I know everyone complains about luck and that when people do there is a general groaning from the chorus followed by sarcastic mockery, but I'm going to put myself out there. I'm very happy that the luck graph is available because more often then not my loses are directly attributed to my rolls and not my strategy.

That being said my luck is consistently well below the margin. My record is below -53 and (I'll have to do fact checking) but my luck seems to usually be well below the norm. It really feels sometimes that the gods of wargear are out to get me.

Is there something I should know about how the rolls are "randomized." Are there tricks or things I can do to improve my luck. It just seems unnatural that my luck would be this bad all of the time. Could there be a computer glitch somewhere?

I get a kick out of this.  We humans are just really bad judges of our own luck.  LOC, I just checked your last 15 games (choosing "finished" in the games list). You ended up with a positive luck stat in 7 of them.  With a total of 4058 dice thrown in those 15 games, you lost a whopping 2.96 more armies (net) than expected.  That's about a statistically as even as it gets.

And the weird thing about luck is luck within luck.  You can end up with neutral luck but if you are unlucky the first two rounds you lose the advantage, and then the game, even when you get lucky later in the game.  Or you can get unlucky when your opponent gets unlucky and there is no bad effect.  Or sometimes your bad luck with placement or opponents moves make you make attacks your are not favored to win, and you don't, but its not unlucky in the stats. etc etc 

There's so much other luck involved in the game too - starting position, turn order, getting 2 pair (or having a set with 3).  And as Squint Gnome alluded too, there is luck within luck on the luck graph.   One fortunate roll of the dice could wipe out a huge bonus for someone, but would show up as a small tick (if even that) on the graph.

Sometimes you get in one of those ruts where you have bad rolls in every game it seems for a few days.  It's easy to forget the times when the dice were going good for you.

M57 wrote:

I get a kick out of this.  We humans are just really bad judges of our own luck.  LOC, I just checked your last 15 games (choosing "finished" in the games list). You ended up with a positive luck stat in 7 of them.  With a total of 4058 dice thrown in those 15 games, you lost a whopping 2.96 more armies (net) than expected.  That's about a statistically as even as it gets.

M57, you've proven my point about "groans from the chorus and sarcastic mockery." Thank you for doing the research I should have done previous to posting and also for proving my point. Now I'll get back to you on the total mean of all my games played, rather than just 15 recent. 

I'd like to point out that in this instance I've been referring to the luck I've been having lately in unfinished games not recently finished. And if all things considered we discover that I'm bitching for no reason than so be it. But as always you've contributed nothing to this thread except for bitching about my bitching. So thanks for nothing.

Excellent points by both Squint Gnome and Cona Chris. Thank you.

I guess my real question should be how are rolls arrived at? And not to second guess the programmers but as a sincere point of curiosity. I'm a player and not a programmer so this information piques my interest.

Also I don't mind hearing other people complain about their luck, whether warranted or not. It's an integral part of the game (although good strategy can counter-act it) and sometimes it just helps to vent.

Lord Of Cups wrote:

But as always you've contributed nothing to this thread except for bitching about my bitching. 

"As always" (whatever that was supposed to mean), I've contributed a number of things to the thread, including doing research that you were seemingly unwilling to do. 

So thanks for nothing.

Thank you for doing the research I should have done previous to posting and also for proving my point. 

So which is it?

Lord Of Cups wrote:

M57, you've proven my point about "groans from the chorus and sarcastic mockery."

Hmm.. I must have missed that one.  Perhaps it is because I'm having trouble understanding the metaphor.

LOC, sorry if my initial comments seemed insensitive (they certainly are now), but I have a pretty high confidence in whatever seed and algorithm tom has chosen to roll the dice, and I took your initial post at face value.

I'm much more interested in finding a way for us to put our luck scores in perspective.  I.e,, just how unlucky is a LS of -50 in a game where you had 1000 rolls? How does this compare to the player who has accumulated a luck stat of -5000 over 100,000 rolls. There have been a number of threads devoted to this subject and the mathematicians on the site have stated that the number of calculations involved in computing the standard deviation (which is generally accepted as a good solution) are too processor consuming.  I've offered a very basic formula for estimating something close to it, and suggested that there might be a kind of backdoor way of calculating it, but until we come up with an agreed method of analyzing Luck Stats, these threads are destined to be not much more than, as LOC says "bitching about bitching."

