Book of Openings

[VictorIn_Pacific]

Is there any taste for putting together an "openings book", which sets out opening moves for specific maps? For example, I have written down a standard opening for Germany for New World Order, so that I don't have to recalculate it each time. That was pretty easy. But I also have a very detailed description of the first couple turns in Middle Earth, which I have spent literally months on, and it's still far from finished.

[beelee]

@VictorIn_Pacific Hi Victor. I think war club is where you wanna post strategies.

Or are you thinking of scripted moves for the AI ?

But yea, I think there would be interest. Actually, come to think of it, some games have their own threads. You could post there or start one if needed.

[redrum]

@VictorIn_Pacific Yeah, openings for maps would be welcome. IMO, actual openings are best displayed as a save game and then general strategy for nations added in text. A good example is this TWW guide: https://forums.triplea-game.org/topic/661/total-world-war-strategy-guide. Which then links to some threads that have save games that show openings and examples.

[Captain Crunch]

good idea!

[VictorIn_Pacific]

In response to beelee:

I am not proposing a strategy, but rather an addition to the game. Not exactly a part of the game, but a supporting document.

Can you write fixed moves for the AI? Let’s consider some cases.

1. Chess. It’s been around for hundreds of years. In principle, zero randomness. In practice, some randomness. Approximately 5 legitimate opening moves for White; others are inferior. Approximately 10 each replies for Black; others are inferior. Then for several moves, no more than 5 reasonable candidate moves, often as few as one. Overall, the Book of Openings compiles hundreds of years of study by thousands of people; it’s a very thick book. A human player cannot be considered decent unless he has mastered a large fraction of that information. Today, common chess computers routinely beat anyone not a grandmaster; cutting edge programs routinely beat grandmasters. Those computers use built-in opening databases for at least a few moves: no calculation is involved.

2. Standard AAA. Much randomness. Strategy is at the same time simpler than in chess and more complex. The randomness makes exact opening theory impossible. You can, however, formulate objectives and try to implement them. But it would be quite different from what can be done in chess. It would be like the difference between thinking and memorizing.

3. Low luck AAA. In many situations, zero randomness. In many more, the randomness has little effect. Now, this is the situation I was referring to in my first post here.
   I think it should be possible to write down a definitive move for the first player. There might be exactly one, with all others being inferior. There might be several. Or players might be willing to tolerate slightly inferior variations for a number of reasons.
   For the second player, we can suppose that there might be a few moves that are optimum or nearly so. But if there were 3 reasonable possibilities for the first player, and 3 each for the second player, now your database is starting to get quite large, and it’s still early in the game.
   Here’s where it gets tricky. Even if your game engine is able to provide a large openings database, how is the program going to select among the several possible responses? It would be easy for a human, but not for a machine. How is the machine going to recognize which variation was played in the previous move? What if there was a small deviation from the exact variation? What if it was a critical deviation? Overall, I think it is impossible to create an Openings Book for an AI to use that would be comparable to what is today easy for chess.
   What should be possible is to create a scripted opening for the first move only, simply for the purpose of reducing calculation time and effort, not that it seems to be a problem. It may also be possible to do this for some unconnected later moves. There is also the advantage that you can force the AI to make what are known to be good moves.

[VictorIn_Pacific]

With respect to the first move for Germany for NWO, you have about 50 units, and to accomplish the maximum, you need to move specific units to specific locations, and it's not generally obvious which unit should go where. So if you start at one end of the map and work your way across, you will find that you have insufficient units somewhere and too many somewhere else, and because you can't just teleport the extra units to where they're needed, you have to undo the whole move and start over. Even if you've worked out the best overall arrangement, you won't remember it, next time you play. Thus, you write it down. And you may as well publish it.

With Middle Earth, it's a lot harder. Time and time again, you find that you have to move exactly the right number of units to a specific location, otherwise bad things will happen. And I'm not talking about calculating that 4+1+1=6, and 3+2+1=6, but I choose the first option because if one of my infantry die on the attack, I still have an infantry left there to block.

