17 crusaders (MP report)

[VetLegion]

Recently in a MP game of TOT I lost 17 crusaders in a single turn to 2 fortified phalanxes and one fortified legion in a city that had barracks and was on a river. I didnt kill any of them, just managed to get them "red".

This is the calculation:

2*1.5(veteran)*2(fortified+river)=6
my crusaders were all non veteran, but still with a 5:6 ratio I should have kicked them and still have some army left. I tried to repeat that offline later and I had 7 crusaders left after conquering the city.

My questions are:
Anyone else had it happened?
Is there a difference in MP battle calculus?
Which side are the battles calculated on(host or client)?
Could it happen because of lag or connection problems?


I am worried because I noticed this only because it was so obvious. Who knows how many times I didnt notice it? Perhaps some times I lost a unit and thought I lost it to a veteran defender/attacker and but it was a bug?

[John Possidente]

Every battle in Civ has a random element. Events like the one you describe are therefore possible, but they are extremely unlikely. Sounds like you ran into the wrong end of the laws of probability. Someone has to; that's how they work.

Hope you're enjoying the game otherwise.

[tobyr]

A simple, possible explanation for what happened to you might be that the game's random number generator (or the programmers' use of the generator) is not really random, that there is a bug causing it to repeat the same value in consecutive calls more often than it should.

For example, the Microprose original release of Magic the Gathering (and first game patch) could not deal randomly from a card deck. In a deck with less than 1/3 land cards, it tended to produce 40%+ land VERY frequently.

Programmers tend to use random number generators thoughtlessly, without:
(1) Testing to see how random they are; some compilers come with poor generators.
(2) Considering whether their conversion of random values to small numbers removes some of the randomness.
(3) Relying on the low order bits of a random value, which tend to be less random than other bits in the value. That may be the case here.

I have experimented a lot with opening huts, saving, restoring, and reopening the same hut. I have not been systematic about this, but it seems to me that I am getting the same result two and three times in a row more often than I should. (But my evidence is very informal, and it is even possible that the game makes this happen intentionally.)

For example, suppose a programmer wants to select, randomly, one of eight possible outcomes from popping a hut. This line of C code will do it:
index = nrand() % 8;
but MUCH safer is this line, which ignores the low order three bits:
index = ( nrand() >> 3) % 8;

I usually ignore the random number generators that come with runtime libararies and use the well-tested Lehmer algorithm instead.

- toby



------------------
toby robison
criticalpaths@mindspring.com

[VetLegion]

I won that game so I dont complain too much. I wanted to see if anyone else reported this especially after what people reported in civ2 forums....hehe I like that about civ, the legends built around it.

tobyr
I am not a c programmer, but I think 3 bits are just enough for 8 combinations You think civ2 was written in c? I liked what they did with civ1 animated water. That was probably assembler. What do you think why they didn't animate water in civ2?

[Mark von Wagner]

Every battle in Civ has a random element. Events like the one you describe are therefore possible, but they are extremely unlikely. Sounds like you ran into the wrong end of the laws of probability. Someone has to; that's how they work.

Damned unlikely. My battle-odds calculator gives the 17 attacking crusaders a 99.99999999999998% chance of winning. Even with only 4 attacking crusaders, you have an 82.5% chance of winning.

Mark
[This message has been edited by Mark von Wagner (edited January 16, 2000).]

[VetLegion]

Mark I agree with you. I dont know. Tobyr hinted that randomize function might be badly programmed, and since I dont know in which language, I cant check.

Anyway...I have more bugs to report.
I never played MGE online so I dont know are these known to the people.

2. The Repeated Action thing
I noticed that sometimes I get to move my units twice. Just this night I built a road with settler and then moved him!? Sometimes my warriors gets a second move (I play single movement) Fun eh?

3. The Popping Screen
Negotiating screen keeps popping out although no human initiated talks. It pops out a few times in a turn and gets annoying.

I somehow think all these problems are related to move synchronization. That is why I asked who calculates the battles and all. What is the host role?

Oh, and tobyr I think I get your point. it makes sense since rand() probably returns word or byte. What do you think, maybe randomize_timer() function is only used at begining of turn, so the seed is same in all turn? These are just gueses, since I dont know c.

[Orz]

Hmm... Actually in Civ2 and in following add-ons/mods the random element is much smaller than in original Civilization.

Civ2 has hitpoints and units make several "rounds". In civ1 they made only one: remember - chariot was pretty good a weapon even in modern time!

So in civ1 I've made great battles with chariots and vice versa - have lost armors due a single chariot attack.

[VetLegion]

Tell me about it All I ever built was chariots and settlers. Some times I even built barracks, but then changed my mind just in time to build a chariot