Fun Hypothetical

[binTravkin]

Sure but it kinda defeats the purpose to need units that cost 5 to 6 times the mineral cost in order to survive attacks by even the lowest cost enemy

Not sure I got your point, but if it was that it's not fair to fight with a 2 row unit against a 4-5 row unit I answer - nothing is fair in this life.
Such a combo is a nice tactical trick and taking advantage of it is kinda knowing rock-paper-scissors rules.
Those rules are the path to victory for any strategist IMO.

Also you are sacrificinh you mobility in the fungus if you have to wait for the conventional unit

Have you ever had your opponents rating the tech of Centauri Meditation low?
I don't.

[Flubber]

Originally posted by binTravkin


Not sure I got your point, but if it was that it's not fair to fight with a 2 row unit against a 4-5 row unit I answer - nothing is fair in this life.

My point is that it can be an innefficient use of resources to be using a 4-5 row unit to kill a one row unit sucessfully a bit more than half the time. Add in a necessity to defend it with a 2-3 row unit ???

Originally posted by binTravkin
Such a combo is a nice tactical trick and taking advantage of it is kinda knowing rock-paper-scissors rules.
Those rules are the path to victory for any strategist IMO.

I'm well aware of the various tactics available-- ask anyone that played me . I still think it is correct to point out that a worm is relatively weak alone and is the equal of units much cheaper than itself.

Originally posted by binTravkin


Have you ever had your opponents rating the tech of Centauri Meditation low?
I don't.

I like free worms but I its evry situational when I see the value in building them

[Commy]

Results of worm tests...

Each test was out of ten battles...A locosts of chiron attacks a scout patrol...the independent variable is the reactor of the scout patrol...the morale is the same and no additional bonuses apply

Fission--odds=3-2...locosts won 9 battles

Fusion--odds=3-4...locosts won 9 battles

Quantum--odds=5-10...locosts won 8 battles

Singularity--odds=3=8...locosts won 9 battles

This is very strange...of course, it makes sense that the odds should be no different, but then why would the game say the combat odds are different...I stand corrected...

[Flubber]

Originally posted by Commy
but then why would the game say the combat odds are different

The battle odds are broken for native combat against higher reactors. Its actually an easy adjustment to make. If my fusion unit would attack a worm at odds of say 9 to 7, I know the true odds are something like 9 to 14. Bottom line is that if a unit doesn't attack with odds in its factor by a factor as big as its reactor, the actual chance of winning is less than 50%

Originally posted by Commy
I stand corrected...

Glad I could help -- We all learn from others on here--

[Commy]

It is interesting...I have noticed this before...I don't have the actual data at hand, but if odds are against you, say 2-3, you almost always lose, every time pretty much...

In civ, these odds would mean you would win 2 out of 5 battles, but in SMAC, its more like 1 out of 10 or 15...I don't think it is a battle odds error, rather the random generator is different...

[Flubber]

Originally posted by Commy
It is interesting...I have noticed this before...I don't have the actual data at hand, but if odds are against you, say 2-3, you almost always lose, every time pretty much...

In civ, these odds would mean you would win 2 out of 5 battles, but in SMAC, its more like 1 out of 10 or 15...I don't think it is a battle odds error, rather the random generator is different...

I believe ( but am not certain) that the democracy game folks found some odd things when they looked at battle results. For example if a unit with a legitimate 1 in 3 chance at a particular time, won that battle, that result could be replicated a disproportionate number of times if the same turn were played in the same order by other players.

If you did the same attack but had changed a number of other unit movements and the order of doing things, your attacker might die and then if THAT turn was replicated a number of times, the attacker might die a disproportionately high number of times.

It seems that repeating the same attack a number of times will not give results that approximate even the correct batle odds.

I suspect that if you repeated your locust attacks on units of various reactors, on a different turn and using a differnt tile, you might get results where the locus LOSES far more than it should.

Its as if the game has already calculated, before the turn, the random numbers to be generated on each attack. The only problem with that theory was that the democracy gamers reported (IIRC) that the results weren't 100% repetitive

[Petek]

Originally posted by Commy
It is interesting...I have noticed this before...I don't have the actual data at hand, but if odds are against you, say 2-3, you almost always lose, every time pretty much...

In civ, these odds would mean you would win 2 out of 5 battles, but in SMAC, its more like 1 out of 10 or 15...I don't think it is a battle odds error, rather the random generator is different...

I think that your observation (that battle odds don't reflect actual battle results) is correct. The reason, IIRC, is that each battle is actually a series of "mini-battles" between the two units. The loser of each mini-battle has its hit points reduced by one. When a unit's hit points are reduced to zero, it loses the battle and is eliminated.

The battle odds, I think, are the probability that the attacking unit will win one of the mini-battles. If the two units involved in the battle both have the same number of hit points (10, say, for a fission reactor), then the first unit that loses 10 mini-battles will be destroyed.

If the battle odds are, say, 3-2, then the attacker has a 60% chance of winning each mini-battle and the defender has a 40% chance. I think that it's intuitive that the unit with the 60% chance of winning each mini-battle is much more likely to survive than the unit that has a 40% chance of winning. That is, the probability of of winning is much greater that 60%.

There may be other factors involved, but I think this is the primary reason that the displayed battle odds don't seem accurate.

[Commy]

Originally posted by Flubber

Its as if the game has already calculated, before the turn, the random numbers to be generated on each attack. The only problem with that theory was that the democracy gamers reported (IIRC) that the results weren't 100% repetitive

This might be because the string is already chosen, but the seed number isn't...perhaps certain strings will lean towards a certain outcome, but not all the seeds agree with that outcome...

