Quick Notes on Blackrock Mountain Wing 4: Blackwing Lair

I’ve been noting down my experience with, and strategy for, the Hearthstone Adventure encounters each week on the EJ forum.  I just realized there’s no reason not to put them here for people working on the fights now.  I know they’re mostly over, but I’ll put in this week’s anyway, and maybe paste in the old ones sometime for people trying to do them later.

Descriptions of the bosses and their decks can be found here.


Razorgore: Probably lots of easy ways to do this.  I went with typical Freeze Mage-ish board clears.  His only value advantage over you is creating 4-5 HP worth of stuff to beat down every turn, which is not hard to overcome just by establishing board control in conventional ways.  An early Doomsayer and a late Flamestrike help you lock it down though.  He Corruptions big threats, so typical midrange creatures to own the board work great.

Continue reading

Quick notes on Int/Spirit trinket changes

Quick notes for now since I’m at PAX East. I’m I’ll give this more discussion going forward, probably on the next podcast (hopefully soon) especially.

As many healers have noticed, in Patch 6.1, all non-Spirit items were removed from the healer classes’ lootspecs.  So healers can’t get them from any source that respects lootspec (bonus rolls, personal loot in dungeons or raids, LFR, mission tokens, challenge mode rewards, and so on).  This seems like an odd choice since Int vs. Spirit, especially on trinkets, was generally considered an interesting loot choice and there was a lot of variation in people’s views on it.

To briefly summarize why it happened: tanks have a similar situation with Bonus Armor, and for them there isn’t as much choice.  Bonus Armor is unequivocally better for tanks, and so this change was needed in their case–it prevent them from getting non-Armor items they’d never use.  But for healers, Spirit is not always unequivocally far better than Int; in fact many healers have situations where they choose Int on trinkets.  So this change has the odd effect of removing an interesting loot choice.  Blizzard’s stated reason is that they want “healer” and “DPS” trinkets to be clearly delineated, so there aren’t items with overly broad competition in a raid, which is frustrating.  That makes sense, but overall I don’t think the 6.1 solution is a good way of addressing it.

There are some mana balance issues unrelated to all this.  Disc has a super undercosted spell so they have excess mana (separate class toolkit issue). Paladin doesn’t have heals with enough marginal gain over Holy Light casts (separate class toolkit issue).  For now, let’s take a class with a very well-working mana game, and a good spellcasting options which are influenced by mana constraints (Shaman or Druid).  Right now, the weird thing is, things should be totally fine for these classes.  Int and Spirit (on trinkets) are reasonably well-balanced options.  That’s a hard thing to do and it’s done surprisingly well.  There are good cases for both stats and it can vary with content/encounter.  Furthermore, they removed the choice from most slots (4 where you always want Spirit and many where you never do) to prevent being able to swing it too much, but left the choice in 2 important slots (trinkets).

Everything about this, so far, is so good and healthy that it’s a little bizarre that they feel a need to stamp it out.  We’re talking about removing actual, good, gameplay in the form of meaningful stat choices (one of the most prominent places such a thing remains).  The biggest problem with the solution is that it seems to ignore that the class/stat design doesn’t support this notion of Spirit being unequivocally far better in every situation.  That’s not even close to true, even for a class like Druid (even if Spirit is often slightly better in many situations).  That might take an expansion-level change, again.

And furthermore, they’d have to massively buff Spirit in a way that didn’t accelerate mana flooding (which is okay in Warlords, but touchy, as I got into here: Healing Theory: Warlords Spirit Update | It’s Dangerous to Go Alone ).  That’s okay–for example, the recent Monk change was careful about this.  Watcher’s suggestion in a recent interview of making Spirit also give SP to healers would do it for all healers.  But totally reworking the function of the stat seems like next-expansion territory.

If they do make an big fundamental change, wouldn’t it better to still have the option of regen stat vs. throughput stat–an option that is already designed, implemented, and well-balanced (rather than mashing them into one stat)?  It would be perfectly fine to have Spirit and Heal Power trinkets, both of which were not shared with DPS.  Or simply go the Mists route of having some trinkets that only proc on damaging events and some that only proc on heal events.  There are various potential ways to have clearly-defined “DPS trinkets”, “Healer Spirit trinkets” and “Healer throughput trinkets.”

