Great Generals

[Kuciwalker]

Settled GE's? Seriously? Why?

[ColdPhoenix]

I quite often settle great engineers if I get them early enough and there aren't any wonders that I desperately want but it looks like I may not get. You get a great payback from them over time especially if you use them in a city with lots of multipliers like ironworks or (more often for me as it's early) your capital when running bureaucracy.

[Velociryx]

I have also won battles with 1% chance of winning. The reason why most people don't remember such battles is that they don't attack with such odds.

1 out of 100 isn't nothing considering that they are way over 100 battles a game. For my taste that's too much for a warlord. Considering that most battle will rather be around 95% they just don't fight long enough to pay of.


This is a true statement, for sure, but I do not believe that it explains the phenomenon away.

In a recent game, I lost both of my warlords.

The first lost with odds of 99.6%, and the second with 99.8%.

So statistically, if I were to re-fight those battles a thousand times each, I should be quite surprised if I lost either of them more than once.

What's interesting here is that I MAY have fought ~800 battles in the whole game, making the loss of two warlords excruciatingly "unlucky."

If this had been the first and only time such "unluck" had occurred, one might be tempted to write it off as just that, but this type of "unluck" seems fairly rampant....so much so that it's unlikely that we're all just really "unlucky" which naturally leads one down the path of wondering what other force might be at work here.

Not that it's a show stopper, but it IS frustrating, and having witnessed the phenomenon yet again, it certainly has the effect of making those super medic promotions more attractive.

-=Vel=-

[Kuciwalker]

What's interesting here is that I MAY have fought ~800 battles in the whole game, making the loss of two warlords excruciatingly "unlucky."

Except that you've played many games.

If this had been the first and only time such "unluck" had occurred, one might be tempted to write it off as just that, but this type of "unluck" seems fairly rampant....so much so that it's unlikely that we're all just really "unlucky" which naturally leads one down the path of wondering what other force might be at work here.

I challenge you to actually run 1000 battles at, say, 99.8% and record the results. If there are significantly more than 2 battles (at the 95% confidence rate), come back and post it.

[Velociryx]

Yep, I grant you...I've played many games, and curiously enough, I've encountered similar losses pretty consistently. Again...I'm not going to cry over spilt milk...stuff happens and you win the game anyhow, but it is amazing, and apparently, I'm not the only one who has observed it.

If I can scrape together enough time to do a 1000 battle sim, I'll take you up on the second part of your reply...

-=Vel=-

[couerdelion]

You can do a test with only 30 battles.

With odds of 99.5%, one defeat is statistically significant

If you do 100 battles, you need 3 defeats for a significant result.

But you can't do too many tests since the more tests you do, the more likely you are to find some spurious proof for something that does not exist.

[Krill]

*Shrug* I played a game the other night where I destroyed every AI coastal city with CR3 tanks. Every single time I attacked with the first tank I lost ,never mind if it was 75% odds or 99.8% odds. (Rifles defending in cities that have had the cultural defense bombed away by battleship are pretty crap against CR3 tanks even with the amphib attack penelty.)

Doesn't really matter, I still managed to raze every coastal citiy in about 15 turns and killed the AIs in 20.

[ColdPhoenix]

I've heard people talking about this before. Someone did a few tests as they were convinced the odds were fixed (using worldbuilder to set up a couple of hundred battles or so). To their surprise they found that the results showed that the percentages given were correct. Someone can do a search for the thread as it was on Apolyton that I read it.

[ColdPhoenix]

Just searched, think it was this thread though I haven't read the whoel way through:

http://apolyton.net/forums/showthread.php?s=&threadid=146545&perpage=30&pagenumber=1

[Arrian]

My GG useage:

1) Settle in HE city. Since the HE requires only a lvl3 unit (10xp), I typically can manage that just fighting barbs. It really shouldn't require a GG. Back when it took a lvl4 (17xp) unit, things were different.

Getting this set up early is quite powerful. Your HE city will pump out lvl2 units, which can give you a major advantage, or at the very least level the playing field in terms of unit quality (if you are non-agg facing an agg, or perhaps a prot civ).

2) Warlord attached to something strong so I can build West point.

3+) It depends. I may add them to my WP city, or I may build an academy or two... it doesn't really matter at this point.

-Arrian

[Kuciwalker]

Originally posted by couerdelion
But you can't do too many tests since the more tests you do, the more likely you are to find some spurious proof for something that does not exist.

If you do more tests you have to combine the answers of the tests. At the 95% significance level you can expect 1 out of 20 tests to give a significant result indicating that the game is rigged.

[Kuciwalker]

You can also search the game code for any rigging, but you won't find any.

[Velociryx]

Right right....at the end of the day, it doesn't really matter what you attribute it to, it is what it is, and in my experience, a Warlord unit with a 99+% chance to win doesn't tend to survive much beyond 12-15 battles, even tho the odds indicate that it should (this is an average over LOTS of games and LOTS of Warlords, and not just a one-off statistical blip).

Now, whether you attribute that to an extremely long running statistical anomoly, REALLY bad luck, a hidden game mechanic, the presence of the One Ring in the enemy camp, or elfin magic and pixie dust, it doesn't matter at the end, cos it still is what it is.

I only brought it back up because based on my experiences (whatever the ultimate cause) it has prompted me to reevaluate my Warlord usage, and I am now MUCH more inclined to go the super medic route for the Warlords I create (which is the minority of GG's I might generate in a typical game).

-=Vel=-

[Kuciwalker]

Right right....at the end of the day, it doesn't really matter what you attribute it to, it is what it is, and in my experience, a Warlord unit with a 99+% chance to win doesn't tend to survive much beyond 12-15 battles, even tho the odds indicate that it should (this is an average over LOTS of games and LOTS of Warlords, and not just a one-off statistical blip).

No it's not. It's your impression over LOTS of games and LOTS of Warlords. It's very well documented that people very selectively remember such things.

[Velociryx]

Actually, no...it's not.

It's my hand written notes about lots and lots of games.

The same notes I use when writing my strategy articles here.

I seldom rely on memory, preferring to monitor trends the old fashioned way, so one thing this isn't, is me mis-remembering, or suffering from selective amnesia, but again...it is what it is, and what it is, has prompted a change in the way I use what Warlords I create.

-=Vel=-