Example: I have a group of 10 cavalry units. I want to send them to guard the flank of an allied faction for an attack that has not happened yet but is expected to happen in the future. Should I send 10 or less? If I send 10, I will be short one critical unit to guard my capitol next turn, assuming that 2 other factions in the future do exactly the right thing to be able to attack my capitol. Should I send 8? Then my guard force is subject to an attack, which has a 15% chance of wiping out my guard force. Then the faction I am protecting would have to fight some battles it was not expecting, and then, after several other player factions react, a stable situation is reached. Can I tolerate that? Maybe not, so it looks like the optimum number to send is 9. I also have to consider if my home areas can handle sending all my cavalry but one away. Most of this could not be worked out over the board, and that means you will almost certainly be making inferior moves almost all the time.

This is not really a strategy guide. It is a totally specific description of exactly what 700 units must do to survive turn 1. It comes with detailed descriptions of the bad things that happen if you don't follow the instructions, assuming your opponent has read the document.

[beelee]

@VictorIn_Pacific yea i was thinking more like 4 or 5 optimal 1st moves for the AI and it would randomly choose.

Seems as as if you got a good handle on what you're after.

On a side note. Chess championship was happening recently. Might still be going on. I haven't followed that closely. Was 8 draws in a row last i heard.

[VictorIn_Pacific]

Of course, this would need to be a collaborative effort. There are people who are experts on certain maps, and they might be willing to offer their knowledge. Otherwise, it would be too difficult to do.

[VictorIn_Pacific]

We should start to solicit contributions.

I have

1. A very basic first move for Germans and Italians in NWO. I did that because it's too complicated to be obvious, and saves a lot of time, instead of recalculating it every time you play the game. It is optimised according to certain metrics.

2. An insanely detailed turn 1 for LotR, which I finally completed after many months of hard work. It is mostly deterministic, that is, any deviations, even to the actions of a single unit or PU, can be shown to be inferior. However, there are sections that remain unclear, that is, it isn't reasonably possible to determine the final outcome. There are also sections where divergent choices are possible, each having advantages and disadvantages.

I cannot find a clear path to victory for either side, but I still think that the Free peoples have the advantage.

[Captain Crunch]

well maybe this thread will fill up with 1st moves for most of the maps!

Ofcourse I played the WWIIClassic map and board game of it so I guess I will give my strategy for that one and goodluck collecting and documenting all the 1st moves!

WWIIClassic map: Allies - the 1st moves for the Allies is usually have Russia buy mostly Infantry and 1 Tank and push back Japan as much you can but concentrate on Karelia and holding it and then pressure Germany through Eastern Europe while your ally United Kingdom builds a Navy and gets into Western Europe together with your ally the United States who is sending Infantry on Transports to Western Europe and so basically if the United Kingdom can hold its Navy and Transports then eventually the 3 Allies can squeeze Germany and take its capitol first and then team up and take down Japan's capitol

Axis - the 1st move for Germany usually dictates the rest of the match vs the Allies. Germany tries to assault and take Karelia with as much as it can (save something to take out some of the United Kingdom's Navy) except for a Transport and Battleship to help start the takeover of Africa. As Japan most people do the "Pearl Harbor attack" to start which I do (use a couple Fighters and most of your Navy) and then you want to build a Factory quick and start marching as many Tanks and Bombers you can to go across Asia and assault Russia and if Germany can hold off the 3 Allied countries before Japan takes Russia's capitol then the Axis should be able to take down the Allies

[VictorIn_Pacific]

@Captain-Crunch Now, that is a good start: a clear, concise, and useful explanation. But I must ask if you are willing to take it to the next level.

Take the Pearl Harbour raid, for example. WHY do you do this? A possible reason would be that you inflict more TUV damage than you take, after all subsequent counterattacks result in a stable position. There might be other reasons.

Then, WHAT do you send? All available ships? That seems like a good choice. What about airplanes? Probably, every plane that could make it there could do something useful somewhere else, so it is a real choice.

And HOW MUCH do you send? Assuming this is a dice game, do you send exactly the average amount to ensure success? Or do you send extra units?

[Captain Crunch]

@VictorIn_Pacific ok I understand now. I doubt I'd take the time to dictate every attack move to start if thats what you meant! I have save games of the exact scenario for WWIIClassic map in the AI Discussion thread and could post the game but thats not for your book of openings document so good luck documenting all the maps lol!

