Free Allegiance Forums

Naning would be good to add into the stats, but... probing.. not so much...
cause i could probe the entire game, just spamming and spamming em! and all of a sudden, be at the top without a single kill!

I think you'd have to make it so you get points when a probe you drop independently spots an important enemy unit (con, miner, bomber), with the caveat that more points are awarded the further into the lifespan of the probe (so you can't spam probes in their home just to earn points, they'll get killed too quickly to be effective). Also, no points for probes that are within the scanrange of another friendly unit (such as a garrison).

Ive got pictures of Baker's monkey anyone interested?

No thanks.

Sealer wrote:QUOTE (Sealer @ Jan 23 2009, 05:47 AM) And nans on those cons/miners do not matter? How about bomb runs, they don't matter either?

There will always be "thankless" jobs, and no system will ever be able to perfectly capture them.

Get over it.

No.

TeeJ wrote:QUOTE (TeeJ @ Jan 23 2009, 04:53 PM) I actually like the more precise numbers of AllegSkill, I just hate the fact that it kind of looks to support whoring more than some of the more tedious jobs.

If you look, and read, more carefully in to the subject matter, you'll discover that the core AS ranks have nothing to do with the specific role you preform in-game. Granted, the whore rating is pretty much kill-centric, but if you read the not-so-small print, you'll notice that it has no effect on your overall rank. If there were something I could do to remedy this now... I would have done it months ago.

In fact, the very nature of the underlying mathematics creates a system that rewards fairly whichever role you perform in-game. If you're a scout whore and constantly spot enemy cons/miners/attacks, and this is consistently of benefit to your team, AS will increase your rating accordingly. As always the converse also holds. This, along with an innate support for team play, is amongst the numerous reasons we (Ok... I - long and amusing story for another day) chose to implement Trueskill over the other systems we were considering.

In case the last paragraph was too opaque: Rating systems based on in-game actions (more rank for more kill/bases bombed/asteroids crashed into etc etc) are sooooooooo difficult to implement correctly, fairly and consistently in a vastly dynamic game such Allegiance that it's really not worth the effort in trying. Furthermore, such systems are wide open to flagrant manipulation - as wise folks from the MS Zone days of Alleg stats will attest.

Anyhow... whilst I'm here I'd like to thank the community as a whole for not turning this into a massive Drama™ fest. I really was expecting more useless flak over the past few days, but contrary to my expectations this has failed to transpire. I've been in touch with Microsoft Research again. They're not only genuinely pleased that an old MSR project has met a new one, but I can also reveal that Allegiance is the first game to implement Trueskill with 'partial play' (fractional games times - joining late/leaving early). Um... GO ALLEGIANCE!

sgt_baker wrote:QUOTE (sgt_baker @ Jan 23 2009, 09:22 PM) If you're a scout whore and constantly spot enemy cons/miners/attacks, and this is consistently of benefit to your team, AS will increase your rating accordingly.

As if all the conses, miners and bombers a scout eyes got killed...

Oh, the beautiful pain of Allegiance. If you see the cons you can't kill it. If you can kill the cons you can't see it.

I've given the system more though and i'm wondering if I make sense with this possible problem when you want to use AS to balance games:

Even if we asume that AS was perfect at giving all players a correct relative rank(all rank 30>all rank 29> all rank 28... etc), I don't think that ranks(Or Mu,whatever) are an "absolute" measurement of skill.
I'd like to know if I'm right if I say that you can't take the sum of the ranks of the players on 2 teams, compare it and say that one team is better because the sum of its rank is higher.

Example:

100 vs 100.
Team 1 has 99 (0) and 1 (1). Total rank: 1
Team 2 has 98 (0) and 2 (1). Total rank: 2
The sum of the rank of the 2nd team is 2x higher than the sum of the skill of the 1st, but put them against each others, and you'll have a balanced game. This is an extreme example

To be able to say that a team is better than another because the sum of the ranks of its player is higher, a (2) must be 2x better than a (1), a (16) must be 2x better than a (8), etc. I don't think it's the kind of info that is given by AS(It says that the (16) is better than the (8), but I doubt it says that the (16) is 2x better than the (8).


Edit: I'd just to mention that I'm not questionning the system, but how we should read it.

All mathematical operations (including the forthcoming autobalance system) in AS use both mu and sigma. NOT RANK. A newbie (25/8.333...) could have any skill between (rank)0 - 50. Rank is only there to please simple minds who can't think in mu/sigma.

See the graphs on the first page of the wiki. It's presented in rather simple terms. C'mon.

Edit: Perhaps I'm being harsh here, but it's clearly stated that (rank) is only ever used for display purposes. Were you taking about manually balancing games?