How terrain selection works

[batesmotel]

This is a paraphrase of an explanation I received from Keith when I asked about what seemed to be anomalous map selections based on what each player had requested:

1) Both sides choose a terrain preference.
2) The winner of the initiative gets their score times 3 plus the losers score all divided by 4 for a terrain score. This produces a terrain score from 1 to 100.
3) The system now looks at how many maps there are (currently 100 30 x 40 maps for example) and sorts them by terrain density.
4) With 100 maps and without the chance modification, if the net terrain score was 75 then the 75th most crowded map would be chosen.
5) The program then randomly selects a map where the weighted chance of getting any specific map is based on how close it is to the terrain score*100/total number of maps. In this example, the 75th map is most likely, the 74th or 76th the next most likely the 73rd and 77th the next most likely and so. But any map could still be selected.

Essentially there is something like a bell curve of possible results. The wide open map when both sides chose very crowded terrain is one chosen from the tail of the bell curve.

This is a fairly sophisticated algorithm and it does make it hard to intuit what is happening, especially with a somewhat limited number of data points. And due to human nature, the exceptional results are the ones that stand out in memory.

Chris

[Morbio]

This is a paraphrase of an explanation I received from Keith when I asked about what seemed to be anomalous map selections based on what each player had requested:

1) Both sides choose a terrain preference.
2) The winner of the initiative gets their score times 3 plus the losers score all divided by 4 for a terrain score. This produces a terrain score from 1 to 100.
3) The system now looks at how many maps there are (currently 100 30 x 40 maps for example) and sorts them by terrain density.
4) With 100 maps and without the chance modification, if the net terrain score was 75 then the 75th most crowded map would be chosen.
5) The program then randomly selects a map where the weighted chance of getting any specific map is based on how close it is to the terrain score*100/total number of maps. In this example, the 75th map is most likely, the 74th or 76th the next most likely the 73rd and 77th the next most likely and so. But any map could still be selected.

Essentially there is something like a bell curve of possible results. The wide open map when both sides chose very crowded terrain is one chosen from the tail of the bell curve.

This is a fairly sophisticated algorithm and it does make it hard to intuit what is happening, especially with a somewhat limited number of data points. And due to human nature, the exceptional results are the ones that stand out in memory.

Chris

Re. Point 2) What is 'their score'? Is this the horse plus commander plus die roll?

[Blathergut]

Thanks for finding this out. At least when you see what you think is a weird map result you know why!

[SRW1962]

The system for choosing the maps sounds fair enough, its just a shame that there are:

a. So many VERY similar maps

b. So many VERY similarly ridiculous maps

c. A designers fetish for terrain that diagonally bisects the maps

d. A designers fetish for lots of water features

[Triarii]

Tangentially - how will this work.
Immortal fire will have 64 more DAG maps. So is this the future mechanism for terrain selection?

Can I assume that if I own Storm of Arrows (SoA) and Immortal Fire (IF) and multiplay an opponent
a) who owns IF but not SoA in an IF game the terrain selection is only from the IF 64 DAG maps?
b) who owns SoA but not IF in an SoA game the terrain selection excludes the 64 IF DAG maps?

[batesmotel]

This is a paraphrase of an explanation I received from Keith when I asked about what seemed to be anomalous map selections based on what each player had requested:

1) Both sides choose a terrain preference.
2) The winner of the initiative gets their score times 3 plus the losers score all divided by 4 for a terrain score. This produces a terrain score from 1 to 100.
3) The system now looks at how many maps there are (currently 100 30 x 40 maps for example) and sorts them by terrain density.
4) With 100 maps and without the chance modification, if the net terrain score was 75 then the 75th most crowded map would be chosen.
5) The program then randomly selects a map where the weighted chance of getting any specific map is based on how close it is to the terrain score*100/total number of maps. In this example, the 75th map is most likely, the 74th or 76th the next most likely the 73rd and 77th the next most likely and so. But any map could still be selected.

Essentially there is something like a bell curve of possible results. The wide open map when both sides chose very crowded terrain is one chosen from the tail of the bell curve.

This is a fairly sophisticated algorithm and it does make it hard to intuit what is happening, especially with a somewhat limited number of data points. And due to human nature, the exceptional results are the ones that stand out in memory.

Chris

The score is for the the amount of terrain they chose. My guess is that this ranges from 0 for very open to 100 for very crowded or possibly the reverse of this. This way the amount of terrain chosen by the winner of the initiative has 3 times the weight of the loser's choice.

Chris

Re. Point 2) What is 'their score'? Is this the horse plus commander plus die roll?

[Morbio]

This is a paraphrase of an explanation I received from Keith when I asked about what seemed to be anomalous map selections based on what each player had requested:

1) Both sides choose a terrain preference.
2) The winner of the initiative gets their score times 3 plus the losers score all divided by 4 for a terrain score. This produces a terrain score from 1 to 100.
3) The system now looks at how many maps there are (currently 100 30 x 40 maps for example) and sorts them by terrain density.
4) With 100 maps and without the chance modification, if the net terrain score was 75 then the 75th most crowded map would be chosen.
5) The program then randomly selects a map where the weighted chance of getting any specific map is based on how close it is to the terrain score*100/total number of maps. In this example, the 75th map is most likely, the 74th or 76th the next most likely the 73rd and 77th the next most likely and so. But any map could still be selected.

