Monday, January 4, 2010

Critical Strike vs. Spell Damage

I recently spent the time to earn the 210 Triumph badges to buy myself a second set of T9. It really didn't take all that long, since I'm cueing as a healer anyhow. But now that I've got a good set of shadow gear, what gems do I put in it? Aside from a little bit of spirit, the holy T9 is almost identical to the shadow T9. The set bonuses are different, of course, but the stats on each piece are very much the same. The equip bonuses though, offer more hit rating, as well as crit and spell damage. So I began wondering if I should gem for +crit or +spell dmg.

After doing a bit of research, I found that the answer was a resounding "+spell dmg"! Make sure you're hit capped first though. Crit rating only increases dmg by 150%. So raising crit strike chance by 1% only increases total damage by 0.5%. However, increasing spell damage by 1% increases total damage by 1%. Even when you take into account the damage coefficients and the fact that the spell's base damage isn't increased by +spell dmg, you're better off with spell dmg.

The break even point is at about 35% crit strike chance. That's pretty hefty for a shadow priest. To even get that high, you'd have to sacrifice quite a bit in other areas that would most likely make the return worthless.

Monday, December 21, 2009

Effective mp/5 from spirit?

Ok, so a while back, I posted about which was better, intellect or spirit. I eventually concluded that spirit was best for healers but not for dps. DPS generally stay in the FSR because they're casting as often as they can. While healers only need to cast when someone's hurt, they can benifit from spirit-based mana regen.

Now, with that said, which is better, spirit or mp/5?

Instead of comparing the mathematical forumlas, wowwiki describes the relationship between spirit and mp/5 in terms of enchants.

"For example the two top mana regeneration enchants for chest items yield either 15 spirit or 6 mp5 ([Enchant Chest - Major Spirit] and [Enchant Chest - Restore Mana Prime]) - this shows the factor of 2.5 very clearly."

They surmise with this, that 1 mp/5 = 2.5 spirit. I don't agree. I think that mp/5 is worth far more spirit. To demonstrate, I've done some calculations using excel again with the same input parameters as I did in the intellect vs spirit discussion. Assuming a character has 1000 intellect and 500 spirit, the following graph shows their equivalent mp/5, depending on their time spent outside the FSR.

Realistically, the value of 1 mp/5 compared to spirit depends largely upon time spent outside the FSR. Let's assume that a priest spends 25% of their time outside the FSR (which is fairly generous, since we decided that 13% was the break even point last time). At 25% outside the FSR, the effective mp/5 from spirit is 79. Now, if we divide the spirit required by the acquired mana regeneration, you get:

500/79 = 6.33 spirit = 1 mp/5

mp/5 is worth 6 times more than spirit, not the 2.5 that wowwiki suggests.

Now, say you think you're an exceptional healer, and get 50% outside the FSR:

500/132 = 3.79 spirit = 1 mp/5. That's still higher than wowwiki's 2.5 suggestion.

In fact, if we solve the equation 500/x = 2.5 (x = 200), and compare that mp/5 value to our graph, we find that we'd need to spend 75% of the time outside the FSR to achieve this mp5 to spirit ratio.

So, spirit is disproportionally availiable on gear. Take gems for example. The highest epic gem granting spirit give +20 spirit. While a +10 mp/5 gem is availiable. The mp/5 is worth +63 spirit according to our calculations, and doesn't have the FSR limitation.

I've now geared my priest so that it has 150 regular mp/5, with 965 spirit and 1182 intellect unbuffed which equates to 139 mp/5 at 25% FSR. A total of 289 effective mp/5. I'd love to swap out all my spirit for mp/5 if I could. At an exchange rate of 6.33 spirit per mp/5, I'd get 152 additional regular mp/5. The problem is that it's just not availiable in those quantities. Since gear is lacking in mp/5 equip bonuses, it is impossible to swap out all your spirit for mp/5. I could trade out some gems and upgrade a few pieces, but it's tough to come across, especially below raid-quality gear. Healing gear comes pre-loaded with spirit, just not at the exchange rate we'd expect. I assume this is to keep the difficulty level of the regular and heroic 5-mans.

