Comparebot is a useful analysis tool that takes World of Logs parses and winnows that information down even further. It does not display more or better information than World of Logs. It does, however, do more legwork for the user than WoL, pulling out important bits of information and presenting them in ways that are easy to access and understand.
Analyzing Details Pages
At the top of the Comparebot page you see a section titled “Compare Parse Details.” Below it are lines for entering URLs of WoL details pages. You can’t put in URLs for a whole fight or for overall damage done; you have to select a player or npc by name and get the URL for their details page. The details pages for the whole night, for all attempts on one boss or for just one attempt all work.
You have the option to input up to three details pages to analyze side by side but putting in only one is fine too. Let’s say you put in the URL for your details page on a boss kill. Once you click “Compare Parses” below the input lines and the analysis is done, you’ll be taken to a summary page where you will see important stats. Did you work in all the Rapid Fires and Call of the Wilds that you could? Did you click on the Lightwell? Beyond the Summary page there are other tabs related to damage taken, damage done, buffs, debuffs and so on. Some of the information is useful, some isn’t. Overall, a Comparebot analysis is helpful for letting you know whether you did a good job or not when it comes to things that you can quantify, such as damage done, avoidable damage taken, potions used and the like.
Example 1: vodka links to their WoL parses and so I’m going to act like that makes their hunters fair game as examples. The link in this example takes you to Cev’s stats from a 3-22-11 kill of Omonotron 25H as analyzed by Comparebot.
You can also run the details page of an NPC, like Blood of the Old God on the Cho’gall fight. You can look at its debuffs to see the uptime on Ice Trap. You can look at its damage taken to see how well your AoE stacked up against others. You can also see which players did the most damage to the adds of the given type under the Targets tab.
Example 2: This is a link from the same Omnotron kill as Example 1, but it analyzes Poison Bomb adds rather than a particular player.
The real utility of the “Compare Parse Details” section comes in when you actually compare parses by running more than one details page. This is where Comparebot goes beyond WoL’s functionality and starts doing something pretty cool: juxtaposing different details pages. The same information and formatting is present from the single parse analysis, but extra columns are added for every details page included beyond the first. Just like before, you can analyze any WoL details page. It can help to add your own names to the pages on the extra text input lines next to the url entry lines in order to be able to tell them apart.
Example 3: Here we have the same parse of Cev’s 3-22-11 kill of Omonotron 25H as linked above, but juxtaposed with it are Cev’s stats from a kill of the same boss from a week before. Note that there are differences between the parses, such as the fight duration, that affect things like dps, Rapid Fires used, etc.
First let me state that you don’t have to make actual comparisons. You can throw in different details pages that are incomparable but are still just nice to have all in one place. For example, you can juxtapose the details page for yourself for all attempts made on a boss for a night with the the details page for yourself on the kill. You obviously can’t reasonably compare damage done between columns,but you do have a lot of nice data in front of you all at once.
Say you do want to make comparisons, though. There are some caveats. As with all theorycrafting and data analysis in WoW, you want to be as scientific as possible. Since strict laboratory controls are impossible, we might instead say that you want the comparison to be reasonable. For example, comparing the dps of two hunters in the same raid on the same fight is sort of reasonable because most variables are held constant (that is, they’re the same) between the two details pages (fight duration, buffs, etc.). The only things that vary are skill, gear, spec and maybe a bit of rng and so you have a somewhat reasonable basis for comparison. Roles may also vary and so it helps to be aware of those fights where hunters can be assigned to do particular things.
The more you can hold constant and the more you isolate a variable of interest as the only thing that actually varies, the better the comparison and the more you approach what scientists call a controlled experiment. An ideal parse comparison would be between two hunters in the same fight on a tank and spank with identical gear and specs, leaving the only difference as their skill. However, expecting that sort of comparison would be unreasonable. Raids are not labs. We have to make do as best we can when we look at logs and all the while keep in mind the blatant flaws in what we’re doing.
What can we do to increase the quality of our parse comparisons, even if they’ll never be as neat and tidy as we’d like?
- Hold constant what you can. Make sure it’s the same fight in the same mode. Make sure the fight durations are the same. Make sure the roles are the same (comparing a kiter and a dpser is just silly, for example). Basically, you want the comparison to be fair and reasonable. Consider concepts like face validity, internal validity, and external validity. Whatever relevant variables you can’t keep equal between the logs you have to take under consideration as a possible explanatory factors for differences between the logs.
- Compare things that are straightforward to compare. Contrasting dps measures is difficult because of all the things that go into composing dps; you can really have little idea as to what is contributing to a difference and what is not. Comparing the numbers of Rapid Fires used is simple, though. All that goes into that stat is skill, spec and fight duration and so it is quite easy to evaluate.
- Compare stats that actually mean something. Caustic Slime damage taken on Chimaeron is a silly stat to look at since it’s all either random or raid-wide. It means almost nothing. Corrupting Crash damage and/or hits on Cho’gall, however, is very useful to look at since it is almost 100% avoidable and should be avoided.
- Know a stat well before you try to analyze it. For example, if you compare two hunters’ uptime on Aspect of the Hawk you might be tempted to say that the person with the greater uptime is doing the better job. However, anyone who knows hunters knows that it isn’t that simple. A hunter who spends more time in Aspect of the Fox on a fight like Atramedes will have lower uptime on Hawk but may actually be doing a better job because she’s getting more shots off on the move and is thereby doing more dps. You really have to know hunters in order to analyze them, just like with any other class.
- Know the the fight that the stat is taken from. You can’t really understand the data without also understanding the context.
- Keep an open and incisive mind. All sorts of things can explain stat differences and not all stats are visible on Comparebot.
- Conclude only so far as the data allows. You don’t want to be Wile E. Coyote, running out into thin air, thinking there’s still a bridge (data) supporting you. You also don’t want to be one of those skeletons that a real hero comes across later, a failed adventurer who strode out onto a rope bridge only to have it give way under the slightest breeze.
Below the section on details pages there is the section entitled “Full Raid Summary.” This section’s text input line will accept dashboard pages for full nights, full sets of attempts or full fights. The data synthesis laid out in a Full Raid Summary is quite useful for analyzing raids (who’s using potions, who’s using their cooldowns, who took damage from what, etc.). It again does not (it cannot) display any more information than WoL displays; it merely displays useful information in an accessible manner. The raid summaries can be helpful for comparing different hunters in the same raid, but they really are better used as snapshots of whole raids than as an analysis tool for looking at a specific player in depth. Note that it takes a while for Comparebot to calculate full Raid Summaries.
Example 4: Here again is an analysis of that 3-22-11 vodka kill of Omonotron 25H, but this time using the Raid Summary tool.
Questions? Concerns? Please leave a comment below.