Soren Johnson: The Chick Parabola

Does anyone else like this solution (not sure if its original or not)?

Keep a hex or square tile world map, that's not so important. Reinstitute stacks, but limit units per stack to say 10 max. THEN when any battle takes place, it takes place on a new (let's call it strategy map) consisting of Hexs, or even octagons based upon the terrain of the world map tile.

Each army would deploy, at one unit per hex/oct tile of this strategy map. Then they would move in turn to engage each other. This would allow flanking units, early archery units, artillery units, infantry, tanks, airstrikes, you name it, ALL to have sensible and valid uses. It would also make it in principle much easier to program a decent ai opponent.

It would stop the (and lets face it, however you dress it up it's ludicrous) fact of a battle where an Archer can fire 130 miles, a horse unit can dart in 240 miles as a flanking manouever etc.

It would also allow much easier army coordination for the ai, and would probably vastly speed the whole game up. (Outside of the battles that is).

For those who didn't like to fight battles as such, there could be some auto-calc button for the battle, which took factors into consideration, and produce a winner.

Let's face it, battles have always been fought between armies. This approach would reinforce that fact, while allowing tactics and strategy in combat, and keeping the 1upt in battle. Trying to use the world map, as the same map as the battle map, and limiting 1upt, is I'm afraid an idea doomed to failure at the concept level.

If you're going to make CIV basically a war game, then at least have a seperate battle map, and make it a decent war game.

Thoughts anyone?
 
Does anyone else like this solution (not sure if its original or not)?

Keep a hex or square tile world map, that's not so important. Reinstitute stacks, but limit units per stack to say 10 max. THEN when any battle takes place, it takes place on a new (let's call it strategy map) consisting of Hexs, or even octagons based upon the terrain of the world map tile.

Each army would deploy, at one unit per hex/oct tile of this strategy map. Then they would move in turn to engage each other. This would allow flanking units, early archery units, artillery units, infantry, tanks, airstrikes, you name it, ALL to have sensible and valid uses. It would also make it in principle much easier to program a decent ai opponent.

It would stop the (and lets face it, however you dress it up it's ludicrous) fact of a battle where an Archer can fire 130 miles, a horse unit can dart in 240 miles as a flanking manouever etc.

It would also allow much easier army coordination for the ai, and would probably vastly speed the whole game up. (Outside of the battles that is).

For those who didn't like to fight battles as such, there could be some auto-calc button for the battle, which took factors into consideration, and produce a winner.

Let's face it, battles have always been fought between armies. This approach would reinforce that fact, while allowing tactics and strategy in combat, and keeping the 1upt in battle. Trying to use the world map, as the same map as the battle map, and limiting 1upt, is I'm afraid an idea doomed to failure at the concept level.

If you're going to make CIV basically a war game, then at least have a seperate battle map, and make it a decent war game.

Thoughts anyone?

That's not entirely unlike Call to Power's system. You could stack up to 9 units, if I remember correctly, and battles opened up a separate window. Now, that window didn't allow you anymore control over the battle than in canon Civ games, except that like Civ Rev, you could withdraw. It did stack ranged units in back, infantry in front and flanking units at the margins.
 
Crown of Glory and Forge of Freedom do this. You move large units around on a stratey map. When enemies are in teh same province/hex/tile, a fight can occur. If it does, you can resolve it instantly, or via a detailed hex-map battle randomly generated. It works well.

On the strategic map, stacking is unlimited though you have to play for supply when you get large concentrations. At the tactical level it is 1UPT.
 
Although I agree about the difficulty to create a good combat AI for a game like Civ with its countless units, hexes/tiles and whatnotever, the calculation of the computer opponents could be done during the human's turn. After all, there is multi-core support now.

All what would have to be done is to check whether any action of the human changes the related "threat-level". As long as this does not happen (not having met the human or the human actions are not considered threatening for the respective AI opponent) the calculation can just go on.
That is only partly true and even less, later in the game. You forget the AI not only have to count in YOUR actions, but also react on actions of the OTHER AI=players in the game. The more AI's there are, the more you have to WAIT and the LESSER the AI can predict pre-calculate. So i think you underestimate the problem, or overestimate the pre-calculations the AI can make. 8 civ game = 7 AI's and you. That means seven time a AI-civ calculation and then you the player, every turn. And then you have to add the civ-states also.