Last Night's PuG

So usually a post starting with this title would be a long rant about some terrible group that I ended up with over the weekend. However, I am pleased that is not the case today. In fact, it is exactly the opposite.

I've decided that groups are notably better in the hours when kids are not home or are not allowed to play on the computer. Groups during the middle of the day 9am-3pm are pretty good. But the exceptional groups are those that occur after 10pm.

After running a few groups with a paladin friend of mine, he decided to run on his mage for some badges. So we cued in the dungeon finder for a tank. We ended up with a paladin named Chunchi on Stormreaver. He had 36k hp unbuffed, which is pretty decent but isn't phenominal or anything. That's totally attainable for paladin tanks who try. However, throw on a few buffs and he was pushing 43k hp. The truth was that this guy just was amazing. He made me stop and wonder about all the other tanks in our battlegroup (Reckoning). All of the other tanks should be ashamed of themselves.

We ran back-to-back instances for about 4 hours. Let me remind you that this was a PuG for the most part. Chunchi was pushing 3.5k dps while tanking. I'm really not sure how that's even possible. The best part was at the beginning of H HoL, I told him, "We have a pretty geared group, let's take this quickly." And he responded, "Ok." Now, I originally thought he'd pull two groups at once, or something of that nature. Instead, he pulls entire hallways. He runs to the very end, and hangs onto every last guy. There were no runners taking out the clothies, no spell-casters trailing behind. Once he got to the end of the hallway, he beat the crap out of them while we all AOE'd. At times, our 3900-gearscore mage was pushing 11k dps, and ended up with an average somewhere around 6k dps.

You'd think with that many guys beating on our fearless paladin, that he'd be taking significant damage. But for some reason he really wasn't. Well, he had 26% dodge, 20% parry, and 15% block. Beyond that, at 27k armor, he's already mitigating 65% of the received damage. That means, that he wasn't even getting hit 61% of the time, and when he did get hit, he only took 35% damage. (0.39 * 0.35 = 13.65% overall damage taken)

Now, there's an exponential principle here. When one person in a group becomes better (better gear or skill or both), the group does not simply increase by that much. (ie: it's not mathematically 'addative') Instead, everyone becomes better. But it's not mathamatically multiplicitive either. It's not that everyone in the group increased by some delta x. Instead, there is exponential increase... let me show you why.

When the tank becomes better, he takes less damage. Thus, the healer does not need to heal as much. Which in turn means that the priest has more mana (which can be turned into DPS if the group is good). Also, an increase in a tank's gear means that he can hold aggro on more mobs because he can sink more overall damage. This means that the DPS can use more AOE and do increased damage to more mobs at once, killing them faster. Hence, dungeons are run faster, more loot is obtained, and more badges are earned. Exponential increase.

I was lucky enough to get to DPS for the last two instances of the night. I popped up shadow spec, and went at it. During H ToC, I peaked at around 7k dps, and averaged about 3.5k dps. That's not too shabby, considering that half of my gear is shared with my holy healing set.

So, when all was said and done, we averaged about one badge every 3 minutes, totally nearly 100 badges when we finished. I had amassed about 230 triumphs, and collected my entire T9 set in one night. I went by the auction house and dropped about 1,500 gold on epic gems, and re-enchanted my gear. If only I got such an awesome group every evening, I'd have another 230 emblems to drop on my shadow set in about 3 days.

Thursday, December 10, 2009

Smarter Group Matching

So the Random Dungeon matching system in the new 3.3 patch brings up some interesting questions, some of which I'm not too keen about the 'answers.'

