Block war by block city?

[Optional]

This is a question about AI behaviour. In a game I'm in as the Maya I found myself about to be attacked by the Mongols. I was at peace with them, but that would soon change:

Spoiler :

Those swords were heading for me. I didn't mind that, in fact I welcomed a war, except my defense was still a bit thin at 800BC; not many catapults yet.
What I ended up doing was making a guess that the Mongols wouldn't declare war before they were positioned right outside one of my cities. So I used my units to block direct access to the town they were heading for - Tulum, hoping this would buy me some time to get a bit more arty in place:

Spoiler :

It worked; the Mongols tried to circumnavigate my units instead of attacking them, and only declared upon the opportunity to attack a town. Then I already had more catapults in place and it wasn't too difficult for me to deal with those swords.

But I'm wondering: is this blocking tactic fairly safe to use or did I just get lucky? I only used units with defense 2 to block them and thought the Mongols would perhaps attack warriors/horses, but I really don't know. Have other players any experience with this tactic?

 

[vmxa]

As long as the attackers are on par with the blockers, it should work. In fact they tend to go around, even if they had already declared.

This is what happen on my turn set in Rat40. I had 2 spears dead in front of the town and they sent their units around, even though they had already started the war. They only attacked, once they had direct access to the town.

The attackers where the same and the defenders where the same as I moved them into towns after they passed.

If they had horse units or 4 attack units, I am pretty sure they would not have gone around.

 

[undertoad]

I don't have experience of games where the enemy units are far superior to my blocking units, since I play up to Monarch level and always concentrate on preventing the enemy from getting ahead of me on tech (if I can, I try to make sure I'm well ahead).

But that said, yep, enemy units always prefer going round to attacking, even when actually at war. There's another thread called Decoy Units that discusses how obsessed the AI is with this.

Blocking can also be very useful to block those infuriating Settler/Spearman pairs the enemy loves sending into your territory when you have a ROP, or even to "herd" them to the border. Leave them only one direction in which to move - which is towards the big sign saying You Are Now Leaving [your civ] - Thank you for Walking Carefully.

It may be that when the AI does have far better units, blocking doesn't work and they just attack - but that's outside my experience.

 

[Fiddlin Nero]

Blocking is definitly a major tactic for me. In peacetime Workers would have worked in your situation. The ultimate blocker though has to be the army, in 90% of the situations the AI will not attack an army. I love blocking allies with Workers to keep them from getting to enemy towns I don't won't to share.

 

[Lord Emsworth]

But I'm wondering: is this blocking tactic fairly safe to use or did I just get lucky? I only used units with defense 2 to block them and thought the Mongols would perhaps attack warriors/horses, but I really don't know. Have other players any experience with this tactic?

I cannot prove it, but it has long been a suspicion of mine that the AI will only ever lauch sneak attacks against cities, and never against units in the open. You observation confirms that suspicion. If true, it might be possible to avoid war with 100% certainty if only your empire layout permits it.

The only ways the AI declares war are:

1. As a reaction to rejeted demands.
Countermeasure: If you give in to every demand, the AI has no chance to declare war.

2. As a reaction to "Leave or Declare" orders.
Countermeasure: Never ask the AI to "Leave or declare", or better yet, make sure that the AI does not even enter your territory.

3. As a reaction to espionage.
Countermeasure: Don't use espionage.

4a. An MPP partner of an AI is attacked
4b. An AI is bought into a war via a Military Alliance
Countermeasure: Well, both have as a prerequisite that you yourself are at war already. So, just don't start any wars and no MPPs can kick in, and neither can anybody be brought into an MA against you.

5. A sneak attack against one of your cities.
Countermeasure: Block all access to your cities. (A One City Challenge with a landlocked city looks ideal.)

If, on top of that, you stay away from MPPs yourself

==>

Chances of War = 0%

 

[Hellfiredoom]

I spit out lots of warriors early in the game as scouts, to increase my Civs "power" (to reduce AI requests and DOWs), and as blockers in support of my REX efforts. I'll fortify a warrior scout on any chokepoint he comes across and leave him there until the AI boots him out of his territory. And if I have the units to spare (ha, who ever does?) I'll even herd AI settler/spearman pairs off to the nearest coast and pin them there until I get my territory laid out around them.