Count in the fact thay you, the player have not much to do to begin with other then pushing "next turn" for the first 100/200 turns, you leave the AI little pre-calculating time in the early ages for starters.
 
Having battles resolved on a different map is an option, but if it's MoM-like where you can control your units, it will just make the ai worse. If it's CtP like, then it's good because you can't outwitt the ai tactically on this map and still it makes use of combined arms and the ai can be taought that quite easily.
 
In a game with unlimited stack per tile, the tactic is very simplified compared to 1upt.

With unlimited stack, the AI can consider all units as one, and evaluate the value (good or bad) of a target position in complete isolation.
The AI can just calculate all tiles reachable with the slowest unit in the stack, give "points" for each end-state, and go for the one the score highest.
The AI can just evaluate the move of each stack in complete isolation from one another, and still get decent results.

With 1upt the evaluation of a position for you unit cannot be seen in isolation; the AI has to evaluate its value in conjunction with the position of all the other units in the area.
If the AI evaluate the movement for one unit at the time, it will not be able to make the best use of terrain and cross cover (e.g. artillery behind infantry, and not the opposite).
I really get the feeling that this is what the AI is doing: evaluate the movement of each unit one by one.
Spoiler :
Often the AI moves artillery units far too forward because it doesn't see enemy (human) units in front: this is particularly true when the human uses mounted units... the AI gets a false sense of security because it doesn't see/calculate that fast units may attack from unseen positions.


One way around the problem is that the AI should evaluate all moves, give a score to the final position.
If the resulting score is under a specific threshold, then keep in memory, undo all moves, and try again with some variation (adding no-go tiles for some units).
Repeat until a good end formation is found.

This is pretty much what the human player does (different process but similar target of optimizing end position for all units) and the AI should do the same.
BA brute force approach as indicated may work (but it has a cost in terms of computation time for each turn).
 
Does anyone else like this solution (not sure if its original or not)?

Keep a hex or square tile world map, that's not so important. Reinstitute stacks, but limit units per stack to say 10 max. THEN when any battle takes place, it takes place on a new (let's call it strategy map) consisting of Hexs, or even octagons based upon the terrain of the world map tile.

Each army would deploy, at one unit per hex/oct tile of this strategy map. Then they would move in turn to engage each other. This would allow flanking units, early archery units, artillery units, infantry, tanks, airstrikes, you name it, ALL to have sensible and valid uses. It would also make it in principle much easier to program a decent ai opponent.

-snip-

Thoughts anyone?

I actually really like this idea. It was something I used to daydream about when mentally imagining what a modern version of MoM would be like. (This was before I discovered FFH2).
 
Does anyone else like this solution (not sure if its original or not)?

Keep a hex or square tile world map, that's not so important. Reinstitute stacks, but limit units per stack to say 10 max. THEN when any battle takes place, it takes place on a new (let's call it strategy map) consisting of Hexs, or even octagons based upon the terrain of the world map tile.

Each army would deploy, at one unit per hex/oct tile of this strategy map. Then they would move in turn to engage each other. This would allow flanking units, early archery units, artillery units, infantry, tanks, airstrikes, you name it, ALL to have sensible and valid uses. It would also make it in principle much easier to program a decent ai opponent.

It would stop the (and lets face it, however you dress it up it's ludicrous) fact of a battle where an Archer can fire 130 miles, a horse unit can dart in 240 miles as a flanking manouever etc.

It would also allow much easier army coordination for the ai, and would probably vastly speed the whole game up. (Outside of the battles that is).

For those who didn't like to fight battles as such, there could be some auto-calc button for the battle, which took factors into consideration, and produce a winner.

