Explain this further. Because the OP was referencing Pantheon statistics, in which the AI would only take over for player's whom left the game early. Said statistics can be found here. As the OP said, there is no objective data to support Erebus being nerfed. Erebus does not stand out in any category.

For the record I play primarily as Rook for the Forces of Light and Queen of Thorns for the Forces of Darkness, and I am very confused as to why so many people are fixated on demigod balance, and further, why the development team would have such a knee-jerk (and extreme) reaction to subjective and unsupported claims on these forums.

erb nerf was too much, only 1-2 of the nerfs were needed

and i dont play as erb

None that 'we' have.  I certainly hope GPG isn't simply looking at win ratios on Pantheon when making balance changes, though.

Pantheon statistics do not tell you which DGs are the most efficient. If they did, then it would mean Hunters in World of Warcraft were kings of everything for the first two years and Shaman never had any advantage in the early days of PvP, but it's widely accepted that Hunters were over-used by gold-farmers and Shaman was a problem until the autumn of 2005. The context is important.

If people say Erebus is overpowered, certain players will be encouraged to use him whilst a minority of macho competitive types might consider it completely unsporting to use him at all because of the perception. It skews the statistics: a good player might win more matches with Erebus, but for each of them there are many more average and below average players that are not making the most of his advantages and bringing the numbers down. The statistics as they are can't tell us anything unless they are broken down more. There are people that win more games on average and people who lose more games on average and there are players that always use Erebus and people that casually drift between DGs depending on their mood. Without knowing anything about the kind of players that use Erebus, we can't draw any conclusions from just the numbers.

Our most accurate metric at the moment is reasonable projections, tested and found to be true. Things like his ability to permanately stay in Mist form if a player invests towards it, his movement advantage given by Bat Swarm over impassable terrain and the amount and extent of the utility co-efficients stacked on his primary damage ability. We know he can do the first two; feats not as easily replicable by other DGs without significant trade-offs and the third is a more complicated matter but seems such a 'must have' that it's drawn suspicion.

You are misrepresenting statistics here, these things DO NOT MATTER, the law of large numbers holds that there will always be regression towards the mean, the stats hold unless they have been tampered with or artificially generated, I don’t believe either possibility to be the case.

To complete the thought you are trying to express, you have to ask "which mean?" If better players tend not to play Erebus (as the person you're responding to conjectures) then regression will indeed be towards the mean, but the mean will be the mean score earned by underskilled players. Erebus will appear to be worse off than he actually is from a balance standpoint.

Pantheon stats don't matter, they're probably broken, and for all we know, it might even include AI in it. I'm pretty sure AI is still in pantheon because I just played a pantheon match with 2v2 with 1 team having an AI. I also got a 2v1 before that.  Not to mention the abundance of bad players who probably decided to play him after hearing he's overpowered.

No stats to back the nerf up? Bite is an ability with a low cost, low recharge, no cast time that did 300 life drain + snare + armor debuff at level 1 allowing your next few attacks to hit for more damage. 600 points of damage+heal+2 other effects, does way more than other level 1 skills of other demigods.

 Erebus is already powerful enough even after the nerf, as he has 2 ways to escape otherwise certain death, and an effective stun.

As you noted, this is pure conjecture; a normative distribution of both good and bad players should be inferred.

Regression trends towards the population mean of the group in sum, i.e. we have some good players, some bad ones, some that play these characters, some that play others, but given enough data (law of large numbers) the result will eventually reach the true mean, your conjecture therefore does not hold, it require a gross discrepancy between poor and competent players to make any sort of impact, and you can not assume this as it is highly unlikely. At current Erebus's stats sit at a 52% win rate after 32,199 games, I suppose you could argue that this is a bad sample, but that would only be because more data is needed.

/agreed

This is why a competitive scene would go far in determining game balance, but we are not there yet.

Seriously one nerf and everyone busts out the kleenex and rage hats.

Imo conserve your rage like Esbon said ... wait till some others get some nerfs/buffs. So much crying incoming ...

Personal experiences represent too small of a sample to support a claim that one Demigod is played by the AI more than another. As said by BeardKing:

In this case, we can assume that the distribution of games played fully or partially by AI will be roughly the same for each demigod.

Back to JagerJack:

Ouchy Dathurts made an excellent response here, and I have one to add. The point here is that balancing changes should be justified by objective information, not by expert (or inexpert) subjective opinions. The OP shouldnt have to bring up statistics at the 11th hour, they should have been the foundation of this debate from the beginning!

Moving on: Busdude says

I edited the first paragraph of your post with underlines to bring attention to how many assumptions you are making here, and that you did not give us any support for them. In the second paragraph I italicised sections and words that you need to support. For example, define "low", lower than other abilites? which abilites? "way more than other level 1 skills", how much more is "way more"? which skills? how do they relate to bite? It is your responsibility to inform these claims, and please be less vague in the future.

Finally: from Gorgeras

"The context is important" applies to your own post as well, in which you compare an MMORPG that features PvE and PvP with a persistant world; to a non-persistant, deathmatch style game that features no PvE play. Class distribution in WoW reflects more than just players interested in PvP, at least that is what i'll assume from the link you gave supporting your arguement (there was none).

edit: formatting issues

honestly, I'm a bit shocked they nerfed him already. Has not even been a month