We're gonna die!

[link120]

thats actually same for me, *karnagez quote*

[raumvogel]

Yes, anything that can be done to promote the game is appreciated. We love this game and want to see it grow in player base. That's why most of the vets will help you. We have a few bad apples of coarse, but we hope the newbs will be mature enough to ignore them. Feel free to PM the more frequent posters including myself if you need any help.
I think it acually help to have an occasional "We're gonna die" thread as it makes the vets panic and play the game more often.

[GoodWill]

QUOTE (raumvogel @ Jul 16 2011, 09:42 PM) Feel free to PM the more frequent posters including myself if you need any help.

Prepare to be boarded! ... Just kidding.

QUOTE (raumvogel @ Jul 16 2011, 09:42 PM) I think it acually help to have an occasional "We're gonna die" thread as it makes the vets panic and play the game more often.

True. On the other hand, from being a "vet" in other games, I know that it sucks
when games start to feel like an obligation. Instead of playing the game and
having fun, you find yourself wasting your free time on having to work through
forum-posts, having to instruct an army of noobs about whatever, having to be
the guy who pulls other peoples weight to keep things going ...
It just doesnt really work out to be the "social worker" or "the guy who takes
care of everything" when originally you started for a simple reason: to have fun!

[Axel Kolle]

I second Goodwill on the generally positive reception.

When something goes right, it gets noticed as much as you get a "read the chat" or a rocket if you don't get it. Thanks for noticing, guys. You know who you are. Just understand that luck plays a part.

I also noticed that commanders seem to elaborate a bit when giving orders. I appreciate your taking the extra few seconds. And being the one to find the miners the other day was what a good dogfight is to an Int driver. A rush.


Apart from that: This game has changed, and will change more, as circumstances and the general gaming culture change. For one other games create a different kind of expectation of what is a "good game". Upholding the standards of the past is only part of the difficulty, a certain measure of tolerance for what the game is now is necessary to avoid disillusionment.

I don't know any better, so in a way I am at an advantage. But being a niche game is not necessarily a bad thing. As long as one embraces the choice. And the inevitable consequences.

[Zero_Falcon]

"Dated graphics"...

I'll say, atm Neo Dreghklar, Neo TF and Phoenix have models equal of the present time, but it's the rest of the factions that need remodelling, especially IC, their ships look really old-style Other factions: GT...

I wouldn't mind having the original MS ships being remodelled, seeing the circumstances right now maybe we should take steps to modernising alleg even more, this being one of them. I'm getting motivated right now to try my hand at their new models, even!

[Lykourgos]

the graphics still look fine to me after a 3 year hiatus

i'm pretty sure it's the way joining a game works, combined with the below-critical-mass existing user base, combined with the super steep learning curve

as it always has been

[Spunkmeyer]

QUOTE (Lykourgos @ Jul 17 2011, 10:15 AM) the graphics still look fine to me after a 3 year hiatus

i'm pretty sure it's the way joining a game works, combined with the below-critical-mass existing user base, combined with the super steep learning curve

as it always has been

QFT, pretty much.

[raumvogel]

I found this on CNET:

The old D/L for M$ Allegiance circa June 2001

Unless someone can tell me why I shouldn't, I'm going to try and get them to host the new and improved version.

[Bard]

QUOTE (raumvogel @ Jul 17 2011, 11:46 AM) I found this on CNET:

The old D/L for M$ Allegiance circa June 2001

Unless someone can tell me why I shouldn't, I'm going to try and get them to host the new and improved version.

As long as someone remembers to contact them again when we replace ASGS...

[MrChaos]

QUOTE (Spunkmeyer @ Jul 13 2011, 06:57 PM) (*sigh*... $#@!, ok gonna trim the mini-novella I just wrote)

None taken. You think what I said conflicts with what you've found but I don't believe it does: I'm not questioning AS team totals are predictive of the outcome while disregarding the commander's influence (or the accuracy of your analysis at arriving at the conclusion). I'm more than happy take your word for it.

But while there could be a number of reasons why that holds true in the current matchmaking environment, it won't necessarily hold true in a new, more structured (and hopefully fairer and less painful) matchmaking environment. We won't know until there is enough historical data, and my "gut-feeling" is that ignoring the commander influence especially in a stacking-free environment is not going to help.

Or to put it another way: we know AS, or rather TS, works perfectly in games where there no external variables beyond player skill influencing the game. Therefore it's logical to either eliminate them or take them into account. I realize eliminating them is more in-line with TS/AS philosophy and any additional fudging has an arbitrary factor, but the commander is not something you can eliminate (others such as game settings favoring one team could be eliminated, potentially)

Cyre: you can measure skill. There will of course be some arbitrary weighting to it - the question is if we can get it close enough to result in an agreeable number. We won't know till we try and model the system.

There is also a huge advantage in providing statistics to highlight good stealth bombers, TP droppers, proxers etc for gameplay purposes.