Maybe I'll update my post above and be more specific there.

So for the Book of Openings make an official announcement somewhere besides this thread I guess but its an ambitious endeavor so keep asking and then ya that is helpful for beginners and would help me if I ever finally tried different maps hehe.

[VictorIn_Pacific]

@Captain-Crunch Well, it's not MY Book of Openings. It's all of ours. So the rules and the format will be decided by anyone who wants to contribute to it. And of course, anyone who wishes to add to anything someone else did will be welcome to do that.

[VictorIn_Pacific]

This is what I have for NWO. Very bare bones, but even so, quite long. Pure text, no intention of doing anything else for this. It would need to be printed, and used as a reference document. Good luck memorizing it. Other formats would be less useful. Purpose is stated below. Corrections welcome. Alternate moves are at the discretion of other people.

---

NWO

The purpose of this document is to set down the optimum first move for the Germans and Italians, or at least an optimum move. We call this the "classic" move. The main objectives are to seize the maximum amount of territory, kill the maximum number of enemy units, and take the least casualties, but across the board, not necessarily in any given battle, although most battles are resolved in one turn. There are other objectives which are harder to explain, but they might be apparent when looking at the map.

This is actually a conservative move. Very few risks are taken, and nothing outrageous is done. The move is deterministic, not probabilistic. Thus, if a unit is to be killed in one round, 6 attack points are sent, to the extent possible. Sufficient of the cheapest units are committed to each battle to absorb all casualties, including from potential counterattacks.

The process used to determine the moves was to send the slow units first, starting with situations lacking choice, and then to add fast units as needed. It turns out that there is very little choice overall as to which unit goes where.

The reason for writing this down is that it takes quite a long time just to execute the instructions, and much longer to do the move if you had to recalculate it each time. Much remains unstated, even for just the two factions covered, but standard TripleA considerations apply. Even so, this document is already quite lengthy.

For the Germans, there is generally enough force available to accomplish all of the primary objectives easily. For the Italians, this is not the case, so their move is less deterministic.

Notation: I = Infantry, A = Artillery, T = Tank, C = Armoured Car, B = Bomber, eF = Early Fighter, S = Sub, D = DD, Bt = Battleship. A combat result of "KIA n" means that the target is eliminated in n combat rounds, although in a few cases, that is only the likely result.

German turn 1

E. Czech 3IA -> Lvov: KIA 1
W. Czech 6I, E. Czech 3IT -> Krakow: KIA 1
Gdansk 3I2AT -> Koenig: KIA 2
Wroclaw 8IAT, W. Czech 4TeF, Berlin eF -> C. Pol: KIA 1
Rostock 6I, Hamburg 5I, Dresden 3T, Berlin 3T, Bremen I -> Denmark: KIA 1
Bremen 4I2T, Berlin 2T, Koeln 5IT -> Neth: KIA 1
Frankfurt 8IT2eF, Bremen eF, Berlin eF, Muenchen 4I2TeF, Nuernberg 3TeF, Austria 2T -> Metz: KIA 1
6S -> W. Ice: KIA 1
4S -> E. US: KIA 1.67
6SD -> S. Gib: KIA 2
3IA -> Alger: KIA 1
2B2eFBt3D2S -> S. Fin: KIA 1
2I2T -> Fin

2T unused

Buy 10 C + truck for France, 2 transport + sub for N. New production infantry cannot reach Paris on turn 3.

Naval battles can be a bit tricky. In the presence of enemy subs, at least one DD must survive till the end, otherwise the subs get free shots next round. Also keep in mind that subs don't coordinate with other units; 5 combat factors each of subs and other units may get zero hits.

Italian turn 1

Note that the Italians could capture more territory in the Balkans. This is deliberately not done, because the Romanian economy needs to be increased, and they have few expansion opportunities.

3I2AT -> Serbia; fight one round (to help Romania)
4I4T -> Marseille, also 2I2A via sea: KIA 2
Bt2DeFS -> E. Tunis: KIA 1
IT from Albania, also 3I3A -> Tunis: KIA 2
2I2eF -> W. Tunisia: KIA 2?

Buy 2 transports at Genova to invade Morocco next turn with 8 units; also fly 3eF to Alger. The rest is not critical.