 

[Optional]

I've seen more references in the forums about the AI going around, but I've seen the opposite also happen a lot in my games; plenty of AI fighting in the open going on, and not just the easy battles, also longbows attacking pikes and such.

So blocking units when at war is something I wouldn't do unless my blocking units were of superior strength. When they're iffy I wouldn't chance it.
In the last screenshot I'm blocking while still at peace, but I've done a test and I now have uploaded the same save again and declared war on the Mongols myself.
Since in this discussion some people didn't seem convinced of the difference between before and after a war declaration, I can tell you what would have happened with my blocking units had the war already been declared:

Spoiler :

They would have been wiped off the face of the earth!

The Mongols had 5 attack opportunities, and took them all 5; 5 of my units dispatched.
I will admit though; there is likely a significant RNG factor involved in these decisions to attack.

But I'll do a bit more testing with these saves. I'm also curious to know whether I would have gotten away with blocking with ordinary warriors, or even a catapult or so - err, while still at peace, of course!

It's just that testing is a bit like... work , so that's why I asked in this forum first, but if the opinions are so diverse, then I'm curious enough to see if I can get a bit clarity on the matter by performing some tests.

 

[Spoonwood]

5. A sneak attack against one of your cities.
Countermeasure: Block all access to your cities. (A One City Challenge with a landlocked city looks ideal.)

I definitely recall in at least one OCC game (well it didn't start out as an OCC, I abandoned all my cities near the end, but this shouldn't matter), with my city on the coast, where I've had my city surrounded with units and still gotten attacked. Pretty sure I've had this happen in true OCCs also.

That said, the probabilty of getting attacked may decrease, and/or may decrease significantly, via blocking.

 

[Fiddlin Nero]

Its been awhile but I just had to save it. An AI and I were at war with this last city of another AI. Those are my Mech Infantry blocking access to the city. There must 30 to 40 Battleships of the AI stacked up around that city trying to destroy it with bombardment. Oh, what an opportunity to blast some Battleships just for the fun of it!

 

[Spoonwood]

Now that I recall more, I know I used magniot lines and got attacked when I first started playing also.

The AI CAN launch sneak attacks against units in the open. The screenshots below suggest this, and the saves confirm that I haven't just put pictures together. Method 5 proposed by LordEmsworth does not always work. It may in some cases work, it may decrease the probability of an attack, but it does not always work. At least in Conquests. Therefore, Optional's blocking tactic may decrease the probability of an attack in general, but as long as this example fits into his scheme, it won't always do so.

Spoiler :

http://i370.photobucket.com/albums/oo146/Spoonwood792000/1274031055.jpg

Spoiler :

http://i370.photobucket.com/albums/oo146/Spoonwood792000/1274031056.jpg

 

[Spoonwood]

VMXA's extra condition that the blockers have at least much strength as the attackers, doesn't quite work out either (given that I've interpreted what he wrote correctly). I reloaded from the auto-save a few turns back in that game and built rifles. Only a rifle gets attacked between 1560 and 1565. The rifle gets attacked by a cavalry. The Greeks though did attack a turn later this time. Check these two saves, and screenshots:

Spoiler :

http://i370.photobucket.com/albums/oo146/Spoonwood792000/1274034294.jpg

Spoiler :

http://i370.photobucket.com/albums/oo146/Spoonwood792000/1274034296.jpg

 

[DWetzel]

I believe that the key is that the units will always (almost, anyway) go around the units to get to a city rather than attack the units. But this requires that there be some way to get to the city in question.

I believe that if you set things up with seven units around the city, and one opening for units to move toward it, you could get the units to dance around trying to get to the opening almost indefinitely.

I'm sure this would work if an enemy is sneak-attacking; I'm pretty sure that if already at war it wouldn't work.

 

[Spoonwood]

I expected I would quickly disconfirm DWetzel's hypothesis. I did not. The Greeks danced around for a few turns and then basically behaved as if they felt disinterested in attacking (before 1615). Does this suggest that if you leave one square open, and they can't attack your city directly (and you don't have workers unprotected, if that ends up necessary also), that the AI cannot attack any unit of yours? I do not know how to confirm this. I suspect one might possibly write a computer program to check that the AI does not attack here. If the "preserve random setting" changes the probability of a sneak attack, this wouldn't come as too hard to check for many cases. Does such a dance come as sufficiently similar to Bamspeedy's Army Wall Tactic/Sirpleb's FoD to make a reasonable comparison between the two such that since Bamspeedy's Army Wall Tactic works, what DWetzel's suggests should also? Some saves:

 

[DWetzel]

I'm certain that units aren't enough of a priority to make the AI sneak-attack, while cities are. I'm not certain about workers but I don't recall ever having those be the first strike of a sneak attack either.

