<< Back to General Forum   Search

Posts 1 - 18 of 18   
Crazy Challenge: 11/4/2015 16:05:53


wêr
Level 28
Report
Greetings Forum,

I am busy doing the crazy challenge,
I am concerned i will mess up this situation,
could you give me some suggestions?

I placed numbers to see how long it would take to get there.



Edited 11/6/2015 13:07:52
Crazy Challenge: 11/4/2015 18:07:21


Cata Cauda
Level 58
Report
Image doest load for me...
Crazy Challenge: 11/4/2015 18:12:08


Riveath
Level 59
Report
That's definitely not bad. Now ensure the two teams are fighting each other, try to keep a balance between them while gaining power yourself.
Crazy Challenge: 11/4/2015 18:26:10


wêr
Level 28
Report
Maybe it loads now, I turned it into a jpeg.

Edited 11/6/2015 13:07:05
Crazy Challenge: 11/4/2015 18:47:47

wct
Level 56
Report
Both images are broken for me. Says "Could not load image" in error log.
Crazy Challenge: 11/4/2015 18:58:25


wêr
Level 28
Report
I changed the host, maybe now it works.
Crazy Challenge: 11/4/2015 19:55:46

wct
Level 56
Report
ok, can see them now.

First thoughts: You are in a good position at this turn, so you *definitely* have a chance to win it before turn 22. *But*, you will need to play in a way that outsmarts the AI's algorithm, and from your current borders, it appears that you might not yet be playing the best way. If you want to see how to beat the AI, take a look at some of the games on this thread: https://www.warlight.net/Forum/114929-something-different

Specifically, you are bordering the AI in Brazil with 6. That means that next turn, *if* green can afford to attack you, then it *will* attack you with *at least* double your 6, which is 12.

Therefore, if you are to guarantee you are going to take Brazil, you need to deploy enough troops in Nigeria to defend from at least 12, and also to *counterattack* whatever remains.

To do this properly, you need to use delay moves (or just be so massive already that your expansion moves act like delay moves also), so that green will attack you *first*, smash against your giant stack in Nigeria and destroy itself.

However, this is all complicated by the fact that you have a 6 on your border. You need to keep your borders much lower if you are going to be able to outwit the AIs consistently. To put this in perspective, if you only had 3 on your border, then green would try to attack with 6 instead of 12. So, to defend and counter-attack against 6 is much easier than to counter against 12. Likewise, if your border was only 1, then you'd only have to counter against an attack of 2.

This is the *general* rule, however there are exceptions. If green's *only* enemy he can see is you, and he can easily already afford all of his expansion moves, then he will attack you with *everything* he's got left over, so one of his attacks will be 'all of his remaining troops', and the attack might be *bigger* than twice your previous defenders. E.g. If he can afford to attack with 13 or 14 or more, then he *will*.

If you simply pay attention to the *exact* sizes of the AI's attacks, compared with the *exact* number of defenders you had in that territory at the end of the previous turn, you will see it's almost always exactly double. There are a few exceptions, and those are the quirks of the AI.

As far as being able to reach all of the territories, don't worry about that yet. Just try to keep expanding at least one territory per turn and you should be fine. The tricky part will be to balance the enemy AI teams so that neither one gets a big income. Whoever has the biggest income, you should try to break their bonuses as soon as possible, and keep them broken. If you do it right, you should be able to get all the AIs knocked down to only 5 income near the end of the game. The faster you eliminate some, the easier the end game will be.

So, summing it up: Predict exactly how much the AI will attack you with and where, and deploy *just enough* to defend and counter-attack, ideally so that at the end of your attacks, all your territories [edited to add: by this I mean all your territories on your *borders* with AIs] look as weak and innocuous as possible (ideally with just a single 1 army in them if possible). Sometimes you'll end up with stacks, which is unavoidable, but just use them to defend nearby territories instead of trying to build them up even bigger. Check out those games I mentioned to see how I do it.

Edited 11/6/2015 22:49:00
Crazy Challenge: 11/4/2015 19:59:01

wct
Level 56
Report
To beat the Crazy Challenge, it's enough to play it a few times, until you are sure you missed the 22 turn mark, and then check the history to see where it is important to focus on with highest priority.