There are a 100,000 plus games to work with iirc. The amount of data isn't the issue.
We did prune the data with an eye to bad records, short game resigns, newbies vs newbies, *insert squad here* vs world, Jimmyesque affairs.
Correctly predicting outcome of the game when using player ranks was(Baker is going to shoot me cause I always get it wrong) mid/high 80s
Using the commander ranks AND well balanced teams (it has been years I can't even begin to give the definition) showed nadah. The players not the commander decided the outcome.

Nope you don't have to throw the last six years of data to get accurate ranks and compare commanders. Why? To put it simply and save a novella back it isn't an issue given the prediction level being so high. Cleaner data could give you a better base to work with but the idea of letting the gen pop lose to modify their behavior to $#@! with ranks meant we quickly discarded this notion. Before the trolls troll (not you Spunk) what this means it will become even more accurate. Could there be a spanner in the works fubarring things enough to change the outcome of the commander matters great debate but it would be interesting to compare consider me onboard

AS/TS works well in complex games where simple points based ranks predicting falls apart... like say... Allegiance. It is EXACTLY this kind of game where players come and go, there is no matchmaking, and the roles are so diverse where it shines. I'm positive on this next one. Points based systems where one receives them for in game interactions effects game balance on a micro level and highly inaccurate compared to the approach taken by MSR. So is MSR btw and part of the reason they went with TS is due to this very game and the whole "Microsoft Zone" experience (@#(! did I remember the name right?). Halo/XBox was the money engine that finally got it done for XBo Live. If you want to take a stab at the points approach you first have to develop a list of game interacts that matter. You are linearizing the game experience to produce a single value rank for the players. Each inteaction must be consider with every other in all combinations along with the everything else we didn't look at factor.


Before the next bit: I apologize if I get the equation layout wrong it has been a bit since I done this stuff by hand and Im already expending way to much energy... again

Even just ten items (kills, not kills in an int vs a scout but just kills counts as an item) right off the leaderboard causes something of CPU crushing size
a b c d e f g h i j ab ac ad ae af ag ah ai bc bd... abc abe abf... abcd abce.... abcde+abcdefghij + stuff

Even this is cutting corners since you didn't look at ab ba abc acb etc etc. not all combos and permutations in a nutshell

after that

and if stuff (everything you didn't consider incluing the whole ba and ab thingie) is large enough i.e. you didn't guess right you go directly to jail, do not pass Go, do not collect two hundred dollars and start over. Plus you need to check its accuracy by spinning the points about the line you created through the data cloud (cause thats all your really doing linearizing things and creating a line), after weighting each point and then check the line's wobble. This is a compound error factor across the entire player base and calculations.

Oh and this approach absolutely requires that the underlying data be normalized which $#@!s with both the data and the alogorithm's accuracy, a pretty significant task, fraught with error filled danger, which you have to check each and every time... well on a regular basis at least, all so you can have the equivalent of this:

lnY=mlnX+b+stuff since the equation isn't afine

You could try the fuzzy logicapproach but its basically the above but you guess at the starting values and then shove massive amounts of data through it (it is dependent on the number of nodes and I cannot recall the details) and hope it works well. Btw

QUOTE Fuzzy logic began with the 1965 proposal of fuzzy set theory by Lotfi Zadeh.[2][3] Though fuzzy logic has been applied to many fields, from control theory to artificial intelligence, it still remains controversial among most statisticians, who prefer Bayesian logic, and some control engineers, who prefer traditional two-valued logic[/quote] It works well when you want, well fuzzy answers (too warm or too cool) but usually like utter @#(! when you want discrete answers ( player X is better then player Y by this much) unless the system is so simple to make the exercise trivial. This is not a trivial undertaking btw

Its a herculean effort either way and highly dependent on having someone with the expertise to maintain the beast to do something akin to the GMMR. Basically done for the sake of doing it, finite, and a maintaince nightmare all to grow food in the desert rather then buying it.


I absolutely agree we should have stats and achievements up the ying yang. Display them, send them t-shirts, anywhere and everywhere and make more of them while giving them public recognition too. It is the only thing that is sustained from game to game with Allegiance and the only way the average player gets the Skinner Box experince. Ranks on the other hand are meant to rank you for match play and should never consider anything else i.e. ranks = $#@! their feelings -> your ugly Stats = makes them feel better -> you have a wonderful voice and big ol cans.

Btw one can measure "skill" using a Baysian... *sigh* never mind, never mind.


tl;$#@! the novella give it to me short and sweet
No one eliminated the commander just gave no special value to his role. We did our best to check and predictive outcomes support this approach
The discussion on balanced game play vs non and the extra influence of the commander is an interesting one, let me know when we have the data to do it
Points based systems are highly arbitary, inherent balance issues, gigantic amounts work to implement AND maintain, and also open to blatant abuse issues
Statistics are cool we need them, achievements too they make epeen bigger but are not the same thing or play the same role as a ranking system
edit: added if Im not mistaken the ten factors discussion is 10! for all perms and combos or 362,880 nodes... that just can't be right