Obviously, once actually at war, all bets are off.

Edit to add: the Funnel of Doom works differently. That just takes advantage of the AI not attacking units of sufficient defensive value (usually armies, but any sufficiently good defensive units should work).

 

[Lanzelot]

I also recall from past games that if the AI is "planning" a sneak attack, you can prevent/postpone that sneak attack via blocking access to cities.

However, even if the AI already managed to move next to your city, there are ways to make a sneak attack more unlikely:

• Obviously, an undefended city is an outright "invitation" for a sneak attack. This is the reason why you should protect all of your cities with at least one unit, if you have a RoP with an AI and railways in place. (Not doing so is asking for trouble... I paid dearly for neglecting this rule in COTM48...) I also noticed in past games, that if an AI is about to sneak attack you, there's nothing to stop them: in one case, when I noticed an AI was about to sneak-attack, I signed a RoP, gifted them stuff (including a luxury) and bought a tech for lots of gpt. --- To no avail: the next turn they attacked me anyway, blowing their rep and loosing the gpt/luxuries.
• If they have the units next to a lightly defended city, they may sneak-attack anyway. However, I have never seen them doing so via RoP/railway?!
• However, if the city is better defended than their attacking force, they won't try the sneak-attack. (Or at least the probability seems to be very low.) You can probably test this by reloading from a sav where they sneak-attacked you and then putting more defenders into the city.

Regarding SirPleb's Funnel of Doom: I think it combines both strategems: the fact that the AI won't attack if chances of winning the battle are like 1%, and the fact that the AI is obsessed with trying to get to undefended cities. I haven't read SirPleb's article in a while, but I recall it involves alternatingly opening and closing access to an undefended city in order to make the AI units march around in circles...

Lanzelot

 

[Optional]

Thanks for the replies! I still want to do some testing myself, but... lazy.

I also noticed in past games, that if an AI is about to sneak attack you, there's nothing to stop them: in one case, when I noticed an AI was about to sneak-attack, I signed a RoP, gifted them stuff (including a luxury) and bought a tech for lots of gpt. --- To no avail: the next turn they attacked me anyway, blowing their rep and loosing the gpt/luxuries.

I'm fairly certain in my game I could have stopped the Mongols from attacking me; by declaring war on the Aztecs and signing an alliance with the Mongols. Aztecs were situated between me and the Mongols, I had 2 techs up on the Mongols, so I could have bribed them, even with an RoP between Aztecs and Mongols.
But then I would not get any war happiness and the risk is that either Mongols or Aztecs would get the upper hand in the war. And probably less chances for my Javelin Throwers to make slaves.

 

[Lanzelot]

I can confirm that the pure action of signing a Millitary Alliance does not stop a sneak attack, if they have the units already close by: I remember in that game I mentioned above I got interested in whether there would be any way to prevent them from declaring. So I tested the autosave several times with different things, and signing an MA against a common enemy was among them. (As well as signing trade embargoes, gifting gpt, resource, luxes. I think I did not try re-negotiating the peace deal, that might be something worth testing.)

I was very weak compared to them (a Demigod game in the early stages), perhaps that has an influence.

In your case it might have worked, if their units were still "on the other side" of the Aztec territory (because then they would rather attack the Aztecs than marching through their territory in order to get to you), but if their units were already at your borders, I doubt that signing an MA would stop them from sneak-attacking undefended/ pooly defended cities...

 

[Optional]

In your case it might have worked, if their units were still "on the other side" of the Aztec territory (because then they would rather attack the Aztecs than marching through their territory in order to get to you)...

Yeah, the only time I tried this something like this was the case; another civ had some weak unit(s) next to the stack that was approaching me. What the territory situation was I don't remember, but the stack immediately attacked the easy target(s) next to them, as you would expect from the AI.
If I'm picking up my save again I'll try testing this bit as well. The Aztecs didn't have units nearby the Mongol stack, so if your guess is they would have marched on in this case, based upon your observations, then that makes my doubt my earlier assumptions - although it could still be a roll-of-the dice thing for a big part.