To beat Insane, you'll need to use many of the techniques I mentioned previously.
Crazy Challenge: 11/4/2015 22:36:22


wêr
Level 28
Report
Those three troops are leftovers from an attack of 4 from Venezuela to Brazil,
so, likely, RED (not green) has more areas to conquer in South America.
Can I assume the 3 troops wil be used to conquer more areas, or will they remain to defend?
RED has maximum 9 troops reinforcements, and I can afford to attack with 25.

I minimized the session for later, and i started a new one and paused it at turn 6 :



Edited 11/6/2015 13:25:44
Crazy Challenge: 11/5/2015 07:29:52

wct
Level 56
Report
Whoops, sorry for the colour confusion.

Those three troops are leftovers from an attack of 4 from Venezuela to Brazil,
so, likely, RED (not green) has more areas to conquer in South America.
Can I assume the 3 troops wil be used to conquer more areas, or will they remain to defend?

Oh, that's good, then. It's very unlikely he'll attack you then, because the AI will pretty much always prefer to attack neutrals if it can afford to before it attacks an enemy.

You can expect that it will deploy a shitload in Brazil, but will spend all of them trying to take neutrals in South America. So, if you have enough delay moves, you should be able to wait until all of its attacks are done, and then simply take Brazil with just 2 armies. Whatever you do, don't attack early in the turn.

Another thing to know is that the AI *never* ever leaves troops to defend its territories *if* it can afford to attack from that territory. The only time it leaves defenders is if it can't afford to attack. And the only other circumstance you might find troops 'left behind' is if they actually were used to attack, but the attack failed and these are the survivors bumped back to their original territory.

Thus, if you can be sure that a) the AI can afford to attack, b) none of those attacks will be of the type 'failed but with surviving attackers, rebounded', and c) you can out-delay the AI; then you can also be sure that by the time your delayed attack rolls around, there will be only 1 army left over in that territory, and therefore a single attack with 2 armies will take it. This rule is how I beat AIs so consistently. You only use the bare minimum to steal their territories, and as a bonus it leaves most of your borders looking completely weak and innocuous with only 1s everywhere.

RED has maximum 9 troops reinforcements, and I can afford to attack with 25.

Do you mean max 9 deploy? Or do you mean including a Reinforcement card?

BTW, you should remember that in Crazy (and probably in Insane as well), the AI teams get Reinforcement cards. They take turns playing them, so it might be the case that Red will be getting a Reinforcement card this turn, so that 9 might be a 15 or something like that. Keeping track of which AIs have played reinforce cards and in what order will help you beat that level faster, since you'll know which AI will be likely to "Surprise! I'm not dead yet!" which is really annoying if you're not expecting it.

For that Africa/South America battle, you might want to deploy 12 (total 18) just in case, but make sure you use enough delay moves, and then one of your last moves you should attack with *only* 2. When you see this work, it's like magic.

Then, the next turn, on your very first move, push all the surviving armies from Africa into Brazil to 'reinforce' it. If you play it right, then the AI will send a bunch of 2 attacks from all of its S.America territories to try to retake Brazil. If you *also* deploy some armies in Brazil next turn (you should), then delay, then you should be able to take all of those territories with just 2 armies each.

But for this turn, use your remaining deployments to expand expand expand as fast as possible. Go for the juiciest income first. Also try to fuck with any AIs you see nearby using the same old one-trick pony tactic as I've been describing. Try to break AI bonuses and keep them broken, while you systematically scoop up your own bonuses one by one. Try to expand one territory towards the furthest map point each turn. It's not as hard as it might sound. The hardest part of this challenge is eliminating the AIs, not just reaching the farthest map point.

I minimized the session for later, and i started a new one and paused it at turn 6 :

Good job, this one looks even better. You know Antarctica is safe, so block Colombia and Venezuela from possible capture. They will probably be attacked with 4 each (if there's an AI there) which would leave 3 defenders in each if you attack late. Therefore, attack them with 5 each to be sure you get them. Don't spend armies developing S. America until you're sure it's safe (unless you can totally afford it and have nothing better to spend the troops on).

