forum.freeciv.org

one small question: is the defender strength (either AI only or general) higher than usual for the last unit left defending a capital city?

I just attacked a capital with marines:
- city has city walls, terrain grassland thus no terrain def bonus, no river
- all units have full health at start of the fights
- first fight: attacker has vet status hardened, defender is pikeman simple veteran status: attacker strength 14 HP 20, defender stength 13.5 HP 10
result: as expected, attacking unit wins easily, with around 70% HP left

- second fight: attacker has simple vet status, defender pikeman simple vet status: attacker strength 12 HP 20, defender stength 13.5 HP 10
result: the defender wins with around 25% HP left

the expected result for the second fight would be a win for the attacking marines unit, with around 30-40% HP left - I first thought the randomizer would have given me a very bad result, but Ive repeated this sequence of events several times now, the result being identical each time;

since the 2nd pikemen unit is the last remaining in the capital city, I thought there might be an additional special factor ("AI's capital last defence" or something) coming into play;
Should this be the case, it would be good to add this to the combat information of the help section.

P.S.: I inversed the sequence of attacks (attacking with the weaker unit first), and meeting the same phenomenon: the first fight is an easy win, in the second one the pikemen prevails although its defender strength should not allow it to do that.

I suppose you are familiar with how combats work, if not it's explained on the wiki (http://freeciv.wikia.com/wiki/Combat).

Ive repeated this sequence of events several times now, the result being identical each time.

I think freeciv has a predictable random number generator. That is, if you repeat exactly the same sequence of events, you will get the same results. It's useful for debugging purposes (to reproduce bugs more easily).
To explain what can happen, let's have a look at the sequence of random numbers produce by the generator. Say 'G' is a number good for you, ie the defending unit loses the round. Let 'B' (bad) be the opposite (your unit loses). The sequence might look like this : The first fight uses the first numbers in the sequence. As they are well-balanced (around on half 'G'), you win as expected.
The second fight uses the next numbers. What happens is that they are very bad for you, so you lose much more rounds than on average. So your unit dies.

The unit is not stronger ; it is just much more lucky. One way of getting around would be to attack (twice) somewhere else before (you will likely lose the second combat), or simply to wait for one turn (provided there are enough AIs to call the random number generator enough).

Louis

thx a bunch louis;

what you are saying is basically that the randomizer is just set to produce a very bad result for the attacker for the fight that will occur at that time, correct?
means some unit has to consume this result;

what you are saying is basically that the randomizer is just set to produce a very bad result for the attacker for the fight that will occur at that time, correct?

Yes

Does the defender still have same veterancy level in the second battle, or did it gain veterancy from the first?

cazfi wrote:Does the defender still have same veterancy level in the second battle, or did it gain veterancy from the first?

the first defender dies in the first battle; both pikemen units have vet lvl 1 from the start;
but I think louis clarified the issue;

I've just experienced many times, attacking in early game with too little army, situations where most of the defenders in a city die in their first battle, but the one that survives then gets really hard to kill because it gains veterancy levels from the battle(s) it survive.