"The new system will now work to match at least one experienced player for the assigned dungeon with less experienced players in the group. It will more carefully match class types, and will more consistently match players of equal levels for the dungeon chosen. The interface will also allow players to see how many players are looking for groups as well as the average wait time of groups that the tool has formed." --www.wowhead.com

"The Dungeon System will make sure players are of the appropriate level, have the appropriate attunements for a dungeon (if applicable), and look at the average item level of each player. The system will put together a group according to the roles (tanking, healing, DPS) players have selected to fill and attempt to match characters as best as possible according to the items they have equipped." --www.projectlore.com

So there seems to be a bit of discrepancy between these two quotes from the Blizzard Q&A session. The first seems to indicate that there is a sense of "experience" that Blizzard can surmise about one's character. There's no details on what distinguishes one from the other. Is it achievements? Total quests completed? How many times you've run an instance? Or is that simply meanig that someone clicked the button voulenteering to be the group leader? Furthermore, it says that they will be intentionally mixing these experienced and in-experienced players together on purpose! Ugh.

While, on the other hand, the second quote suggests that the average item level on the character is going to be examined. Sounds like Blizzard is implementing their own sort of gearscore. The quote, however, does not say how the average item level would be used. Is that what they mean by "experienced"? I sure hope not. At least with pre-3.3 PuGs, I got to screen certain people out in an effort of getting the best possible group that I could. It sounds like Blizzard is, in effect, making sure that you don't end up with a "Dream Team" group and that you never will.

It's important to note that it's a Random Dungeon finder, not a Random Group finder. The group is not random at all... it's been mettled with using some criteria that no one knows exactly. I can't friend-list these cross-realm folks, so there's no way for me to keep track of who on what server is actually any good and play with them again. I'm not out there to challenge myself by playing with idiots and see if I can still make it through an instance. I'm out to make my life easier by grouping with people I know can get the job done, and not annoy me to death while doing it.

I still maintain that some people make it to 80 not because they're good, but rather because they're persistant.

Patch Day (v3.3) Recap

So patch day came and went, yet I fared better than most.

I logged on early Tuesday morning to find that the patch had gone live... and that my background downloader hadn't done that great of a job. It had only downloaded about 120 Mb of the 680 Mb patch. So, it spent the next 2 hours downloading. It finished installing just after 11:00 am, at which point I tried to log in and found that daylight savings time had struck again, and that Arizona is an hour off from California where my server was based. So, I had to wait till noon.

Maintenance was extended by about an hour and a half, so I read the patch notes and some patch 3.3 stuff while I waited. However, twitter was a great resource for finding out when the realms were coming online without having to constantly be typing in my password over and over again.

I finally got logged on at about 1:30 pm and started in on the random dungeons. The powers that be decided that the first dungeon would be Heroic Occulus, which everyone in my group throught was pretty lousy. However, patch 3.3 just nerfed Occulus a bit, so we went at it. We had zero problems with it, and it was actually pretty enjoyable.

I'm sure though that once this tapers off a bit, you could easily get grouped with a bunch of noobs who have no clue what they're doing in Heroic Occulus, and that it could go very, very wrong. My groups have all been wonderful through the random dungeon system so far... I'm not sure of the algorithms behind it. However, I have been selected as Healer every time so far, even though I always select Healer and Damage. I guess there's still a shortage there. I don't think there's anything in place for keeping out undergeared players, but I suppose you could always /votekick them.

The second dungeon that was randomly selected for me was the brand-new Icecrown Citadel wing. We pounded our way through those, even though no one had ever been there before. We assumed tank-and-spank on every boss, which proved to be efficient. I was a bit let down with the lack of complexity of the boss fights, especially for heroic. But then again, I was most likely playing with the hard-core WoW'ers who signed on as soon as the realms came up, and not the 12 year olds who were still in school. My group made it through the first two, but then the authentication server went down and we lost 3 out of our 5 group members before Halls of Reflection.

I was able to go back and do my dailies by myself while 90% of the WoW population was stuck offline. I considered myself lucky to be able to play at all.

Once the authentication servers came back online, there were all sorts of trouble with the instance servers being too over-crowded, and not able to launch more instances. Boo. Anyhow, I still came away with many more badges than I could get before, and a few frost's too.

Monday, December 7, 2009

Patch 3.3 Buffs Up Priests

Reading the patch notes on 3.3, I noticed the hefty buff to priests, yet the 1-liner for warriors. Regardless, priests have waited their fair share for this one. Albeit, the changes across all classes are mostly to deal with the cross-realm arena changes to spell cooldowns, but shadowpriests are going to enjoy this one.