Take only the South Pole in Ant. Again, don't waste troops developing if it's not safe yet.

Red is likely to attack Iraq with 4, and it's possible an AI will attack Spain with 4. So, both of those you should attack with 5 late. Do not take Saudi Arabia this turn, since you don't need to, and it doesn't gain you any new neighbours in the fog.

One other thing to know: The AI always does:
1) All attacks against neutrals first
2) All attacks against non-neutral enemies next
3) All simple transfers of left-overs last, moving toward closest enemy or neutral.

Therefore: You should move extra defenders into vulnerable territories as your first moves. For example, move the 3 from Egypt to Israel on your first move. That way, if the AI ends up attacking you there, those 3 will be the first ones to die, leaving your attackers available to carry out their attacks.

With all of that detail being said, here's what I would do *ideally* (if I had more income than you do):
[deploys]
+4 Novolazarevskaya
+9 Brazil
+4 Algeria
+ the remainder in Israel (at least 6)
[urgent reinforce moves]
Transfer 3 from Egypt to Israel
[delay moves]
Trx 1 from Mauritania to Algeria
Trx 2 from South Africa to Novolazarevskaya
Trx 1 from Namibia to South Africa
[relatively safe attacks]
Attack 5 from Novolazarevskaya to South Pole
Attack 5 from Brazil to Colombia
Attack 5 from Brazil to Venezuela
Attack 5 from Algeria to Spain
[most critical attacks last]
Attack 5 from Israel to Iraq
Attack 2 from Israel to Turkey
[just to be sure your front-lines are minimal, in case Turkey not taken]
Attack remainder from Israel to Saudi Arabia

Unfortunately, you're a few armies short on the income. So, you'll have to do a bit of a trick to use fewer armies taking those neutrals. Instead of attacking with 5s *late*, you'll have to attack with 4s *early*. This is a wee bit riskier, but in fact it's what I do more commonly. So, revised suggestion, fit into your budget:

[deploys]
+2 Novolazarevskaya
+7 Brazil
+3 Algeria
+7 Israel
[urgent neutral captures]
Attack 4 from Algeria to Spain (to block Red from possible Europe capture; keep their bonuses broken)
Attack 4 from Brazil to Colombia
Attack 4 from Brazil to Venezuela
[urgent reinforce moves]
Transfer 3 from Egypt to Israel
[delay moves]
Trx 1 from Mauritania to Algeria
Trx 2 from South Africa to Novolazarevskaya
Trx 1 from Namibia to South Africa
[relatively safe attacks]
Attack 3 from Novolazarevskaya to South Pole
[most critical attacks last]
Attack 5 from Israel to Iraq
Attack 2 from Israel to Turkey
[just to be sure your front-lines are minimal, in case Turkey not taken]
Transfer remainder from Israel back into Egypt

Edited 11/5/2015 07:30:14
Crazy Challenge: 11/6/2015 09:00:55


wêr
Level 28
Report
Thank you very much for your awesome contributions! I Highly appreciate it and am learning much! I am processing all your information and have been practicing it. First session is in the state below.

It's unclear whether Canada, Australia, East Russia, East China and Indonesia are combined to a bonus, during turn 13, within my view, Light Blue deployed 18 reinforcements, Grey did 3, Red did 3, and Purple did 2. I assume Light Blue had Australia, East China and Indonesia, and played a card in turn 13. Which makes Light Blue's income 16 reinforcements in turn 13, and was deploying 4 troops somewhere else also. My guess is, that Canada and East Russia were contested, and me and Light Blue were the only ones with bonuses. What I think: Team B had 15 income and Team C had 26, unfortunately my attack on East China was nerved by Light Blue. Noteworthy: Light Blue decided to attack my 1 troop in Pakistan from India with 12 troops, and Tibet with 1.

I have to move towards Queensland, Northern Territory, South Australia, Papua New Guinea, Alaska, Yukon, and Anadyr within this turn. Thus far my attacks should be Hong Kong (Papua New Guinea/Northern Territory), Baja (Yukon), South Pole (Queensland) and Irkutsk (Anadyr).

