Grim_Reaper_4u wrote:QUOTE (Grim_Reaper_4u @ Jan 17 2008, 07:16 PM) Eh that seems to be your opinion and not scientific fact Baker /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" /> a win/loss system is a lot easier to implement though and of course a good points based system might not be possible with current logging technology /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />
Give me *one* way of accurately scoring situational awareness and I'm all ears.
QUOTE a) the more people on a team the more games you need to accurately judge a persons skill, peeps that only play 20 vs 20 will never be ranked accurately before 2015 /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" />[/quote]
That's supposition and exaggeration. I asked you to read up on the facts. If you had you'd know that accurate ranking depends not only on how many games you've played, but also *who you play against*. Opponents with low sigmas lead to bigger decreases in one's own sigma.
QUOTE b) all the theory quoted in your articles fails to take into account anti-stackers : peeps with great skills that don't mind losing and join the underdog team that is almost dead, even though AB says teams are even (cause it defies logic /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" /> )[/quote]
I'd like you to expand upon this statement. I've explained this to you once already (I think). /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />
QUOTE c) despite your claims to the contrary I think that the new Trueskill was only tested between teams of similarly ranked players (in which it is much easier to identify players with higher skills, esp. with random team assignments)[/quote]
OK. I'll email Ralph Herbrich @ MSR and explicitly ask this question. Will that settle this? (Q: How does testing against the Alleg database constitute 'similarly ranked players'?)
QUOTE d) preferably the teams should be random every time when calculating those stats (to prevent for example a skill-less noob to always join the same team as Aarm+Snack+Frag and thus freeride the skill-ladder on their backs)[/quote]
The first part of this statement doesn't make sense. In retort to the second part: It is well documented that Trueskill, or any other win/loss system for that matter, is unable to differentiate between players who *always* play on the same team. There is not one instance of this in the whole of the Alleg DB.
QUOTE trueskill in combo with autobalance will never be better than the "eye" of a seasoned vet where balance is concerned, however it is probably the next best thing /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" /> but ..........[/quote]
See my point regarding human aptitude in parallel estimation.
QUOTE Regarding a points based system : IMHO if we balanced out int whores and miner killers over both teams then 95% of allegiance games would be balanced. In most games the balance is ruined by several whores joining 1 team and having a few good miner killers (who aren't ints whores too) on 1 side also ruins game balance often /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" /> If your AB would divide whores and miner killers evenly over both teams then most games would be even (esp. since all "follow the int hos" stackers could now chose between 2 teams /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" /> )[/quote]
"Eh that seems to be your opinion and not scientific fact Baker" <-> "IMHO" Surely some mistake? /wink.gif" style="vertical-align:middle" emoid=";)" border="0" alt="wink.gif" />
