Blood Pact: Using CompareBot to see differences in logs

Every week, WoW Insider brings you Blood Pact for affliction, demonology, and destruction warlocks. This week, Megan O'Neill didn't waste time getting a Whole Body Shrinka' so she could re-enact the Genie describing his living arrangments.

Originally, the last part of this column's series on World of Logs (WoL) would be talking about the other parts of RaidBots (EpeenBot and DPSBot) with WoL rankings, but as time went on, I realized I wanted to look at that data over a patch or tier rather than over just a week. Besides, I believe it's time next week for a break in these serious warlock matters.

The previous four posts went over a very basic introduction to WoL, the graphs of WoL important to warlocks, looking at buffs and debuffs on a timeline, and digging deep into the combat log. This final post will look at comparing two logs more easily than flipping between browser tabs.

What is CompareBot and what can it do for me?

RaidBots (also know as DPSBot) is a site that pulls data from public logs in World Logs and compiles information about overall damage done (or healing done) across all the specs, classes, and fights. In the end, it's a good tool for looking at trends in the overall raiding population from scrub to pro, at least according to what's publicly available on World of Logs. (It's well known that many top guilds log privately; on the flip side, casual players who aren't fully involved with reading up on WoW blogs generally don't even log at all.)

However, DPSBot's graphs are not necessarily great at determining how well a spec is performing its intended job, despite what QQers might want to claim. There are a lot of caveats about generalizing or minimalizing the differences between raid groups in spec stacking and strategizing. RaidBots doesn't measure utility or even target-specific DPS, so padded damage also needs to be taken into account. CompareBot, however, uses very specific parses; you must choose a specific boss fight or time frame AND a player's detail page.

Who should you compare? You can compare yourself one week to yourself another week. You can compare yourself on a fight to a top warlock parse of the same fight. The most often comparison I make is myself with another guilded or applicant warlock after we've played in the same raid. The more you keep the same about the compared parses -- spec, gear, fight length, fight type, particular raid parse -- the more direct your comparison can be.

Example: Demonology on Lei Shen

Demonology is literally the middle spec: it is listed between affliction and destruction. But it also is a good hybrid of DoTs and direct damage -- affliction is an extreme of DoTs and destruction is an extreme of direct damage -- and it have plenty of AoE and pet options to boot. So it makes a pretty good spec to chose when you're learning how to look at various types of warlock damage by the numbers.

When I put parses into CompareBot, I like to note which is which in the optional note box. CompareBot's tables will always use "1" and "2" unfortunately, but it provides a table at the top with the details of the parse numbers should you forget which is which. If you are comparing two distinct raid nights, try to pick either two kills or two similarly timed attempts. If you'll remember from talking about the WoL graphs, comparing an attempt to a kill is often ill-advised, since an attempt will likely be missing one or two phases from a kill. Comparing two similar attempts helps you figure out how the two parses dealt with that particular phase mechanic, but comparing two kills will help show the entire toolkit at play.

As you can see, my demonology spec would prefer my affliction gear had far more mastery on it, and I am definitely not a superb demonology player, but it helps get the AoE job done on Lei Shen. To combine all the lessons at once, I did a comparison on Lei Shen 25N between myself (Ponerya / Parse 1), the other demo 'lock in my raid (Gv / Parse 2), and a ranked demo 'lock parse -- specifically #60 of demonology warlocks on 25N Lei Shen US & EU -- of similar time frame (Rayelle / Parse 3).

Tabs by the numbers

I'm actually going to go a little bit backwards on the tabs today. The tabs are fairly straightforward in what information they provide, but we can talk a little bit more in detail about what tables provide what kinds of insight for warlocks.

Buffs and Debuffs: Uptime is typically the set of columns that does the job for you. Having a buff overwrite itself is not very good, so sometimes the Amount or Amount per Minute can be misleading as to how well buffs were handled. The buffs page also unfortunately lists every buff, including all the static raid buffs that last for 60 minutes, so you'll have to dig through the list to find your trinket and potion specific buffs.

Healing and Damage Taken: Typically this would be a healer-oriented table, but you can use the bottom table on the Healing Targets tab to judge either how well you avoid damage by looking at how much healing you took from whom or how trigger-happy your healers are to heal your Life Tap.