East US, Scandinavian Peninsula and West russia are bonuses i can get within 2 turns, Gulf Coast, Tennessee, Norway and Arkhangelsk would be secondary attacks? And Southwest, Itseqqortoomiit, Tura, Mongolia/Inner Mongolia/Shaanxi would be tertiary attacks?



Edited 11/8/2015 10:04:17
Crazy Challenge: 11/7/2015 00:42:55

wct
Level 56
Report
I won't post too much about your specific game, since it takes a while to think through, and a lot depends on what you've seen the AIs doing previously (including how they attack each other and where they prefer to deploy). Plus, it's *your* crazy challenge, after all. But I'll give some tips regarding how the AI works that will help not just for this game but anytime you meet them.

One thing I notice right off the bat is that Red looks like it's on its last legs. If that 4 is its only remaining territory, then you know for certain it will deploy all 5 there. And if it happens to be playing a reinforce card this turn, then it might be 11 or whatever.

Remember that the hardest part of this challenge is eliminating the AIs quickly enough. It's common to have access to the entire map, with huge income, but with pesky little 5 income AIs constantly dodging your attempts to squish them, and you end up missing the 22 turn mark by one or two turns.

So, if possible, one of your goals which you'll need to decide priority on is to a) defend against red, in case it decides to deploy all in its only territory and attack you (or, almost as annoying, attack a nearby enemy territory you were hoping to steal but can't now that red's junk is all in there); and if possible b) force red into a position where it runs into too many walls and gets eliminated.

One subtle bonus of eliminating a teeny-weeny AI is that it stops earning card pieces for its team, so the *other* AIs on its team will get fewer reinforces to play, making them easier to eliminate too. The earlier you eliminate your first AI, the better, generally.

Light blue is definitely your main target. Aside from eliminating red, put less priority on fucking with red's team. That way they are more likely to fuck with light-blue's team. But you have to balance that with picking up poorly defended bonuses; it's not a hard rule.

One difficulty with the strategy/tactic I describe is that it usually requires a lot of deployment orders on all your borders, so if you have lots of borders, it can strain your income to try to deploy to them all. This can make the going much slower and more tedious. Therefore, look for opportunities to take certain sets of territories such that by taking these key territories it *reduces* your overall number of border territories.

For example, in this turn, in Caucasus, you have to defend both Georgia and Western Kazakhstan. But if you captured Ufa, it would protect both of them, so you'd only have to defend one territory instead of two in the future.

In a similar vein, even if you can't *reduce* the number of borders you have, often it's good enough just to keep them the *same* number.

For example, if you chose to attack Norway, and only Norway, then you would now have defend Norway, Sweden, and Belarus, instead of just Sweden and Belarus. However, if you *also* captured Finland on the same turn, then the combination of Norway and Finland would protect both Sweden and Belarus, so you still only have to defend 2 territories there in the future.

[Generally speaking, reaching the map borders is a really good way of reducing the number of territories you have to defend. On those Big USA games in the other thread, you may notice that as soon as I reached the northern and southern borders, the game play accelerated a lot, because I could keep just pushing east and west without having to increase my number of borders at all. On your map, the three Russias have the same kind of effect. There's no need to defend the northern border. Just push east and west from there. So, reaching the northern border should be one of your goals.]

On a slightly related note, you'll find that the AI tends to prefer certain kinds of territories. I don't have a strong rule for this, but to me it seems to prefer different kinds of territories: 1) Territories with the most connections/neighbours, and 2) Territories that are hard to get to, and therefore easy to defend.

For example, remember I suggested to take Ufa? Well, Ufa also happens to be one of those territories with lots of connections (8). So if the AI is feeling tenacious, you might find that you'll have to defend Ufa harder than the average territory because it will keep getting attacked. The South Pole is another really good example of this. The AI will fight to take/keep the S. Pole much harder than other territories.

On the other hand, one of the territories which the AIs seems to *really* grab hold of with enormous tenacity is Tiksi in Central Russia, even though it only has 2 connections. It's just that it's hard to get to, and thus easy to defend. If you manage to kick the AI out of there, the bonus will be yours. *But* you'll probably find that every time you try to take that territory, especially if it's a weak 5 income AI, it will deploy all of its troops there, and sometimes won't attack if it's surrounded by big stacks, so you might have to send a whole stack in there just to pound it to death, instead of the usual 2 vs 1 trick.