Essentially there is something like a bell curve of possible results. The wide open map when both sides chose very crowded terrain is one chosen from the tail of the bell curve.

This is a fairly sophisticated algorithm and it does make it hard to intuit what is happening, especially with a somewhat limited number of data points. And due to human nature, the exceptional results are the ones that stand out in memory.

Chris

The score is for the the amount of terrain they chose. My guess is that this ranges from 0 for very open to 100 for very crowded or possibly the reverse of this. This way the amount of terrain chosen by the winner of the initiative has 3 times the weight of the loser's choice.

Chris

Re. Point 2) What is 'their score'? Is this the horse plus commander plus die roll?

Given that algorithm then I think the selection process is flawed. I'll explain my logic.

I believe that people will tend (i.e. most times, but not always) pick an extreme variation. My rationale is that either your army is suited to flat terrain (HF and cavalry armies) or it is suited to non-flat terrain (MF, skirmisher armies). So, where both parties agree then that's fine. The maths will give you a result of 1 or 100 and in theory, you should get a battlefield that suits both. But where players have picked at the opposing end, then the maths is either going to give ((3*100) + (1*1))/4 = 75 or ((3*1) + (1*100))/4 = 26 so you are going to see the few maps that are centred around this numbers very frequently. Indeed, given that horse armies will more often win initiative, and will probably choose a flat terrain, then you will get a bias to 26 . (Of course if open is 100 and crowded is 1, then the numbers are reversed - but the argument is the same).

So, looking at the overall process you will see a lot of maps around 1, 26, 75 and 100 which tends to support the comments that people are seeing the same maps often. I think some sort of random element (+/- an amount, with a bias to lower numbers, so it doesn't distort the winners choice too much too often) would help increase that variation.

Personally, I still like the idea of being sequentially presented with a few maps (perhaps using the algorithm discussed above to generate the options) - just with an additional random element each time, so it could get better or worse, would be best.

[76mm]

I find it very hard to believe that there are currently 100 maps available (are those the large maps?)...In probably 75% of the games I play, I see the same half-dozen maps. I don't think I've seen more than about a dozen maps altogether, after playing several dozen DAG games. I usually play 800-1000 pt games, if that matters.

[batesmotel]

My guess is that the 100 maps are the 40x40 ones used for 400-600 points.

Chris

[Morbio]

I find it very hard to believe that there are currently 100 maps available (are those the large maps?)...In probably 75% of the games I play, I see the same half-dozen maps. I don't think I've seen more than about a dozen maps altogether, after playing several dozen DAG games. I usually play 800-1000 pt games, if that matters.

Or it supports my theory of the maths giving the same answers every time

[SRW1962]

A a test for terrain selection myself and my regular gaming buddy both picked extremely open terrain for a few games and both picked extremely dense terrain for a few other games and guess wahat, we ended up with some of the same maps in both sets of games, and furthermore, the density of terrain was the same in all of the games. Personally I have found that no matter what terrain choice you pick it all ends up pretty much the same and most of them are very silly maps to say the least.

With the exception of encounter battles the battlefield was usually chosen by one or other sides, so at least one of the CinC's was happy, but way too often the way it pans out in FOG neither of the Cinc's are happy with the battlefield and they battles are more like encounter battles than the setpiece battles usually associated with Ancient/Medieval warfare. Less maps but better terrain layout is far better than more maps with poor terrain layout.

Historical battlefield maps are a good place to start when it comes to designing maps.






edited for a spelling mistake

[deadtorius]

Thats a much better idea, maps with open centers and less weirdness blocking things all over the place would be a nice change of pace.
Perhaps a different option would be for the game to show the winner of the initiative 12 maps and let them choose which one they want. The maps could be chosen with the current random method but instead of just popping up at least one general would have a choice of battle field. This might help one of the generals get to the happy place to fight.

[maximvs]

A refinement would be to let the winner of the initiative chose say, 5 maps from the larger selection, then the defending general could have the choice of which of the 5 he most liked?

Yes?

[deadtorius]

that could be more complex and make the start of the game take longer than letting the initiative winner choose one form a list and let the other guy suffer. Winning initiative is like being able to choose the ground you want to fight on and waiting for your enemy to arrive, they get little say in things although in the TT rules you choose the terrain type but both players lay the terrain down depending on what is available from the type list. So both players have some more control over what goes on the table. you then roll to see what stays and what goes so the table does not always work out for you either, even if you won initiative.

[Malty666]

I have to agree with SRW1962 and 76mm - you do tend to see a lot of similar maps - no matter what.

I play a broad variety of armies, make a lot of different terrain selection choices, yet inevitably see the same maps.

And what is that diagonal fetish about?

I guess the only solution is to wait for more maps, or to open up the game map generator to allow you to import custom created maps. The community would soon create a few hundred I'm sure, and most would be based on historical maps.