Priests
Divine Hymn:
The cooldown on this spell has been reduced to 8 minutes, down from 10 minutes.
Power Word: Shield: This spell can now be cast on non-raid/party friendly targets.

Shadow Talents
Improved Devouring Plague: This spell now deals 10/20/30% of its total periodic effect instantly, up from 5/10/15%.


This will not increase the total damage that the spell does. Instead, it just makes the initial cast of the spell more powerful, and by so doing, the DoT weaker. It will likely up over-all DPS for priests though, as many DoT's don't take full effect because the target dies before the duration of the DoT. Since the initial spike of damage is higher, you're unavoidably going to get more damage off on the opponent. You could potentially get more killing blows or something like that, but likely you'd use Shadow Word: Death instead.


Mind Flay: The range of this ability has been increased to 30 yards, up from 20.

About time they up the range on this spell. Still not quite comparable with arcane missiles. I don't forsee much of a DPS improvement based on this change, although it could help slightly. It just allows us to be further from the melee damage. Mostly just convienience, but it is possible that since we don't have to keep repositioning ourselves, that we may put out a slight amount more DPS on that.


Shadowform: This talent also now causes Devouring Plague and Vampiric Touch to benefit from haste. Both the period length and the duration of these spells will be reduced by haste. In addition, the mana cost has been reduced from 32% to 13% of base mana.

Here's a major change. Haste always means higher DPS. The implications of making the DoT's tick faster though, is that it will be re-applied more often, causing increased mana consumption. You might note that the mana cost has been decreased by 19% of base mana, but I'm not sure what that entails in the long run. This works well if you've got good mana regen based on MP/5, but you're most likely still going to be running lower on mana than you were before.


Vampiric Embrace: This talent now provides a 30-minute buff that cannot be dispelled, instead of a target debuff and only generates healing for single-target shadow damage spells.
Pet

This is how I always thought the spell should work. Since VE no longer needs to be applied to every target, you automatically gain spell-casting time. One more spell can be cast for each mob which will deal damage instead of being an inactive cast. This obviously means an increase in DPS, but also implies that your first cast will be damage dealing, and thus you will have higher initial agro. Another consideration is that many shadow priests don't use VE because of the loss of DPS so a sudden resurgance of the use of the spell might cause trouble for shadow priest aggro. You will now be generating more threat based on DPS as well as healing aggro. This is indicitive of multi-mob groups where the tank must get control of all of them. But then again, maybe those shadow priests out there will actually use it and help us holy priests heal a little.


Avoidance (passive): Now reduces the damage your pet takes from area-of-effect damage by 90%, but no longer applies to area-of-effect damage caused by other players.

A nice tidbit so that you can use your shadowfeind without losing it to other AOE damage happening around it. I take this to mean that it can now get AOE'd in PvP, but makes it more viable for PvE.

Account Security

Recently, it has come to my attention that there is a major shortcoming in Blizzard's account security. In years past, a simple password has kept many people's data safe. Recently, companies have increased the rules on user passwords regarding length and alpha-numeric characters (neither of which Blizzard requires). Those types of increases in security only affect brute force attacks, where a hacker attempts to break your password by applying dictionaries or sequentially generated passwords over and over.

In scenarios where key-loggers are used, the player's password is stolen as they type it in; regardless of how good of a password they chose. How could Blizzard keep players' acounts safe when viruses, trojan horses, and other malware is stealing passwords on the user's end?

A solution Blizzard came up with was the "authenticator." The device works by counting the seconds from an arbitrary date. The device then encodes this number with an encryption key. The authenticator's serial number is registered with a blizzard account, thus informing Blizzard of which encryption key it is using. When the 6-digit code is entered along with the password, the code is compared against blizzard's records. It decrypts the 6-digit number with the key associated with the player's account. If it decodes correctly, it should have the number of seconds elapsed since some given date. It then can compare that number of seconds to the blizzard server, and tell if the code was generated within the last 60 seconds. Once used, the 6-digit code is no longer valid. If a hacker acquired the 6-digit code via key-logging, and tried to use it, the code would no longer be valid because the user had already entered the code previously, and a new code from the authenticator would be required for a subsequent login. Likewise, attempting to use a code from a different authenticator would mean that the number which was decoded would not likely be within 60 seconds of the actual time, and therefore would be interpreted as an invalid code.