It's these kinds of preferences and 'stickiness' that make finishing this challenge in 22 turns difficult. Just when you think you got em, they dig themselves into their trenches, or take a territory with a lot of connections (and therefore a lot of escape routes for the AI and borders for you to defend against).

Another general note, especially when dealing with AI teams, is that you should try to drive a wedge between allied AIs' borders. For example, pushing up into Scandinavia will drive a wedge between blue and grey, effectively increasing the number of borders that *they* have to defend and thin out their income over.

Almost the flip side of this idea is to try to 'corral' enemy AIs together in a tight/compact space, so that they mutually keep their bonuses broken instead of you having to do all the work. Another way of putting this is *don't* drive a wedge between *enemy* AIs. Instead, force them closer together, so they fight each other more instead of you.

For example, grey and purple are enemies, so (generally) *don't* split them apart. E.g. in case you decide you want to take Irkutsk in East Russia, then probably *don't* take Tomsk and/or Tura at the same time. Leave some area of contact between them, so they waste their income fighting over it.

Also, light blue and purple are enemies, and light-blue is especially your #1 target, so try not to split l.blue and purple up. E.g. taking HongKong is probably your #1 priority this turn, since it will break that bonus, making the rest of the game easier. However, this risks driving a wedge between l.blue and purple. At the same time, l.blue is big, and it probably has lots of contact with purple. But it also has lots of contact with you. So if you can keep its contact with purple, but reduce its contact with you, then it will be more likely to waste income on purple instead of you.

Therefore, if you're going to take HK, which you definitely should, then make it a priority to also clear out SE Asia before pushing deeper into China. Luckily, this also fits into the idea of 'decreasing, or at least not increasing your # of borders'. E.g. if you were to take India, then you have not increased your # of borders, but you *will* have decreased the number of borders l.blue has *with you*. Likewise with Myanmar and (if you also have HK) Thailand. As a bonus, you get ... the bonus (no pun intended).

When you look at the map that way, you can see this as 'corralling' l.blue and purple together, as opposed to splitting them apart.

On a somewhat tangential note, it's the nature of these kinds of maps (and networks/graphs see https://en.wikipedia.org/wiki/Graph_theory for what I mean) that you don't necessarily have to advance toward the furthest node *on every path* to reduce the total distance to the furthest node. For example, taking Gulf Coast in E.USA would reduce distance to Alaska by 1, but so would taking Southwest or Baja. Therefore, if fighting yellow is lower priority (and it is) then taking Gulf Coast (if all you are considering is reducing distance to Alaska) is therefore lower priority than taking Baja or Southwest. On the other hand, if you have lots of troops to spare, certainly go for it.

A further example. You mention taking S.Pole to get closer to the Australian territories, but technically, taking Mawson would also get you one closer.

... cont'd
Crazy Challenge: 11/7/2015 00:47:13


Potatoe
Level 57
Report
Wct, I really appreciate what your doing for this individual :) Thank you for helping
Crazy Challenge: 11/7/2015 01:22:57

wct
Level 56
Report
Okay, one last neat trick and then I'll sum up.

*If* you have enough income, and you might, then there's a neat trick to use the AI's quirky preferences against it and to take the entire of Antarctica in just 2 turns from your current position. However, to *guarantee* it, it will require a lot of income expenditure, so you'll have to prioritize it with all your other goals. Simply waiting a turn or two to get another bonus or two might make it much easier. It's up to you to decide.

The 'trick' is really just the same one-trick-pony method of beating the AI, but applied in this specific case where you *know* where the AI prefers to attack.

Unfortunately, you have 5 in Novolazarevskaya, that will make it harder (require more income).

Turn 1:
1) Build a big wall on Novolazarevskaya to withstand all expected attacks, with enough troops left for counter attacks. You should expect attacks of size 10 (2 times 5), and there are max 3 possible attacks (but limited by AI's income) so max of 30 armies coming at you. You want all to die, so you need minimum 14 defenders in last attack. Plus 6 die from each attack, so need 14 + 6 + 6 = 26 in order for all attackers to die.
2) Out-delay the AI with other moves.
3) Make 3 late attacks of 2 against S.Pole, Mawson, and Halley.

