The 90% rule

[LichKing]

Basically you're telling me that in reality there isn't any reroll even now? I thought there was. Then ok.

If there is a reroll, why not to take my case 2 (if outcome is in the worst 10%) or 3 (for outcomes in the best 10%) directly, without reroll? Just for faithfulness of the battle rolls feedback?

[KGB]

LichKing,

There is definitely a re-roll now. That's done so that the battle screen and the detailed battle report all match what you see in the game.

I agree with your idea of Case 2 and Case 3 and not needing a re-roll (that's how I would have coded it). I would simply change the battle screen so that if something end up in the best/worst 10% to display a message saying 'Extreme Battle Result! Reducing to the Mean' and then show the closest valid result and use that.

So for example if 3 Dragons attacked a Bat and somehow the Bat killed a Dragon (or even 2 or 3) it would display that message and then show the closest valid result is 3 Dragons surviving which it would use. With the 8 Giants vs 8 Giants if one side won with 7 Giants left, the message would be displayed followed by it showing 4 Giants surviving which is the closest valid result.

However that may still confuse players especially ones who aren't using English as their first language. I also think it would annoy players on the times when they are getting their battle result reduced (ie when you overachieve by either losing less men than expected or killing more men than expected). Right now you can't tell when a re-roll happens and when it does happen you don't know whether you or your opponent is the benefactor of the re-roll.

KGB

[LichKing]

Anyway I still think that the reroll makes critical strike more useful (if units with critical are first in the fight order). Example:

pala+medusa vs kraken (let's say more than 90%)

1st roll: battle lost -> reroll -> medusa kills kraken with critical.

In this case the medusa has had 2 chances for critical. Since it's impossible to lose that battle, and since it's even probable to lose it if the medusa fails its ambush roll, you have that almost always both units will survive. I mean, it's not that I dislike this in the end, it makes medusa slightly more useful since it's not an amazing unit right now.
In a less extreme measure, also for orcs you have a slightly better chance of free ambush rolls (if it doesn't fall in the best 10% fraction that is discarded). But I'm not sure.

[KGB]

LichKing,

Not sure why it matters/makes any difference.

1) Pally/Medusa v Kraken. On the re-roll what difference does it make whether the Medusa kills by ambush or misses the ambush and kills the Kraken by regular combat. Either way the result is the same, both units survive. This will be reflected in the winning percentages that put this battle over 90%. It will also show the Medusa as highly likely to survive simply because of the ambush roll.

2) Pally/Medusa v Crow/Kraken. First time Medusa ambushes the crow but the stack loses the battle. On the re-roll the ambush is missed and you can pick whether the Medusa survives or not. But either way, second time around the ambush was missed.

So not seeing anything here that matters or makes Ambush more/less useful. Ultimately the only thing that matters is the final result.

KGB

[LPhillips]

KGB,
I think Lichking's thought process is that the outcome of any battle can be greatly swayed by the Ambush ability, which short-circuits the normal battle mechanics. In the case of combat where the ambushing' unit's opponent has much greater strength than the ambushing unit and those behind it, then missing the ambush means the ambushing unit's side comes up with an extremely unfavorable result. With that outcome rejected, another opportunity is provided for the Ambush ability to do its work and produce a more moderate battle outcome where strong unit(s) are eliminated without normal battle rolls.

However, Lichking, the faults in this logic are threefold:
First, the battle simulations which determine statistical outcome are run entirely without application of any sort of modification to their own outcomes. So the selection of outcomes rejected is always fair, and outcomes either greatly favoring the Ambush ability or greatly disparaging it are equally rejected if they are unreasonable (read here "obscenely unlikely").
Second, each reroll has the same chance of activating the Ambush ability for any particular unit as the last roll. All rolls are fair; it is only the outcomes which are judged. The game logic doesn't consider the path taken, only the destination. This was KGB's objection to your argument.
Third, all players have equal access to unit selection and employment. We are free to use Ambush, Swarming, or other abilities. This is a particularly potent argument to consider when addressing basic battle mechanics, as everyone plays the role of aggressor and defender in the course of any game.

[Ahla]

Exactly what does the 90% rule say? You do a lot of simulations (1000?), sort them, then roll a real battle and reroll if it is within the top or bottom 10%? Or do you just pick a number within the 80% that was eligable and use the corresponing simulation result as the battle result? And I guess this makes it meaningless to attack somethin really powerful with an orc, hoping for a 6% ambush. Or am I missing something?

[KGB]

Ahla,

Exactly what does the 90% rule say? You do a lot of simulations (1000?), sort them, then roll a real battle and reroll if it is within the top or bottom 10%?

Yes

And I guess this makes it meaningless to attack somethin really powerful with an orc, hoping for a 6% ambush. Or am I missing something?

Yes. Unless the ambush chance (6%) + regular battle chance > 10% it's pointless. However 2 Orcs will give you a chance to kill something really powerful since the chance of any unit surviving 2 ambush rolls is only 88%..

KGB

[Ahla]

Are you sure about that KGB? There are several ways of making sure that you avoid the "extreme" results, and which one you choose makes a difference to the overall propabilities. Found some old threads from late 2010 where this is discussed. Not sure if this still applies. Warbarons seems to be in rapid developement.

I generally don't like avoiding randomness in games like this, since it can trigger strange side effects and make some units stronger or weaker if they often appear on the right or wrong side of the 90% barrier. Some games will be won or lost on battle luck, regardless of how you try to avoid extreme results. On smaller maps, running good or bad in the first few battles between players often decide the game.

At least, if ambush should be a part of the game (and I think it should), the tolerance for "extreme" results should be high enough that these chances doesn't vanish.

[KGB]

Ahla,

Yes I am sure. Threads from 2010 are very old now and quite out of date as you have realized.

On smaller maps, running good or bad in the first few battles between players often decide the game.

That's why I don't play 50x maps. Luck plays a much bigger role there than on larger maps.

KGB

[Argammon]

[