Let's face it, battles have always been fought between armies. This approach would reinforce that fact, while allowing tactics and strategy in combat, and keeping the 1upt in battle. Trying to use the world map, as the same map as the battle map, and limiting 1upt, is I'm afraid an idea doomed to failure at the concept level.

If you're going to make CIV basically a war game, then at least have a seperate battle map, and make it a decent war game.

Thoughts anyone?
This solution has been proposed several times already and (except for replacing the term "strategic map" with "tactical map", as the tactical level is where the combat takes place, whilst the strategical map is the "normal level". Furthermore, octagons don't work as they require some squares in between) I agree with it.

Actually, I always thought this would be a logical consequence to solve the dilemma of having units (regardless of being being stacked or kept individual ones) being maintained on a empire-wide map.

That is only partly true and even less, later in the game. You forget the AI not only have to count in YOUR actions, but also react on actions of the OTHER AI=players in the game. The more AI's there are, the more you have to WAIT and the LESSER the AI can predict pre-calculate. So i think you underestimate the problem, or overestimate the pre-calculations the AI can make. 8 civ game = 7 AI's and you. That means seven time a AI-civ calculation and then you the player, every turn. And then you have to add the civ-states also.

Count in the fact thay you, the player have not much to do to begin with other then pushing "next turn" for the first 100/200 turns, you leave the AI little pre-calculating time in the early ages for starters.
We are still talking about a TBS.
That means, players are making their turns one after the other.

For the sake of speeding up the game in total, we want to compute the AI's turns during the human turn, which typically takes longer than the total of the AI's turns.

Therefore it is valid to have the AI only check with the human actions (during the human's turn) while in the background computing AI's A to G.
The worst what could happen is to have no improvement over the current state, but this would require the human to act in a way which makes ALL the AI's having to rethink their actions. And this is quite unlikely under normal conditions.
 
This solution has been proposed several times already and (except for replacing the term "strategic map" with "tactical map", as the tactical level is where the combat takes place, whilst the strategical map is the "normal level". Furthermore, octagons don't work as they require some squares in between) I agree with it.
Firstly, thanks for the comments everyone :)

I didn't think it was a completely original idea.If Octagons leave a gap, then hexagons are fine, and of course it is a limited battle map. It would allow really interesting battle terrain, within the world map tile basis.

Someone mentioned Call To Power, and yes I remember that, but I was thinking more of a (much limited) Total War Style battlefield, yet simplified by using hex tiles, and moving in turns, not real time.

Well at least a few folks posted to say they liked the idea in principle.....also by it's very nature of introducing more units to the game, it would naturally favor faster construction times, more buildings, more techs (see where I'm going here;)?)...basically giving the player a lot more to do, and helping to sate the "lacking feeling" that a lot of people have atm, of just hitting next turn far too often...
 
They cant intuitively on algorithm only, but they can beat humans - even GMs if they have good a games/openings library.
That couldn't be true.

The openings library is there to avoid silly move that cause the AI being beaten by a human GM at the beginning of a game (when there are still many tokens of both sites on board). The openings library actually does NOT help in beating human grand master.
 
That couldn't be true.

The openings library is there to avoid silly move that cause the AI being beaten by a human GM at the beginning of a game (when there are still many tokens of both sites on board). The openings library actually does NOT help in beating human grand master.

Your message is kind of contradicting itself. You make the point that the opening library stops the AI losing but doesn't help it win. That is a little confusing.
 
Opening libraries help the AI's but, as some GM's had learned from their own experience, making suboptimal moves to force the AI's out of book, usually ends up badly.
 
Chess players have spent enormous amounts of time and energy analyzing common openings. They play those openings "better" than if they had to make all of their decisions and discoveries about them at the board, because they have access to all of that information acquired before they ever sit down at the table. Chess programs have opening libraries for exactly the same reason, to put them on an equal footing.
 
We are still talking about a TBS.
That means, players are making their turns one after the other.

No.
There are TBS where players give their orders simultaneously. For instance, consider Diplomacy. Orders are given, and then all orders are resolved.