Lord Of Cups wrote:

I guess my real question should be how are rolls arrived at? And not to second guess the programmers but as a sincere point of curiosity. I'm a player and not a programmer so this information piques my interest.

Also I don't mind hearing other people complain about their luck, whether warranted or not. It's an integral part of the game (although good strategy can counter-act it) and sometimes it just helps to vent.

What do you mean by "how rolls are arrived at"?  Did you mean how are the dice results determined or how is the luck graph calculated?

I think you mean the former, and while I am not a programmer either, I imagine it's a random number generator - so for 6-sided dice, there is a 1/6 chance of it being each of 1, 2, 3, 4, 5 or 6.

Luck is definitely a factor - more so on smaller boards or boards with lots of players. 

I've lost a couple games (but also won some) due to insane luck results - so I definitely know how you feel when the dice are just punishing you.

Mad Bomber wrote:

ther should be a first 10 move luck stat

MB, on the chart, you can set the slider to see where your stat was after ten moves.  And honestly, I don't see that such a stat would be that helpful.  Sometimes it seems that the player who has the best luck stats in the first ten moves is summarily targeted by other players.  I've been on both sides of that equation.   I'll bet many of us here have been in multi-player games where the player with the worst luck stats wins the game. Accurate analysis of LS often requires deeper analysis of not only "sub-luck", but often the workings of game theory.  How many times have we heard the better players mention that it is often good strategy to not have the lead.  I feel that recently my game (and resulting ratings) has deteriorated for this very reason.

Being one of the non-mathematicians on this site I'll offer my two cents on how I deal with luck. 

Firstly, I believe that if my strategy is effective and constant (i.e. consistently good) and my luck is variable, then the variation in my luck should even out over time and my results will basically end up as a function of my strategy vs. a function of my luck.

Secondly, if over time my luck doesn't seem to be evening out and reducing to my strategy, I change my strategy to one that doesn't depend so much on luck (i.e. make attacks that are less luck-dependent).  It's hard to say how many times I've had to do this, but I'm guessing not many.

Thirdly, if my luck is evening out and reducing to my strategy, but my strategy isn't getting me the results I want, then again I shift my strategy.  This happens more often and takes the most creativity.

Overall I feel it comes to these two points:

1) If my style of play is walking the fine line of luck to where the luck makes or breaks my result, then there is probably a better strategy for me to play with.

2) Hmmm...what was that second point? Oh, yeah: If my style of play is NOT walking the fine line of luck and I'm still not getting the results I want, then there is probably a better strategy for me to play with.

Since in both cases it always comes back to my strategy, I throw up my hands and get down to work on thinking about, researching, and testing my next best strategy.  I believe that 1 minute spent reviewing a top player's finished games is worth 1000 minutes studying the luck stats.

~ATH

I minored in statistics, which is not to say I'm any great authority about luck, but I can tell you a couple of things:

1) Probability and outcome have zero to do with each other.  If you plan to flip a coin 1000 times, the probability it will come up heads every time is ridiculously low.  However, if you've flipped a coin 1000 times and it came up heads every time, the probability it will come up heads the 1001th time is 50%.  And as small the probability is that a coin will come up heads a million more times, there's absolutely nothing to prevent it from doing so.

2) No pseudo-random dice generator in existence is either truly fair or truly random.  I'll say it again:  Every dice generator ever devised is neither fair nor random.  Every one produces skewed and/or predictable results.  This is caused either by flawed algorithm design, confines of the generating mechanism, or a host of other arcane reasons.  (The only potential exception I've ever seen was a generator based on a physical reading of radioactive decay.  Unfortuantely, I'm pretty sure Tom doesn't have a geiger counter and a lump of Caesium-137 laying around.)

That said, I don't know what your to-date luck average is (I don't even know what mine is because Tom won't tell me), but I'll bet you $100 I've got a luck average far, far below yours.

I'll also tell you a couple of things I've noticed about the luck generator used here:

1) If your luck starts out bad, it generally stays bad throughout the game, more often than not, if not even get worse.

2) If your luck starts out good, there is no similar expectation that it will stay good throughout the game.

3) Believe it or not, your emotional state while playing has an effect on your luck.  See http://en.wikipedia.org/wiki/Poker

4) Also believe it or not, sometimes bad luck appears to be pre-determined, right at the worst possible time.  See http://en.wikipedia.org/wiki/Murphy%27s_law

5) WRT to #3 and #4, you cannot fool the computer.  See http://en.wikipedia.org/wiki/Thiotimoline

6) Almost invariably, attacks between small stacks tends to favor the defender, but attacks between large stacks (50+) frequently favors the attacker.