In short, this change feels a little reactionary in response to players exercising a choice of stats (which is of course a good thing).  Not only does it sacrifice a solid, if minor, bit of decisionmaking, but it doesn’t even solve the problem.  Healers will still want exactly the same items they wanted before, but now, in certain cases, it will be much more annoying to get them.

Quick Notes on Efficient use of Bloodlust

You probably hear this all the time for various encounters, especially from DPS: “Bloodlust at the start of the fight, because everyone’s standing still and [some mechanic] isn’t happening yet, so you get the best DPS value out of it.”  This post is slightly tricky because I want to talk about 1) a subtle fallacy here (that Bloodlust is inherently better at times when raid DPS is higher), while also noting that, 2) in the end, it often is correct to use it at the start of the encounter (because it will have the strongest intersection with everyone’s cooldowns and trinkets).

There are two reasons that (1) can be incorrect.  First, it can be more important to get through a hard phase of the fight faster than to end the whole fight more quickly.  This is important but I’m not going to get into it too much here.  It’s generally well-recognized and, when Bloodlust isn’t used at the start, this is typically the reason.  What’s more interesting and counterintuitive though is that using Bloodlust when the raid’s doing the most DPS isn’t even necessarily best for ending the fight quickly.

How is that possible?  It’s (roughly speaking) a uniform %-based DPS increase.  It’s most valuable when the underlying damage is the highest! (You might be saying).

Basic Example

Say you have a 2-phase encounter.  Phase 1 goes until 50%.  It does nothing–it’s a target dummy and the raid does its maximum DPS (our hypothetical encounter designer is very amenable to setting up weird examples for me).  Phase 2 has lots of abilities that interfere with DPS, movement and the like.  Say the raid does half of its maximum DPS here.  The boss has 20 million HP and the raid’s maximum DPS is 100,000.  And forget about potions and trinkets etc. for the moment (I’m keeping things very abstract and continuous for now, but will revisit some of these assumptions below).

  • Without Bloodlust: P1 takes 100 seconds, and P2 (at half DPS) takes 200 seconds.  Boss dies in 300 seconds.
  • With Bloodlust in P1: the raid does 130,000 DPS for 40 seconds, doing 5.2 million damage.  The remaining 4.8 million of P1 take 48 seconds.  P2 still takes 200 seconds.  Boss dies in 288 seconds.
  • With Bloodlust in P2: at the start of P2, the raid does 65,000 DPS for 40 seconds, doing 2.6 million damage.  The remaining 7.4 million damage of P2 takes 148 seconds.  P1 still took 100 seconds.  Boss dies in 288 seconds.

“Ok Hamlet,” you might say, “what kind of math voodoo is this?  Bloodlust to add 15000 DPS for the 40s duration is equally effective as Bloodlust to add 30,000 DPS for the 40s duration?”

Continue reading

Quick Notes on Using the Wrong Gems

This was prompted by two recent conversations I had–first, the one surrounding this tweet, and second, someone telling me about a guildmate of theirs complaining about a different guildmate who was using bad gems for their class.

So, the odd thing about someone using completely wrong gems (or some comparable character setup choice) is that it doesn’t matter, on its face.  Gems (especially now) are such a tiny fraction of your overall character strength, and the difference between two secondary stats is such a small fraction of that, that you’re pretty hard-pressed to conclude that someone using the wrong gems has any effect on anything whatsoever.  Maybe in some unusually strong cases, that 1 person out of the 20 in your raid will have a 1% shortfall in their output?  Even that much is probably rather rare.

Basically, if you imagine that you couldn’t inspect people’s character sheets or stats directly, I’d posit that there’s no experiment you could conduct that would reveal to you whether the people in the raid had optimal gems or slightly suboptimal gems.

So why do they matter?  I’ve always only thought there was one reason, and this is including all the way back when I was in a very serious high-level guild and would review applicants.  It’s a way of showing that you’ve researched your class.  That is something you want to know about an applicant or raid member, and is kind of hard to find out directly.