You can use the Healing tab to dig into the numbers of self-heals or the massive tables of the Healing Taken tab to dig into what the raid's healers did on you by specific spell. There is a note at the bottom that absorbs are guessed at, which is the usual problem of trying to parse healing absorbs.

Finally, you can dig into the numbers of what specifically was damaging you with the Damage Taken tab. You'll notice that Rayelle took a quarter of the damage Gv and I took from Ball Lightning. This might be because my guild's strategy is to stack up with Ball Lightning and power our way through with AoE damage and heals.

Targets: This is an important tab for DPS. A high damage done overall could be because the warlock is fantastic at single-target or it could involve damage done on multiple targets through AoE or multidotting.

Amount is the raw damage done number. On targets that would be focus-fired down (as opposed to AoE), typically higher damage is reflective of better buff and cooldown usage. This means either cooldowns were used during focusing on that target or when appropriate procs happened outside of primary DPS done, DoTs and other affects were refreshed on the target.

Percent is the percent of damage on that target of that player's overall damage dealt, and it's useful for figuring out how much time a player spent on a particular target. Because of DoTs, warlocks can have a higher percentage of damage done on a boss when not actively targeting the boss, so a high boss percentage in a multi-bodied fight is typically the beginning of a sign that the warlock multidotted well.

Damage: Finally, the big daddy of all tables. This breaks up by spell and column category your damage done to anything. I like to check off all the boxes at first and then turn on specific columns rather than trying to scroll through the entire table. Here's a summary:

• Missed / Immune: Hopefully this is all dashes all the way down. Your warlock should be hit-capped, even if affliction can fudge it a little under. If your warlock is dodging or parrying, maybe ask the question "Why is the warlock tanking?"

• Damage/DPS: Damage from Hits and Damage from DoTs might be easier to read than the Total Damage column. The caveat here is that the Ability DPS number comes from the Total Damage divided by the parse length in seconds, rather than representing the DPET or Damage Per Execute Time (DPET helps us measure efficiency of abilities in a rotation).

• Crit %: This is usually where you'll find your RNG argument if there is one. It's also super-easy to tell if a warlock used the Lei Shen trinket correctly (as Rayelle did).

• # Hits/DoT Ticks: This is not as useful as the Per Minute version.

• Hits/Ticks per Minute: This speaks to the activity (or haste levels for DoTs) of the warlock. A warlock who dies mid-fight might have a lower overall hit count, but if she has a high per-minute rate of hits of the right spells, she's likely a skilled damage dealer (despite having floor problems).

Context and caveats

The first tab CompareBot presents you with is the Summary tab. This tab shows the obvious reasons why one player did better than another: major cooldowns.

You'll see from the comparison that Gv and I played similarly when it came to major cooldowns. We actually both potted as well, as you can see in my parse, but for some reason it's not showing up in CompareBot. Rayelle, on the other hand, is currently a troll, but she had to have been an orc at the time of the parse to make use of that wonderful racial 4 times.

Rayelle also has something weird going on with her Bloodlust count. Normally this box is either 1 or 0, and sometimes you'll see a 2 if the fight lasts longer than 10 minutes. But Rayelle has three Bloodlust counts in a 9-minute fight. How does the even work without dying?

It's actually a simple and nonmalicious answer: three shaman cast Bloodlust within milliseconds of each other. When you look at the buffs and debuffs graph for Rayelle's parse, it's clear she only got Bloodlust's effect once. Since each shaman's Bloodlust applies one Sated debuff to the player, the combat log saw three Sated debuffs applied in total (even if they overwrote each other!), and therefore CompareBot figured that three Bloodlusts were cast.

In summary, pack your salt shakers and beware of what one specific resource for spying into raid parses will tell you. Context is key when evaluating raid parses, so using multiple sources and analyses of the data will help you form a more complete picture of what happened.

Blood Pact is a weekly column detailing DOTs, demons and all the dastardly deeds done by warlocks. We'll coach you in the fine art of staying alive, help pick the best target for Dark Intent, and steer you through tier 13 set bonuses.