Finally, to quote someone smarter than I, but whom I forget, "you make your own luck."

csecu wrote:

(The only potential exception I've ever seen was a generator based on a physical reading of radioactive decay.  Unfortuantely, I'm pretty sure Tom doesn't have a geiger counter and a lump of Caesium-137 laying around.)

 

I thought there was a RNG based on the Atmospheric Noise that was pretty random...http://www.random.org/

A while back tom said he was using PHP's "Mersenne Twister" implementation. I found an interesting site comparing random.org's to PHP's old rand() function. The Mersenne Twister is mt_rand(), so hopefully that is what is being used. Anywhere, here is a nice visual comparison of bad pseudorandom (PHP's rand) versus physical random (random.org):

http://www.boallen.com/random-numbers.html

Note, however, that the developer could detect no such patterns with mt_rand(). The mersenne twister is supposed to be a significant improvement over older PRNG's, but is still not random.

I like this random # generator based off a lava lamp:

http://www.lavarnd.org/what/index.html

edit:  Looks like that place no longer uses actual lava lamps.  They used to, but now they just basically use a camera with the lens cap on, and then use the resulting noise from that image to see their random generation.

Before we go down this path, is the difference even noticeable between a true random and a very good pseudo-random?  As in, are we arguing between whether it's 100% or 99.99999999999% random?

AttilaTheHun wrote:

Before we go down this path, is the difference even noticeable between a true random and a very good pseudo-random?  As in, are we arguing between whether it's 100% or 99.99999999999% random?

The problem isn't how "close to random" a pseudo-random generator is.  It doesn't work that way.

The problem is two-fold: skewness and predictability.

Skewness can manifest itself in a number of ways.  For instance, if the numbers generated over an (theoretically) infinite data set demonstrate any sort of bell curve or if they are completely flat across the spectrum of numbers to be generated.  (In practice, computer-generated random numbers are generally only a fraction that falls between 0 and 1; all larger numbers are a product of that fraction.)  In most cases, from a distance it looks flat, but if you look reeeeeeeaaaalllly close, it's actually a fractal, with many, many peaks and valleys.

Another, more common artifact of skew is that certain numbers may never be generated.  For instance, out of a set of 1000, the number 623 might never appear.  So, the generator is "99.9% random," but if you're number 623 in line, pardon my French, but you're ass-out.

Predictability is the other serious problem.  By definition, no random data set can ever be duplicated.  Conversely, computer-generated psuedo-random data sets can always be replicated.  It might take some serious programming gurus and supercomputer horsepower to replicate, if you're a third-party, or it might just take resetting the system clock to the same seed point on a pocket calculator.  Predictability allows, however hard it might be, someone to game or break the system.

So, it doesn't matter how "close" to random any algorithm-and-machine combination is.  It's still not truly random, and never will be.

Hugh wrote:

A while back tom said he was using PHP's "Mersenne Twister" implementation. I found an interesting site comparing random.org's to PHP's old rand() function. The Mersenne Twister is mt_rand(), so hopefully that is what is being used. Anywhere, here is a nice visual comparison of bad pseudorandom (PHP's rand) versus physical random (random.org):

http://www.boallen.com/random-numbers.html

Note, however, that the developer could detect no such patterns with mt_rand(). The mersenne twister is supposed to be a significant improvement over older PRNG's, but is still not random.

Yes, mt_rand() is used by all randomization functions in the game engine - includes dice throws, seat assignment, capital assimilation etc.

I recently heard an article on NPR about a study done regarding human interpretation of luck and risk. They did a study using a fairly basic concept. 

I'll flip a coin and if it comes up heads I give you $5 and if it comes up tails you give me $1.

Most people will take this bet.

When they increase the tails price to $3 there is a drastic drop off of people willing to take the deal. Even if the are offered to do it 100 times which should earn them statistically $200 even if they lose 50% of the time.

I think that I got this about right. It was interesting because they believed it is a genetic trait in humans to weigh minor risk much more heavily than probable gains in deciding to do something because of in our early days it wasn't money but living or not.  I wonder if this comes through in our interpretation of our own luck as well.

I've really never been able to understand the luck stats but that is ok since I am convinced I am really bad at playing here especially if it is one of M57s attempts to see just how complicated a game can I create with the new factories.. ie graffiti. 

Thread Hijack GO!