First, I want to emphasize that I can't be 100% sure about all this.  I may be setting myself up for another classic beatdown by Nilbog.  That disclaimer aside, here goes
The Warrior epic proc should be an AC/ATK debuff
I was having a conversation with Vonhammer (Arteker on these forums, hopefully I can get him to post here) and he said that the warrior epic was upgraded in early Luclin.  There is considerable evidence that the original proc was an AC/ATK debuff (I mean, the spell text is 'weakened' which certainly doesn't match a +hate proc). 
http://web.archive.org/web/200208271...&pg=1&sortby=1
	Quote:
	
	
		| 
					Originally Posted by everlore
					
				 EFFECT: Rage of Vallon ( Target debuffer spell that reduces HP/AC/ATK by 100/40/11 for about a 2min period ) | 
	
 http://everquest.allakhazam.com/db/item.html?item=2630
	Quote:
	
	
		| 
					Originally Posted by allakhazam Jan 2001
					
				 Effect: Rage of Vallon  DMG 100  ATK -40  AC -35 | 
	
 
	Quote:
	
	
		| 
					Originally Posted by allakhazam Feb 2001
					
				 Rage of Vallon Decrease Current Hitpoints by 100 (L1)
 Lowers you on the target's hate list by 500
 
 as per the spdat.eff file
 | 
	
 
	Quote:
	
	
		| 
					Originally Posted by allakhazam jun 2001
					
				 The point of the proc is to raise hate since warriors had a much harder time to taunt in Kunark post 50. And you're not reading the spdat correctly, for spells that lower you on aggro it says Negative numbers, not positive. For example, Jolt/Cinder Jolt lower your aggro and the spdat says "Lower you on hate list by -500", whereas Rage of Vallon says "Lower you on hate list by 500". It's just bad wording from EqCaster. | 
	
 http://www.thesteelwarrior.org/forum...hp/t-6905.html
	Quote:
	
	
		| 
					Originally Posted by Steel Warrior 4/2004
					
				 Rage of Zek - 500 (used to have ac debuff)Rage of Vallon - 500 (used to have ac debuff)
 | 
	
 http://www.castle-grounds.co.uk/eq/epiclink.htm
	Quote:
	
	
		| 
					Originally Posted by Oldschool
					
				 The Proc is Fire based, and the effect reduces the targets AC by 12, as well as providing a magic DD of 175 damage. The AC Debuff lasts 3 minutes.
 The Blade of Stategy Proc is also Fire based, and is 100DD as well reducing a Mobs ATK by 40 and their AC by 11 for 2 minutes.
 
 The Blade of Tactics is a right clickable effect, which increases Dex by 40, ATT by 30, and AC by 12. This also lasts 2 minutes.
 | 
	
 Conclusion: The information from the last link looks by far the best to me.  -40ATK and -11AC makes far more sense then the reverse (40AC is a huge increase in damage).  My guess is that the original intent was +hate (as evidenced in the spell file) but they didn't get it implemented in time so they went with the AC debuff.
Most procs (and by extension spells in general) are generating far too much hate
I did the following test with Cucumbers and Coeur (and even Dagorn for a bit).  Our method was: Warrior 1 would attack the mob until he procced.  Warrior 2 would then attack with nonproccing weapons until the mob turned to face him.  We added up the number of primary and secondary attacks, weighted them by damage bonus+weapon damage, and any difference should be the hate value of the proc.  Our results:
	Code:
	Warrior A                        	Warrior B	APH	APD	ASH	ASD	AHATE	BPH	BPD	BSH	BSD	BHATE	Diff	Spell	     Estimate								
Epic+Star of the Guardian	Dagas+Ixi Sceptre	22	14	21	14	844	46	11	31	15	1477	633	Rage of Vallon	600								
Yak Scim+Ixi Sceptre      	GuardianStar+Epic	14	8	0	15	266	28	14	14	14	896	630	Ykesha	        575								
Veridix+Ixi Sceptre         	Dagas+Ixi Sceptre	27	13	16	15	888	56	11	26	15	1622	734	Force Shock	679								
SarnakWarhammer+Shield         	Ixi Sceptre+Epic	84	9	0	0	1680	56	15	37	14	1974	294	Stun     	250								
Malevolent RB+Ixi Sceptre	Dagas+Ixi Sceptre	16	13	9	15	519	47	11	19	15	1319	800	Froglok Poison	750								
Yak Scimitar+Shield        	Ixi Sceptre+Epic	29	8	0	0	551	27	15	31	15	1167	616	Ykesha    	575								
2x Disease Cloud            	Dagas+Ixi Sceptre	0	0	0	0	0	39	11	18	15	1128	1128	Disease Cloud	550								
2x Shadow Vortex           	Dagas+Ixi Sceptre	0	0	0	0	0	40	11	16	15	1120	1120	Shadow Vortex	550								
SarnakWarhammer+Shield         	Ixi Sceptre+Epic	10	9	0	0	200	19	15	13	14	676	476	Stun         	400
 So for example on the first row I had 22 primary hits and 21 secondary hits; Cucumbers had 46 primary hits and 31 secondary hits.  The difference in hate was about 600.  These values in general seem substantially higher than what is reported on the forums:
http://www.thesteelwarrior.org/forum...hp/t-6905.html
	Quote:
	
	
		| 
					Originally Posted by Steel Warrior 4/2004
					
				 The relative amount of hate that stuns generate was derived from many hours of testing stun agro from pally stun spells (the unknown variable) to SK hate spells (the known constant). I forget who did the testing now, but the evidence was conclusive and can be easily reproduced. They where able to come up with ~450 hate for a stun, which also seems to match the "feel" of a stun proc. | 
	
 
	Quote:
	
	
		| 
					Originally Posted by Steel Warrior 4/2004
					
				 ac debuffs should also be added around 400 | 
	
 http://www.thesteelwarrior.org/forum...?t=1157&page=2
	Quote:
	
	
		| 
					Originally Posted by Steel Warrior 6/2003
					
				 Slow is second only to Stun for hate spells (that aren't pure hate) | 
	
 http://www.thesteelwarrior.org/forum...ead.php?t=2057
	Quote:
	
	
		| 
					Originally Posted by Steel Warrior 7/2003
					
				 I would say, from best(top) to worst(bottom)
 - Hate (EB, TB, etc)
 - Stun
 - Slow
 - Debuff (-AC, -STR etc)
 - Snare (Snare or darkness line)
 - Blind
 - Lifetap
 - DD
 - DoT
 | 
	
 Conclusion: Based on this I think we should see stun=450, poison counter = 150 per, debuffs 400.  
Right now Ykesha stuns seem to be 500, while pure stuns are 400.  Poison counters seem to be 225-250 which is extremely high.  And debuffs are 550 which again seems much too high.  Enchanters and shaman may actually be able to live after slowing something!
Major HT to cucumbers for helping me dig up some links and run the tests.