Turn 2:
1) Build another wall on SPole, but this time you only need to defend against attacks of 2. There will be 3 such attacks. However, since it's the SPole it's possible the AI will make a big attack, too. So, it's possible 2 attacks of 2 plus a big attack (if you take HK previous turn too, then big attack less dangerous). For the sake of fool-proofness, let's assume the AI attacks SPole in one big attack with its entire income. That's an attack of 16. To kill 16, you need 16 / 0.7 = 23 defenders. So deploy 23 to SPole. For safetey, deploy at least 1 to Mawson and Halley too. That's total of 25, which you could theoretically afford now. (Remember, this is overkill for fool-proofness; actually much less will be required almost certainly).
2) Out-delay the AI.
3) Make three late attacks of 2 against Scott, Siple, and Russkaya.

At the end, you'll also have a big stack in S Pole hidden from the AI, which you can use to reinforce Scott with an early transfer, to ensure that you break Australia next turn.

Now, realistically, you probably don't really need 26 the first turn and 25 the second. And you probably can't afford that anyway. But you just start from the ideal case (or worst case, depending on your perspective) and make certain compromises to work it down to your budget.

For example, if, instead of a big attack on turn 2, you expect only 3 small 2 attacks, then you only need to have your own 3 counter-attacks of 2 ( = 6) plus survive those 3 AI attacks which kill 1 each. 6 + 3 = 9. So, if you're lucky and the AI is too busy elsewhere, you could get away with only 9 armies on SPole, plus two for Halley and Mawson minimal defense. That's 11, compared to 25. Also, if you're lucky, the AI won't attack at all, so you'll have a big stack on the SPole at that point (well, if you consider 5 a big stack).

As a hunch, I think you'd probably be able to get by with deploying 15 the first turn, and 10-15 the second turn. Might not get it in exactly 2 turns, but then again you might.

Before I used this trick, taking Antarctica used to be one of the most annoying and frustrating parts of this challenge. It's worthless to try to take the surrounding territories and *then* the SPole. And it's worthless to take the SPole one turn but fail to defend it sufficiently (with room for counter-attacks) the second turn.

Summing up:
  • Be prepared for red to deploy all (plus a card) and attack Mexico with 8 or possible more. Eliminate red as soon as you get a good chance. Be careful, l.blue might also attack Mexico because it has 4 in it.
  • Definitely take Baja. Don't lose Mexico. Don't worry so much about East USA, but perhaps defend Florida if you have troops left over.
  • Push north in Europe and Russia. Separate grey from blue; keep grey and l.blue together. Therefore, prioritize Scand. and West Russia even though these don't move you closer to the far points. Don't worry so much about pushing into Greenland this turn, but perhaps defend Iceland if you have extra. If not, then move that 1 out of there, it'll attract an attack of 4.
  • Definitely take Hong Kong. Use a big wall on Tibet, as per the one-trick pony. Simultaneously, try to take India and Myanmar to close up your borders with l.blue (will require defending Pakistan too, of course). Don't worry so much about West China or Central Russia yet (well, maybe go for the west portions of Central Russia, to go along with West Russia, pushing north), but perhaps take a stab at Irkutsk in East Russia if you can afford it, just to be sure it's broken. Low priority, since purple will fight l.blue.
  • Perhaps try the Antarctica trick if you think you're ready to afford it.
  • When you get to trying to take Central Russia, beware of Tiksi.
Well, so much for not commenting on your specific game. I couldn't help it. :-)
Crazy Challenge: 11/7/2015 01:25:21

wct
Level 56
Report
Platinum, no probs. It will help anyone who wants to, also, so it's much easier to justify spending the time on it. Took a while to learn how the AI works; might as well save other people the frustrations. :-)
Crazy Challenge: 11/7/2015 01:35:50


Potatoe
Level 57
Report
Thanks WCT, Though I would recommend making a seperate thread as the tittle "Crazy Challenge" Will wave off most people, Afterall 3 different people posted in this thread.
Crazy Challenge: 11/7/2015 01:53:30

wct
Level 56
Report
Noteworthy: Light Blue decided to attack my 1 troop in Pakistan from India with 12 troops, and Tibet with 1.