Someone mentioned Call To Power, and yes I remember that, but I was thinking more of a (much limited) Total War Style battlefield, yet simplified by using hex tiles, and moving in turns, not real time.
The proposed system is actually more like MoM or MoO. In both these games, the ai was a joke because it couldn't handle both the tactical level correctly. Civ is easier tactically (no black hole generators and webs/cracks call spells) but the ai would still be given another stick to be beaten with in such a system.
 
Your message is kind of contradicting itself. You make the point that the opening library stops the AI losing but doesn't help it win. That is a little confusing.

opening library only helps the AI to avoid silly mistakes at early stage (or as we human being understood as bad openning), but it has nothing to do with those calculations which actually defeats a GM at later stages.

i.e. If you have a really good chess AI engine, you can still beat a human GM without an openning library.
 
opening library only helps the AI to avoid silly mistakes at early stage (or as we human being understood as bad openning)

This is just wrong. Opening analysis by grandmasters looks many moves ahead and identifies moves that are slightly better for subtle reasons too difficult to work out at the table. Opening books are not just about "avoiding silly mistakes" (computers are already very, very good at avoiding any silly mistakes, even without any opening book). They are about obtaining a slightly better position with slightly better winning chances in the middle game, by finding the very best moves.
 
And...? Don't you know that Farmville is the most popular game on Facebook? Not that I play it, but actual success trumps whatever you might think of a game.

Poker machines are even more successful than that. Just because its successful doesn't mean its not a scourge on society.
 
This is just wrong. Opening analysis by grandmasters looks many moves ahead and identifies moves that are slightly better for subtle reasons too difficult to work out at the table. Opening books are not just about "avoiding silly mistakes" (computers are already very, very good at avoiding any silly mistakes, even without any opening book). They are about obtaining a slightly better position with slightly better winning chances in the middle game, by finding the very best moves.

In my previous post for what I have quoted in the braces, see it?

In general a bad openning is silly, so I am not wrong when I say it is to avoid "silly mistake", my intention is to make it clear to someone who has very little knowledge about writing a chess AI engine and who has tried to over-emphasize the role of openning library. (as though it is the most important component that enables a chess AI engine to beat a human grandmaster... in fact it has very limited function in that case)

Openning library has nothing to do with chess AI engine in term of calculating the next move to take. It is just a library for seek and match among those recorded openning states to determine next move.

Your statement:
"They are about obtaining a slightly better position with slightly better winning chances in the middle game"
tells me you really know nothing about how a chess AI is being made.
(come on, why would you want to name it openning library if it is used in the middle game? :D:D:D)
Why don't you just make a seek online, I think there is an open-source codes of basic chess engine... :p
 
In general a bad openning is silly

No, this is just not true. Maybe you are not a native English speaker, so you are misusing the words. "Silly" means "exhibiting or indicative of a lack of common sense or sound judgment". However, it is entirely possible for even a grandmaster with excellent common sense and sound judgment to make a slightly inferior move in an opening they have not studied in detail. Such slightly inferior moves are not "silly" at all. But the benefit you get by avoiding them, is important in high level play.
 
No, this is just not true. Maybe you are not a native English speaker, so you are misusing the words. "Silly" means "exhibiting or indicative of a lack of common sense or sound judgment". However, it is entirely possible for even a grandmaster with excellent common sense and sound judgment to make a slightly inferior move in an opening they have not studied in detail. Such slightly inferior moves are not "silly" at all. But the benefit you get by avoiding them, is important in high level play.

Come to think about it, I think I can agree that using the adjective "silly" to describe bad openning chess game is inappropriate in general. But it is definitely the right word to use in the case of a chess GM.

Let me tell you what, even before the time we have computer and thus this thing call openning library for a chess computer program... all professional chess players memorize a lot of openning moves as part of their training. Basically, I think it is appropriate to say, the basic difference between an amateur chess player and a professional one is, the later is required to memorize opening moves (lot of them) and the former only do it at will. So, it is entirely impossible for grandmaster to play a bad openning chess game by mistake, if he did, he is disqualified as GM.
 
Top Bottom