[Senethro]

My observations support the "mini-battles" and the "pre-determined strings" theories. I was wondering about these things myself, but in an undeducated non-statistical way.

[Flubber]

Originally posted by Petek


The battle odds, I think, are the probability that the attacking unit will win one of the mini-battles.

err no they are not. If you had a weapon 6 attack an armour 6 with equivalent morale and no other modifiers, the chance of a win in any mini battle is 50%. If the units have equivalent hit points, the stated battle odds will reflect that. Howver if you give one of the units 4 times as many hitpoints, the odds will skew markedly in that units favor despite the fact that the chances of winning any ONE of the individual mini- battles has not changed at all

[Petek]

Originally posted by Flubber


err no they are not. If you had a weapon 6 attack an armour 6 with equivalent morale and no other modifiers, the chance of a win in any mini battle is 50%. If the units have equivalent hit points, the stated battle odds will reflect that. Howver if you give one of the units 4 times as many hitpoints, the odds will skew markedly in that units favor despite the fact that the chances of winning any ONE of the individual mini- battles has not changed at all

Why do you say that increasing a unit's hit points won't change the chance of winning one of the mini-battles? Off the top of my head, only a higher reactor will increase hit points. A unit with a fusion reactor and a level 6 weapon has a better chance of winning a mini-battle than a similar unit with a fission reactor. Similarly, a unit that has sustained battle damage will have fewer hit points and this will lower its odds in the mini-battles.

In the battle screen, what I'm calling hit points are called "Power". The other factor that affects combat is called "Strength". It appears that the battle odds are calculated by multiplying Strength by Power for each unit, then taking the ratio. It still seems to me that the battle odds represent the odds for each individual mini-battle.

[Commy]

Originally posted by Petek


Why do you say that increasing a unit's hit points won't change the chance of winning one of the mini-battles? Off the top of my head, only a higher reactor will increase hit points. A unit with a fusion reactor and a level 6 weapon has a better chance of winning a mini-battle than a similar unit with a fission reactor. Similarly, a unit that has sustained battle damage will have fewer hit points and this will lower its odds in the mini-battles.

In the battle screen, what I'm calling hit points are called "Power". The other factor that affects combat is called "Strength". It appears that the battle odds are calculated by multiplying Strength by Power for each unit, then taking the ratio. It still seems to me that the battle odds represent the odds for each individual mini-battle.

Power represents the health of the unit times the reactor...a fusion unit with 30% damage will have a power of 14...

The odds do not represent the odds of a minibattle...they are the odds of the whole battle...this can easily be proven by having two equal combat units and changing the power, either by reactor OR by health...

[Flubber]

Originally posted by Commy


Power represents the health of the unit times the reactor...a fusion unit with 30% damage will have a power of 14...

The odds do not represent the odds of a minibattle...they are the odds of the whole battle...this can easily be proven by having two equal combat units and changing the power, either by reactor OR by health...

what commy said

Bu mini-battle I assumed you mean each little hit where one side or the other takes damage. It happens quickly but you can see the slugging it out before one unit or the other explodes

[Petek]

Originally posted by Commy


Power represents the health of the unit times the reactor...a fusion unit with 30% damage will have a power of 14...

The odds do not represent the odds of a minibattle...they are the odds of the whole battle...this can easily be proven by having two equal combat units and changing the power, either by reactor OR by health...

The question is whether or not the odds shown on the battle odds screen reflect the chance of the attacking unit winning the battle. I say that they do not (unless the odds are shown as 1:1).

You suggest that they do and that this can be proven (if I understand your above comment correctly) by starting with two units of equal battle value (so the battle odds will presumably be 1:1) and then giving the attacking unit a higher reactor (for example). Clearly the battle odds screen will then show that the attacker now has more favorable odds. However, this doesn't prove that the battle odds screen display the odds that the attacker will win. You would have to play the attack multiple times and note how often the attacker wins.

I used the scenario editor to set up a position where the attacker had favorable odds of 3:2 (according to the battle odds screen). If the true odds of winning are 3:2, then the attacker should win approximately 60% of the time. I played the attack 30 times. The result was that the attacker won 28 times (approx. 93%). I then set up two other units with different weapons and morale, but with the battle odds still showing 3:2 in favor of the attacker. The result was exactly the same: the attacker won 28 out of 30 battles.

This strongly suggests that the battle screen does not reflect the correct odds. You observed the same phenomenon in an earlier post in this thread:

"It is interesting...I have noticed this before...I don't have the actual data at hand, but if odds are against you, say 2-3, you almost always lose, every time pretty much...

In civ, these odds would mean you would win 2 out of 5 battles, but in SMAC, its more like 1 out of 10 or 15...I don't think it is a battle odds error, rather the random generator is different...:

I've suggested that the battle odds screen shows the odds of winning one of the mini-battles that ultimately determine who wins the battle. I brushed off my dusty probability skills and tried to calculate the odds in the case where the attacker has a 3:2 advantage in each mini-battle (and both units have 10 hit points). I came up with the attacker winning approx. 81% of the battles. The tests mentioned above show an even greater advantage, so I'm not sure if my calculation was wrong, or what.

Flubber -- Yes, by mini-battle, I meant exactly what you said.

[binTravkin]

I like free worms but I its evry situational when I see the value in building them

You didn't get my point.

I was talking about XED.
It makes your units move like worms so you can accompany a worm with cheaper unit easily.