even a really stupid AI is hard to program. Coming up with a really good strategy in any complex game is definitely challenging; there may be relationships between game elements that were not part of the designers' vision which could make a perfect strategy harder for the designer to find than a third party. That said, it's always possible to improve whatever you've got a little at a time, by dealing with all the individual special cases you can think of and/or actually encounter; therefore, we'll never be quite satisfied.

With all the work that's been done on heuristics and neural nets and other stuff I don't know about, I would have thought that there would be AI's capable of independent learning available for consumer games by now. It could be kind of interesting if the time we spent in playing the game went toward training up our particular copies of the AI, for better or worse. Of course, then we could have them play each other (My Yang can kick your Yang's butt any day of the week!). A game with some really good "hooks" into the AI logic would open up all sorts of interesting possibilities for any of us with too much time on their hands.

[Fitz]

Just think if there was a file that kept track of the various units you designed, and actually made them available to the AI for use. That would work even better if it only added them permentantly to the list if you won the game. And if they were prioritized by how often YOU build them, the AI would have crawlers all over the map.

[johndmuller]

Do you think that the AI would know what to do with all those Chopper Crawlers?

[King Stone]

Take it from me, I work in the industry. Resources are the primary bottleneck in good AI programming. Most of your free memory is taken up by the art, music, sound effects and I/O of the game (from what Windows leaves you). AI is lucky to get 5% of the processor cycles on most games. Some of the more modern games (Deus Ex, Black and White, etc) have bumped that up to 10% or so, but they are still sadly lacking in the memory needed to implement a really decent AI. Check out gamasutra.com and do a search for AI, they have a few good state of the industry articles which come out every few months. Most of the fantastic advances you hear about are restricted to very high end (task specific) hardware and limited problem spaces. Chess is the classic example, a reasonably simple task with a manageable problem space, but even the best AIs have trouble beating talented players, especially if you remove their opening library, which is simply positional play distilled from a century of human masters.
Black and White has integrated some of the learning aspects of AI, that is the basis for the game, but it takes an indecently long time, and the frustration factor is very high. It will be a long time before AI is good enough to challenge hard core players, if ever. The advent of Multiplayer gaming has taken alot of the pressure off, the AI has to be just good enough for practice, the real challenge is in playing other people.

[Blake]

The SMAC AI is fairly solid, but just needs to be tweaked. Unfortunately these tweaks have never been done, making what should be a good AI mediocre. Some of the tweaks include:
Researching Cent.Eco very high priority
Building formers high priority in the early game
Terraforming changed so that anything worse than rolling/moist is forested.
Once Env.Eco arrives boreholes and condensors are given a small priority to be built. If possible code changed so up to 3 formers will build a borehole/condensor.
{Just those changes would give the AI a solid underlying terraforming, with probably twice the resource output of it's current "terraforming"}
Higher priority for building Rec Tanks, tree farms etc... those facilities which human players tend to build ASAP.
Wartime priorities changed so AI doesn't devote all it's resources to building military units.
{Those couple of tweaks would help the AI to keep up with the human, especially the latter}
Crawler code fixed so AI crawlers extract minerals from rocky tiles with mines. (The AI currently tries to extract minerals from any rocky tile.... which doesn't work terribly well on rocky/fungus, the crawler code is SO close to working)
Building crawlers higher priority.
{The AI already *almost* understands crawlers, a little more work and it would be able to atleast crawl minerals, making use of all those rocky tiles which it neglects to work}
FreeMarket set to be not used when at vendetta.
Aircraft never built under Free Market
Green given a higher priority (and planned lower) as empire gets larger (this threshold determined in a simialler way to buro drones). This change would also help to prevent the AI killing it's economy with inefficency.

Note that I arent talking about improving any "understanding" or "awareness" the AI has, just given it SENSIBLE priorities for building stuff. The AI can handle military fairly well, but lacks the "builder" side of things. Sadly this lacking is almost entirely due to bad priorities, they came so close to making the AI good, then gave up. (prehaps the AI programmer died, hahah)

None of the tweaks I suggested require additional processor time or resources, they are mostly changing a few constants and adding a little bit of "special case" code for dealing with well defined and understood situations. Indeed I would be willing to bet that given half-decently commented source code and a week I could tweak the AI to play a 'decent' builder game.
Apparentely this programmer-week was not a luxury Firaxis could afford. I guess I can accept that.

But the fact that the AI wasn't tweaked in SMAX (over SMAC) shows just how little priority AI gets - SMAX was little more than new graphics and voiceovers, barely no improved functionality at all.

Note that I am in no way saying I could program a better AI than firaxis, I'm just saying the one which shipped with the game could be improved dramatically. These improvments are based on a good understanding of the gameplay and strategy, not being an expert AI programmer. (This is why it doesn't bother me the SMAC AI was bad, at the time of release they didn't have a good understanding of the gameplay and viable strategies, the same is NOT true for SMAX).

[Yxklyx]

Going back to the original question. Wouldn't a good "workaround" be to keep your bombers stationed at an air base instead of in a city?

[Adalbertus]

Going back to the original question. Wouldn't a good "workaround" be to keep your bombers stationed at an air base instead of in a city?

I haven't tried it and I so highly doubt it will work that I won't try it.
The reason is that the drones are at the home base of the bomber and I see l little chance that rehoming a bomber to an airbase would work. And if so, this would be a real bug. Anyway, given the many initiatives in Germany (my home country) against any air traffic, bombers should produce more than only two drones in pseudo-Dem - pseudo-FM - real-Wealth

[Yxklyx]

I thought the bombers had to actually be in the city for the drones to appear. Didn't Civ II work like that?

[Yxklyx]

Double Post!