So what's the big deal, you might ask. Although Blizzard has made authenticators availiable to those who wish to buy them, there are a great deal of people who do not want to purchase a piece of hardware to protect their game accounts. Unless blizzard offered an authenticator to every person free of charge, there will always be problems with account security. Why would anyone buy them if they felt their accounts were already safe? Also, consider the fact that Blizzard makes money from selling the authenticators.

The truth is that Blizzard has neglected to put in place simple measures to protect WoW accounts. Let me explain a simple solution to the problem that does not require a hardware authenticator, and probably minimal programming by Blizz:

Imagine that along with your password, you were required to enter a 4-digit pin, which you could choose. Now, instead of using the keyboard to enter the pin, a digital keypad displayed on the screen, much like the numpad or a telephone keypad. The buttons on the screen would be pressed using the mouse. For all those programmers out there who are skeptical at this point, it is true, that this alone would not defeat the ability to log mouse x,y coordinates and mouse clicks to steal a person's pin.

Here comes the truely tricky part that sells the deal. For each digit that the user needs to enter, the keypad numbers are randomly scrambled, such that the user would have to find the correct button to push each time. While this would be a minor inconvienience, it would prevent anyone snooping on mouse coordinates and mouse clicks to actually know for sure which numbers were pressed. This allows for 10! (factorial) or 3,628,800 possible combinations of digits.

The genious of it is that it uses a totally seperate processing engine (your brain) to decode the scrambled numbers. Since your brain easily does the matching of the location of the button with the desired digit, the randomization of the button locations is, in effect, encrypting the data. Since your brain does not have to evaluate the randomization of the digits, but rather find 1 digit out of ten, your brain on average only has to scan n/2 of the numbers. Most of the digits will be found after scanning 5 digits visually. 5 digit scan average * 4 digits in the pin = 20 scanned numbers to enter the pin. Not too much work for your brain.

It's obvious why this method is not used for entering passwords: The alphabet contains a significantly larger number of characters to scan through to find the letter you want to press. (26 letters + 10 numbers)! (factorial) = 1.06387358923717e+56. Yikes, that's a lot of possiblities on random aphabets. Then do that for each character in your password. (That would be some immense security!) Then your brain needs to scan the alphabet and pick out the correct letter/digit. It would find it at an average of n/2 or 18 scanned characters. 18 scan average * 8 characters minimum in a password = 144 scanned characters to enter the password. 144 is an order of magnitude larger than 20 for a 4-digit pin. It would be a major inconvienience to the user. However, when only 10-digits are used on the keypad, the user could quickly scan through them to find the appropriate number... especially if they only needed to do it for 4 digits in a pin.

Now you might ask... they could still steal your password, and then knowing that the 4-digit pin was always a number, that only allows for 10,000 possible numbers to choose from. A computer could quickly brute force a pin, especially being that the average pin would be found at n/2 or 5000 iterations. To prevent this, all Blizzard would need to do is simply lock down the account after 25 (or whatever number they chose) incorrect password attempts in a row, then brute force-ing the pin would most likely fail.

It's not foolproof, mind you. People could still hack your account using other methods. But you'd think that Blizzard would save a lot of money by eliminating all the time spent on dealing with fraud issues and hackers by implementing such a plan. It could be rolled out in a regular patch, and deployed to every WoW player in existance with absolutely no cost to the player. So why don't they do it? There must be some trade-off that I'm not considering. Either the savings don't outweigh the benefits, or that hackers would simply find some other way to cheat the system. Or perhaps blizzard makes a decent amount of money from the new accounts started by hackers and gold farmers, and they really don't want to cut out a source of revinue... after all, in-game resources are free. They can just create you new virtual items and gold at no cost to them... just to pay some GM $10/hr to do it. And at the pace they respond to such tickets... (3 day response time for me) they must not be paying too many GM's.