It’s a little tricky because they don’t necessarily need to be able to know the theorycraft rationale for why that stat is preferred.  Just like the gems themselves aren’t the actual point of interest, being able to do/understand a lot of math isn’t the right criterion either.  It’s entirely possible for someone to understand the spellcasting decisions, which is the actual important issue, without articulating the theory.  But I think what it comes down is, there’s really no way to develop an good practical understanding of the class that won’t, somewhere along the way, involve the information about what stats are best.

Practically speaking, the only way to get solid information on the best play of a class is by reading resources written by others who used some kind of mathematical tool.  And those always include information on stats (conceivably they might not include that topic, but that would be rather odd).  The only exception is if you’re the person making the tool, in which case you of course can work out the stats for yourself as well.  If you’ve gotten all of your class understanding from sources (including “play experience alone”) that don’t contain stat information, then you must not have looked at things which are rigorous.  And that actually is something worth selecting for in applicants and guildmates.

So gems turn out to be good signal.  It’s not very discriminating, sure.  Someone might be using the right ones just by some whim, or by following their Attunement (which is often right), or from seeing other players do it.  But I’m of course not saying you should conclude that anyone with the right gems will be a great person to play with (wouldn’t be nice if that worked).  I am saying that using wrong ones all but rules out that the person has used any meaningful resources in learning to play their class.  And that is very likely to have more than a 1% effect on their performance.

Quick Notes on Menagerie Pet XP Farming

Today is the Grubbles/Scrags/Stings daily at the Menagerie.  It’s one of the ones that’s suitable for farming rapidly for XP (although not quite as well-known for this as the Jahan daily).  I figured I’d use the chance to mention a few thoughts on Pet Battles.

I captured a short video to make clear what I’m talking about.  This is two repetitions of the fight, with a carry pet, in 2.5 minutes.  It gets the carry pet (Frostwolf Pup) from level 1-12, although it would go considerably faster if I had both levels of Pet Treat on in addition to the Safari Hat.  With those it’s 4 battles to level up to around 22 (at which point it’s easier to get the last few with Training Stones).  So you can grind this as much as you want on the appropriate days, powerleveling as much as 1 pet every 5 minutes.

The basic question is whether or not this should be prevented, but that question brings up a few interesting points about how the Pet Battle system as changed since its introduction.

Levels, Rarities, and Breeds

I see Pet Battles as possibly exceeding anyone’s expectations as far as how rich they wound up being.  As of now, the gameplay inherent in the massive number of pets alone (and their complex combat interactions) would be a great game, even if there were no levels, rarities, or breeds.  I know why those things are needed to some extent (rarities give some excitement to wild pets, levels so that initial progression exists).  But it’s worth pointing out that most pet battling among enthusiasts is all max level and among rares (both PvP and PvE), so these mechanics don’t play much part, and it has no shortage of good complexity.

Continue reading

Quick Notes on Warlords Trinkets

Just copying some comments I’d made on this in case it’s handy to people in figuring out what’s going on with trinkets.  Trying to remember to do this more where the info might be generally handy and the post is clear without context.


I liked the Mists model that generally had a primary stat and a secondary stat on each trinket, in equal amounts, with either one or the other replaced by a proc. Made it easy to conceptualize and evaluate trinkets at a glance. So for example, at equal ilvl, you might have:

X pimary + X secondary (passive)
X primary + [proc for 6x secondary lasting 20s, 115s ICD]
X primary + [Use: for 5x secondary lasting 10s, 60s CD]
X secondary + [proc for 6x primary lasting 10s, 0.92 RPPM]
X primary + [unique proc tuned to equal X secondary]
[unique passive tuned to equal X secondary] + [proc for 6x primary lasting 10s, 0.92 RPPM]

And so on (the above actually decribes most Mists trinkets). Technically yes, it’s all consistent with itemization where Int is valued at 2x a secondary. But you’re not even getting to the point of having to worry about that; you just stick with a basic framework where a trinket has two “halves”.


Warlords started very similar to that, with two caveats:
1) Spirit and Bonus Armor take up the “primary” half where they appear. I’m just going to flag that and leave it alone for now.
2) Some trinkets were off at release, with a proc that was too high or too low, and they’ve been adjusted by changing the passive rather than the proc, leading to some confusion.

Continue reading