Oh yeah, forgot to comment on this. This is one of those quirky 'exceptions' I mentioned to the general rule of 'always exactly double your defenders'.

First off, there are a couple possible reasons why he attacked Pakistan with 12: 1) If the AI has way more income than it needs, then it will randomly pick one of its deployments to be "... and the remainder I'll dump here". And then these will always get lumped into some pre-planned attack so that the attack is much bigger than double-the-defenders. 2) Taking Pakistan would re-take the SE Asia bonus, so I presume there's some bonus/income prioritization at play in the AI's preferences of where to attack with an excess. Something similar may explain why the AI will tend to attack the SPole with more than doubles. 3) Pakistan (or India) might be one of those 'sticky' places that the AI prefers for other reasons (Pakistan has 5 connections, India only 3).

Of these, I'm guessing a combination of 2) and 3). I do seem to recall the same tenacity of the AI to fight hard to keep SE Asia.

Now, as to why the AI sometimes-attacks-with-double-the-defenders-minus-one: It's exactly that. The AI is programmed to sometimes attack with double the defenders minus one. There's a random chance that it will make one of its attacks like this.

But actually, it's a plus-one-minus-one situation. What's really happening is that *sometimes* (randomly decided), the AI will make *one* of its attacks from a territory with 1 extra army, while at the same time making another attack from the same territory with 1 fewer army. Example time. Suppose the ideal double-the-defenders attack list would be like this:
* Attack A with 6
* Attack B with 4
* Attack C with 2
* Attack D with 2

Sometimes, it might mutate this to:
* Attack A with 6 + 1 = 7
* Attack B with 4
* Attack C with 2
* Attack D with 2 - 1 = 1

It also seems to me to be somewhat related to the way it will send any remainders along with one of its attacks. It seems that the +1 tends to go along with the "remainders" attack (I'm not 100% sure of this, just a suspicion. It's hard to verify without investigating it further).

The "remainders" attacks are like this. Suppose the 'ideal' attack list is the same as above:
* Attack A with 6
* Attack B with 4
* Attack C with 2
* Attack D with 2

That requires 14 troops. But suppose that this attacking territory already has 450 troops on it, available to attack. In that case, the AI will pick *one* of those attacks to carry all the 'remainder' troops that otherwise aren't really 'needed' for its attack list. 450 - 14 = 436. So you might get:
* Attack A with 6 + 436 remainders = 442
* Attack B with 4
* Attack C with 2
* Attack D with 2

My hunch, unconfirmed, is that the random +1s tend to go along with the remainders, so you'll see:
* Attack A with 6 + 436 remainders + 1 random = 443
* Attack B with 4
* Attack C with 2
* Attack D with 2 - 1 random = 1

More often than:
* Attack A with 6 + 436 remainders = 442
* Attack B with 4 + 1 random = 5
* Attack C with 2
* Attack D with 2 - 1 random = 1

So, wrapping it all up, I'm guessing the AI dumped its 'remainder' deployments into India, and then randomly (or by some priority scheme) chose one of its attacks to dump those 'remainder' troops into. And at the same time it randomly shifted +1/-1 on those attacks:
* Attack Pakistan with 2 + 8 remainders + 1 random = 11
* Attack Tibet with 2 - 1 random = 1

Not 100% sure, but it works pretty well as a rule of thumb. Those 1 attacks made *some* (small amount of) sense when the game still used luck and weighted rounding. However, since the game recently changed to be primarily 0% straight-round, those same 1 attacks now never work. It's a quirk/flaw in the AI that hasn't been updated to handle changing amounts of luck (esp. no luck straight-round).
Crazy Challenge: 11/7/2015 08:11:38


wêr
Level 28
Report
Your advice is great, many cheers to you, awesome stuff!

Reading your posts a couple more times before I continue game 1.

It seems the topic title cannot be changed, given all your info I would like to.

I might continue the game tomorrow, and today think my moves through a bit more.

Again, thanks for your replies, the effort you've put in those words seem off the scale to me.
Posts 1 - 18 of 18