Combat Odds?

[Nugog]

Originally posted by Blake
...the odds mouseover is reliable enough.

Works for me!

[Targonis]

The flaw with the "it's just a bad streak" in the random number generator is that by doing something else(moving other units that are not a part of the battle), in theory the random number generator should then be different.

If the random number generator is used at the time of the attack(based on the seed), then doing other things in that turn BEFORE you initiate the attack should in theory change the random numbers generated for the combat.

I am NOT saying that the percentage chance is wrong or anything like that, just that in spite of all of the evidence to the contrary, there does SEEM to be the "that defending unit is too lucky for it to be coincidence". I am also not attacking or refuting how the percentages work.

If anything, it may be something funky in the game code itself, not in the XML, not in anything a mod would fix. Most of those who deal with mods and the XML will look at the code trying to find a problem, and when they find the code is "perfect", they assume that everything is working properly. What if the problem I(and others) have noticed is due to some programming bug that has nothing to do with the XML parser?

1+1=2 should always apply, but if there is some very weird glitch that makes it 2.00001 one time per 1000 calculations? It could be a DirectX bug that has never been diagnosed, or something else. This sort of thing is really really hard to nail down, but it is quite possible that is the source of some strange behavior.

For the record, for those who don't remember such things, weird problems have shown up in CPUs/processors from time to time. So, I don't blame the programmers for this, I just wonder if this symptom has been really investigated since it just seems to be something many of us have seen before.

[Risa]

I think the game doesn't draw all random numbers from a big sequence. Rather than that, it sets some seeds to something every turn, so each event draw random numbers from it own sequence, with a fixed seed. That is, no matter you take the combat in whatever order, as long as it is between the same two units, it will use the same sequence of random numbers. And if other conditions are not changed, it will have the same result.

Of course, that is only my guess. I didn't look into codes to prove it.

[rah]

From my experience, it is sequencial.
If you do a task prior that requires a random number it will then be different when you do the attack. I did play with it a bit to check this (not during a real game).
One battle it was at 85% and I lost, reloaded, (same seed) and tried the attack with a stronger unit 91% and still lost, Did it again at 98% and finally won.
Using a cat or a unit with a withdraw % might give you a loss but a retreat. If you want to be real anal, by ordering your attack many different ways, you can maximize your result. (keep on 99% in reserve for when you get the bad roll) But if you're going to do that you might as well use the option to change the seed on reload and keep reloading till you win every battle.

The computer does roll the preverbial 00 eventually.

I don't do it, but whatever floats others boats, since the end goal is personal enjoyment.

[rah]

And for those unfamiliar, most random number generators just start with a seed number and perform and arithmatic operations to generate the next number and use the new number as the next seed.

[snoopy369]

RNG is definitely a sequence. It's a single (rather, two) sequences for the entire game. That's how MP works - each game is running entirely on each separate computer, and you each have one RNG sequence that matches each other's RNG sequence. Then as long as everything happens identically on each computer (ie, nothing happens to make one computer show a catapult attack where another doesn't) it has the same effects on each game.

Moving a unit is not enough to change the RNG seed, you have to actually have a combat or something else that actually pulls a random number. There are also two random number lists - I don't know precisely which does what, i think one is for AI decisions and one is for player results or something like that.

Either way, it's random. Streaks are entirely psychological, except when they're a normal random artifact. You remember the times when you lost a bunch, you never remember the wins. You will have several 0s in a row sometimes.

I'm fairly sure it's still incorrect a bit on some of the shown percentages; I still win more than I feel I should when playing with FS units (archers, particularly FS archers).

[Ari]

Originally posted by NackBarnes

So, this pretty much puts the B.S. to the "experienced but obsolete units can beat newer green units" civ iv help blurb, eh?

I wouldn't go that far. An archer with 3 city defense promotions and a combat promotion (a difficult but not impossible level for an archer) can make an attacking level 1 maceman's job *quite* difficult, especially when maximally fortified behind a walled city with its own cultural defense. Someone else could crunch the numbers but I'm pretty sure a maceman would almost certainly lose that attack. A knight probably wouldn't be too favored either (even allowing for its ignoring first strikes). In both cases, the archer is clearly obsolete, but is more than holding its own. Even in the open, an archer with multiple combat and first strike promotions could make either of those units' lives miserable (though I'm pretty sure it couldn't win).
Similarly a knight with combat V and pinch could theoretically clobber level 1 unfortified rifleman.

[snafuc4]

FWIW WRT the RNG, I've always thought that a RNG 'seed reset' for games should be based upon clock ticks between keypresses/mouse clicks. After all, who can predict enough to misuse the microsecond difference, IYSWIM?

[Kuciwalker]

That would be exploitable in MP. You could programmatically perform mouse clicks at the correct time.

[snoopy369]

Not to mention not being possible in MP with a random number system such as Civ4's (identical random number strings used by all computers in a game)...