Random thoughts on QBs.

[SgtMuhammed]

Is it just me or has anyone else noticed that if you set the force mix to random you tend to end up with infantry? I just generated 15 battles in a row and ended up with an infantry force mix in each one. Not a major issue but it does get old after a while.

Also I have noticed that some really strange maps have started coming up recently. I just had one that looked like a giant waffle, lots of little hills without one tree.

[vedder]

I love the roads that lead nowhere, reminds me of that elevator from the simpsons.

[Other Means]

yeah, i've noticed that. i tend to pick my own forces but i've started making the AI pick armour, if left to itself it wouldn't even attempt to win the armour battle. it would end up with 1 med tank & 10 HT.

so now of course it ends up armour heavy. at least that gives some sort of challenge but i'd rather be facing a good mix.

your point about the map: this seems to tally with a few things i've seen in the forum over the past few weeks.

when people were testing the unhittable AT gun they had large numbers of tanks firing at it, a lot of the tanks shells hit the same craters.

some one was saying how a shrek went on a good run & killed 2 tanks, not missing with 5 shots. this is possible but not probable.

your map ended up being regular in its hillyness.

now, all these factors share a common thread. the random number generator. i was wondering if it's possible that the RNG is not quite as random as we'd like?

this all kind of reminds me of a Dilbert cartoon where he went to heck (a toned down hell) & there was a demon there that acted as a RNG. it just kept saying "nine, nine, nine, nine..." and of course no one knew if it was random or not.

in other words, all these things above are possible but if we plotted the RNG number produced against a graph, would the graph average out or would it have distribution peaks?

so:

a) AI's force mix not good.

possible small glitch in RNG.

<hr>

thanks for the OPFOR info last week BTW.

[ March 12, 2003, 05:01 AM: Message edited by: Other Means ]

[Other Means]

someone's got to have an opinion on this.

[MikeyD]

Wow, this is the first time in a while I've read a complaint about the Quickbattle maps. I've been happy-as-a-clam with 'em! Did you just get one especially bad map or is this something you've started to notice since v1.02?

[demoss]

in other words, all these things above are possible but if we plotted the RNG number produced against a graph, would the graph average out or would it have distribution peaks?

Almost certainly. Even bad random number generators usually return a good distribution. The real issue is usually correlation (i.e., does getting a 1 result mean you are more or less likely to get a particular result next time?).

If you want to read about this until you're bored silly (assuming you're not silly to begin with ), you can start by taking a look at one of the "Numerical Recipes in [C | Fortran | whatever]" books.

[Biltong]

Originally posted by vedder:

I love the roads that lead nowhere, reminds me of that elevator from the simpsons.

vedder - have you tried the BCR maps - over 140 at the moment and within days there will be over 200 of them!

All designed specifically for QB's You download a list in Excel which gives you Year/width/hight/trees/hills/map type/battle type etc etc - choose the one you like and then download the map pack it's in. Once you've generated the battle you simply import the map and bob's your uncle... Great looking maps with all the eye-candy you want and without the predictable flags and setup zones.

If you want to be surprised for a change - try one

See my sig below.

Biltong

[Battlefront.com]

Random unit select isn't totally random. The system uses Rarity as a basis. The chance of getting this or that force type depends on the Nation, Month/Year, and Region. For example, you won't likely see any armor in the Finnish Region if you play the Germans. If you pick Central Regaion in July 1943 I think you would see a much higher chance of getting Armor as either Germans or Soviets, for example.

The same is true for putting Nation on Random. This is so you don't Get Italians in 1945 or Finns fighting in the South.

When you set the map to be Random, these settings are also influenced by other factors. Region to be exact. This way you don't see Steppes in the Northern region for example.

Therefore, in these cases "Random" is perhaps a bit better thought of as "Computer Chosen" since the latter does not imply chaotic choosing. The reasons for this should be obvious. Just in case it is to make sure that QBs "feel right" in a historical sense. Without a guiding hand for the "random" choices the games generated would generally suck IMHO

As for odd QB maps... I don't think there is a systematic problem. Or at least not one serious enough to investigate fixing. Especially since CMBB offers the opportunity to use imported maps, just like Biltong suggests.

Good player designed maps will always kick butt over randomly generated ones.

Steve

[Poor Old Spike]

The great thing is that QB randomly-generated maps ARE completely random and fresh every time,making them excitingly dangerous and unpredictable!

("Take that ground!" says your Commander,jabbing his finger on a map)

By comparison,human maps are artificially tweaked and manipulated,and a tad phoney to my way of thinking :cool: