The “Easy Sorc” build

p00tx · November 2019

I think many beginners and mid-game players misunderstand the reason so many end-game players tend to passionately advocate for bis gearing. It's not because they lack imagination, or because they want to restrict you in any way. No one wants to control anyone (who the hell has time for that?).

It comes from a slightly selfish place, in the long run. It's because these new and mid-game players very likely won't always remain there. They'll start to get curious, and they'll want to dip their toes into the end game waters. There is a strong possibility I could be running with one of them in future content and would want to ensure this individual had access to the tools they needed to be the best possible team member. If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team. Once you reach that point, it becomes imperative to think about how each individual impacts the entire team, rather than just letting each person do their own thing and hope it works out.

Many of us who enjoy mag builds started out with a heavy attack build somewhere along the lines. I think it might just be a natural part of our progression as a dps. I strongly resisted the idea of swapping to a single target flame/flame build, until I realized I just wasn't going to get any better and my dps wasn't going to improve until I made the change. I hated it, and it took me some work to get it, but once I did it, it opened up a lot of doors to me in terms of potential content.

The difference is, I had some fantastic influences who kept encouraging me to make the change, and had a lot of knowledge in terms of which sets would benefit me. I set aside my stubbornness and listened to them. I farmed the sets, put in the time on the dummy, learning and refining my rotation, and ran a lot of content to put it onto practice.They were absolutely right. Now I'm raiding with those people and I'm working content that I would never have had access to before.

If you're happy doing base game vet content, then this set-up is fine. Not generally optimal, but absolutely fine, and you should stick with it if you enjoy it that much. If you ever want to explore the rest of the content or push achievements though, you're going to need to let go of it and start to listen to the players who have been putting in the hours testing this both theoretically, and in actual content. No one wants to control you, but we do want to potentially play the game with you someday (end game needs more new blood), and you're going to need a better set-up if that's going to happen.

Joxer61 · November 2019

can we just put a fork in this...its done.

karekiz · November 2019

p00tx wrote: »

If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team.

I highly doubt most people just "waltzes" into a VAS HM out of random. Most have to parse to just get into those style runs your talking. At that point, they are probably wearing whatever BIS list is out there. The others probably pay for a carry or are carried, and that point when you are basically a carry, it honestly doesn't matter what you wear.

starkerealm · November 2019

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

p00tx · November 2019

karekiz wrote: »

p00tx wrote: »

If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team.

I highly doubt most people just "waltzes" into a VAS HM out of random. Most have to parse to just get into those style runs your talking. At that point, they are probably wearing whatever BIS list is out there. The others probably pay for a carry or are carried, and that point when you are basically a carry, it honestly doesn't matter what you wear.

As someone who runs this content a lot, I can assure you that it gets pugged pretty often, and we do get some weird builds in there and a fair amount of players who are definitely not ready to be running it. Sometimes we can carry them and it's fine, but when you've had to guild-pug more than one of these players to fill the roster, it becomes too much.

Joxer61 · November 2019

p00tx wrote: »

karekiz wrote: »

p00tx wrote: »

If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team.

I highly doubt most people just "waltzes" into a VAS HM out of random. Most have to parse to just get into those style runs your talking. At that point, they are probably wearing whatever BIS list is out there. The others probably pay for a carry or are carried, and that point when you are basically a carry, it honestly doesn't matter what you wear.

As someone who runs this content a lot, I can assure you that it gets pugged pretty often, and we do get some weird builds in there and a fair amount of players who are definitely not ready to be running it. Sometimes we can carry them and it's fine, but when you've had to guild-pug more than one of these players to fill the roster, it becomes too much.

that comes down to player skill, NOT based off a build.

p00tx · November 2019

Joxer61 wrote: »

p00tx wrote: »

karekiz wrote: »

p00tx wrote: »

If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team.

I highly doubt most people just "waltzes" into a VAS HM out of random. Most have to parse to just get into those style runs your talking. At that point, they are probably wearing whatever BIS list is out there. The others probably pay for a carry or are carried, and that point when you are basically a carry, it honestly doesn't matter what you wear.

As someone who runs this content a lot, I can assure you that it gets pugged pretty often, and we do get some weird builds in there and a fair amount of players who are definitely not ready to be running it. Sometimes we can carry them and it's fine, but when you've had to guild-pug more than one of these players to fill the roster, it becomes too much.

that comes down to player skill, NOT based off a build.

Then I don't think you truly understand either the build in question or the content I'm talking about, and you seem more interested in arguing for the sake of shutting someone down than in having an actual discussion, which means I'm likely wasting my time responding to you, but here it is anyway. End game content is primarily long, single target fights (add-pulls are a part of it in between the bosses, but generally easy to get through and far less important than the boss fights and they don't require a highly specialized AoE build to manage) with two healers in carefully planned out gear keeping up all of the needed buffs, so any benefits this build might have are wasted, because it's not bringing anything that isn't already there, which means it's actually doing less damage than the player thinks it is once it's in a group.

T3hasiangod · November 2019

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

Joxer61 · November 2019

p00tx wrote: »

Joxer61 wrote: »

p00tx wrote: »

karekiz wrote: »

p00tx wrote: »

If someone waltzes into a vAS+2 with a heavy attack pet sorc build using this particular build, it's not going to go well, and they're not going to be bringing much to the table, making it harder on the rest of the team.

I highly doubt most people just "waltzes" into a VAS HM out of random. Most have to parse to just get into those style runs your talking. At that point, they are probably wearing whatever BIS list is out there. The others probably pay for a carry or are carried, and that point when you are basically a carry, it honestly doesn't matter what you wear.

As someone who runs this content a lot, I can assure you that it gets pugged pretty often, and we do get some weird builds in there and a fair amount of players who are definitely not ready to be running it. Sometimes we can carry them and it's fine, but when you've had to guild-pug more than one of these players to fill the roster, it becomes too much.

that comes down to player skill, NOT based off a build.

Then I don't think you truly understand either the build in question or the content I'm talking about, and you seem more interested in arguing for the sake of shutting someone down than in having an actual discussion, which means I'm likely wasting my time responding to you, but here it is anyway. End game content is primarily long, single target fights (add-pulls are a part of it in between the bosses, but generally easy to get through and far less important than the boss fights and they don't require a highly specialized AoE build to manage) with two healers in carefully planned out gear keeping up all of the needed buffs, so any benefits this build might have are wasted, because it's not bringing anything that isn't already there, which means it's actually doing less damage than the player thinks it is once it's in a group.

yea I get all that, but cant the same be said for half a dozen other builds that people would bring that wouldn't "fit" said trial makeup? Yes, there are going to builds out there that perform better/worse than others depending on content/skill/lag...many factors.
My point to ALL of this is it has become a witch hunt so much to the point to where its no longer about the "what" but about the "who"..
That's all I'm saying....lets shut down the blatant attacks and why cant you all just simply say, "The build doesn't work for me". End of.

starkerealm · November 2019

EDIT: Quoted Post is gone.

starkerealm · November 2019

T3hasiangod wrote: »

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

How simple do you need this?

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables.

Thing about that is, the drop tables are a really good way to see this in action because players do scrutinize them. The overall output is low, and the results draw direct player attention. So, when you're running vMA for a lightning staff, and have to run the damn thing 300 times, that's not normal. Statistically you should have seen one in the first 30 runs. However, the dice roll that determines your drop is loaded. Not, maliciously, not intentionally, but the result isn't truly random.

Now, I'm using the loot table as an example because it's an easy place to see this in operation. However, the same behavior does apply to every other randomly generated event in the game. Ironically, there's a lot of stuff in ESO that isn't nearly as random as it looks.

As for why it doesn't get a lot of attention? For the most part it doesn't matter. If you're testing your builds in game, and I mean honestly testing them, not just saying, "well, I put this together on paper so it should be good," you'll see some of this in action. More frustratingly, some of this is per account. So, if you don't run an alt account (and I don't blame you), you'll never see some of this occur.

On the other side, crit results get almost zero attention, because we generate so much data. I'll give T3h some flack some times, but truth is, we generate a lot of data towards crit chance. As a result, we're much less likely to actually scrutinize how it plays out in the moment. (And when you do scrutinize it, the results get really weird sometimes.) The best example of this are the people who have functionally perfected their rotation (or, at least, aren't making any further improvements based on rote action), but do still see fluctuations in DPS output. It's your crit dice rolls being weird. And, yes, there can be other factors, do keep an eye on those.

So is crit chance completely broken? No. It's not. However, as I said, RNG in this game is streaky. People who test know it, people who math (but aren't programmers) don't understand.

Does this matter for this model? Actually, quite a bit.

See, when you crit you deal an extra 50% damage on that tick of damage. Some damage sources can't crit, so they don't matter. However, not all damage sources are equal. A light attack will not deal the same critical damage as a Barricade tick. This becomes very important when trying to "make the crit value real," because you cannot count multiple sources of damage communally.

You can't say, "well, I've got a light attack, I've got some spilled lightning, and I've got a wall of zap out, if one of those ticks my DPS goes up by 16.67%." Depending the source of the crit, the affect of that modifier will be different on your overall damage. When you extract that out into a larger model, where you're looking at things like dropping ults, mixing in light and heavy attacks, multiple DoT sources, and you have a mess that can be modeled, but it's not worth it.

More than that, if you want to, "make the crit real," you need a much longer data set to do that. (Hence, the 15 minutes number earlier, which, ironically lines up with what @T3hasiangod was counting, except he was putting all damage sources into a single slurry, instead of keeping them separate.)

Again, as an overly simplistic example, if you have three damage sources that tick each second, and you take 20 seconds of data, you have three data sets of 20 points each. You do not have one set of 60 data points. To get that 1.2% margin of error he was talking about, you'd need a minimum of 14m50s of data (at two ticks a second, which is only true for some abilities and requires 100% uptime. Not impossible, but not likely in a live scenario.) (Which is why I simply said, "screw it," and called the number at 15m.)

This is made worse if you're looking at Ults that do direct damage, because you can easily end up with a situation where you cannot, fully, "make the crit real," because the fight would need to be comically long to do so.

This isn't saying that crit itself is bad, or that it's unreliable (even though, it can be very fickle at times), simply that trying to project your DPS by estimating what your build and rotation should do, particularly trying to contrast to existing, tested, data, is a bit of a mistake.

If you think Mother's Sorrow should outperform UI on paper, that's great. Nothing wrong with that belief, until you actually try testing it in a live environment. (Assuming you're not intentionally misusing the set to fudge your results, because some people do exactly that in an attempt to say, "hey, look, this guy's actually really bad at the game, this set is terrible and anyone farming it should feel bad.")

Heelie · November 2019

starkerealm wrote: »

T3hasiangod wrote: »

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

How simple do you need this?

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables.

Thing about that is, the drop tables are a really good way to see this in action because players do scrutinize them. The overall output is low, and the results draw direct player attention. So, when you're running vMA for a lightning staff, and have to run the damn thing 300 times, that's not normal. Statistically you should have seen one in the first 30 runs. However, the dice roll that determines your drop is loaded. Not, maliciously, not intentionally, but the result isn't truly random.

Now, I'm using the loot table as an example because it's an easy place to see this in operation. However, the same behavior does apply to every other randomly generated event in the game. Ironically, there's a lot of stuff in ESO that isn't nearly as random as it looks.

As for why it doesn't get a lot of attention? For the most part it doesn't matter. If you're testing your builds in game, and I mean honestly testing them, not just saying, "well, I put this together on paper so it should be good," you'll see some of this in action. More frustratingly, some of this is per account. So, if you don't run an alt account (and I don't blame you), you'll never see some of this occur.

On the other side, crit results get almost zero attention, because we generate so much data. I'll give T3h some flack some times, but truth is, we generate a lot of data towards crit chance. As a result, we're much less likely to actually scrutinize how it plays out in the moment. (And when you do scrutinize it, the results get really weird sometimes.) The best example of this are the people who have functionally perfected their rotation (or, at least, aren't making any further improvements based on rote action), but do still see fluctuations in DPS output. It's your crit dice rolls being weird. And, yes, there can be other factors, do keep an eye on those.

So is crit chance completely broken? No. It's not. However, as I said, RNG in this game is streaky. People who test know it, people who math (but aren't programmers) don't understand.

Does this matter for this model? Actually, quite a bit.

See, when you crit you deal an extra 50% damage on that tick of damage. Some damage sources can't crit, so they don't matter. However, not all damage sources are equal. A light attack will not deal the same critical damage as a Barricade tick. This becomes very important when trying to "make the crit value real," because you cannot count multiple sources of damage communally.

You can't say, "well, I've got a light attack, I've got some spilled lightning, and I've got a wall of zap out, if one of those ticks my DPS goes up by 16.67%." Depending the source of the crit, the affect of that modifier will be different on your overall damage. When you extract that out into a larger model, where you're looking at things like dropping ults, mixing in light and heavy attacks, multiple DoT sources, and you have a mess that can be modeled, but it's not worth it.

More than that, if you want to, "make the crit real," you need a much longer data set to do that. (Hence, the 15 minutes number earlier, which, ironically lines up with what @T3hasiangod was counting, except he was putting all damage sources into a single slurry, instead of keeping them separate.)

Again, as an overly simplistic example, if you have three damage sources that tick each second, and you take 20 seconds of data, you have three data sets of 20 points each. You do not have one set of 60 data points. To get that 1.2% margin of error he was talking about, you'd need a minimum of 14m50s of data (at two ticks a second, which is only true for some abilities and requires 100% uptime. Not impossible, but not likely in a live scenario.) (Which is why I simply said, "screw it," and called the number at 15m.)

This is made worse if you're looking at Ults that do direct damage, because you can easily end up with a situation where you cannot, fully, "make the crit real," because the fight would need to be comically long to do so.

This isn't saying that crit itself is bad, or that it's unreliable (even though, it can be very fickle at times), simply that trying to project your DPS by estimating what your build and rotation should do, particularly trying to contrast to existing, tested, data, is a bit of a mistake.

If you think Mother's Sorrow should outperform UI on paper, that's great. Nothing wrong with that belief, until you actually try testing it in a live environment. (Assuming you're not intentionally misusing the set to fudge your results, because some people do exactly that in an attempt to say, "hey, look, this guy's actually really bad at the game, this set is terrible and anyone farming it should feel bad.")

You're right the programs can never do true RNG but this does not change the fact that mothers sorrow is an amazing set. You see no game with crit chance has crit at the same efficiency value as let's say weapon or spell damage. This means that even when you're unlucky you're still ahead on dps. Also I think you massively overestimate just how impactful this random RNG is. Because looking at 10-15 parses in vAS all faster than 6 mins. You find that they're all within 1-2% of their "true" value. Meaning the set is still stronger than anything else out there. Over the cause of an entire full trial like HOF you will probably see this margin shrink even lower

I have still not seen evidence of actual crit values being somewhere around 5-6% lower or higher than the "true" values. Which is what it would take for something like UI to beat MS.

Also if this was the case. This should be the same case no mather your actual crit value. So you imagine having 60% crit or 72%. Both players should expect the same variation. Making it really an invalid argument against stacking crit sets. If anything you want more to get more value from major force. This is because in higher tiers of play major force uptime can sometimes surpass 50% only further solidifying Mothers Sorrow as eso's tier 1 front bar set.

Also I dont think anyone actually think Xynode is bad at the game. Obviously he's a good player. What people don't like is the fact that he markets his builds as something they're not. If Xynode on his website said that his builds are not competitive to min max trial builds. And not uptimized for beginners noone would complain.

starkerealm · November 2019

Heelie wrote: »

You're right the programs can never do true RNG but this does not change the fact that mothers sorrow is an amazing set.

You may also note that I never said it was a terrible set. In fact, I haven't said much about the set at all, except that it's somewhat obnoxious to obtain. I don't like the set for personal use, but it's not really a component of the mag builds I do run.

I will add, hyperfixation on a single set can lead to tunnel vision. This is an excellent example of that. While MS can be a decent component of a build, particularly one that already has solid flat stats, and reasonable sustain, but suffers from low crit, fixating on it can lead you to thinking it's a "sacred cow," necessary for all builds. Which simply isn't the case. If your build has enough crit that you're comfortable with, you don't need to swing Mother's Sorrow. You can use that 5pc for something more useful for that build.

The most absurd example of this was a player I ran across a couple years ago (I think it was shortly after Dragon Bones dropped) who was running Acuity and Mother's Sorrow, trying to figure out why their damage wasn't there.

It's useful for your current build? Cool, run it.

Do you need it under all circumstances? No.

Heelie wrote: »

Also I dont think anyone actually think Xynode is bad at the game.

There are a lot of people who would be very quick to argue that point.

Heelie wrote: »

Obviously he's a good player. What people don't like is the fact that he markets his builds as something they're not. If Xynode on his website said that his builds are not competitive to min max trial builds. And not uptimized for beginners noone would complain.

The problem with this is the builds are competitive. The issue a lot of people takes is that they're not Alcast builds or some near variation of them. Which does bring us back to MS. Alcast loves that set, which, okay, cool. That's his thing, I respect it, not to my taste, but to each their own.

Xy's builds frequently buck, "conventional wisdom," in the endgame community, and that drives a small, but shockingly vocal chunk of it absolutely bat****.

The part where he's also being featured on the website has to stick in their craw a bit too.

Now, I'll grant you, he's not running trial progressions currently. I suspect a lot of that is because producing content is a full time job. He's said as much many times. That said, he has run endgame raiding in ESO. So, it's not like this is some strange mystery he's never examined, it is something he's done.

karekiz · November 2019

Heelie wrote: »

Also I dont think anyone actually think Xynode is bad at the game. Obviously he's a good player. What people don't like is the fact that he markets his builds as something they're not. If Xynode on his website said that his builds are not competitive to min max trial builds. And not uptimized for beginners noone would complain.

Who cares in the end? If its a *** build that takes a long time to farm, then whatever. It just feels as if both sides takes this personal. I don't buy the whole "we say this for the community!" either.

If people want to criticize Xynode for something legit.

1. That horrid website
2. That horrid intro

For the most part I shifted through a his post <my eyes> and the only two biggest pushes were the quotes: " highest AOE available <Assuming for class>". Which can be debated /shrug.

This is built for players of ALL skill levels from a brand new player all the way up to the PROs. I dunno what exactly a "pro" is, but I am going to assume teh l33t players. I guess you can run it, but /shrug as other said there is probably better.

I don't get the hate the build gets as in the end what <To me a casual player> sounds like is people want him to just copy-pasta Alcast/Liko and repost it as his own. Which to me would be 10X more worse than what hes posted. Its a not super amazing build, and some language is slightly irritating to read, but its harmless in the end.

Basically all he would have to do is add "Non Meta PvE Sorc" to the build, then all complaints would stop?

vesselwiththepestle · November 2019

So how is Farming UI easy compared to just using Julianos or Mothers Sorrow? I Never understood.

Could someone point put some ESO Log Reports with competitive Easy sorcs? Maybe here is a misunderstanding what competitive means.

Also I can't remember Xynode Advertising the EasySorc as competetive.

FrancisCrawford · November 2019

starkerealm wrote: »

T3hasiangod wrote: »

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

How simple do you need this?

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables.

Thing about that is, the drop tables are a really good way to see this in action because players do scrutinize them. The overall output is low, and the results draw direct player attention. So, when you're running vMA for a lightning staff, and have to run the damn thing 300 times, that's not normal. Statistically you should have seen one in the first 30 runs. However, the dice roll that determines your drop is loaded. Not, maliciously, not intentionally, but the result isn't truly random.

Now, I'm using the loot table as an example because it's an easy place to see this in operation. However, the same behavior does apply to every other randomly generated event in the game. Ironically, there's a lot of stuff in ESO that isn't nearly as random as it looks.

As for why it doesn't get a lot of attention? For the most part it doesn't matter. If you're testing your builds in game, and I mean honestly testing them, not just saying, "well, I put this together on paper so it should be good," you'll see some of this in action. More frustratingly, some of this is per account. So, if you don't run an alt account (and I don't blame you), you'll never see some of this occur.

On the other side, crit results get almost zero attention, because we generate so much data. I'll give T3h some flack some times, but truth is, we generate a lot of data towards crit chance. As a result, we're much less likely to actually scrutinize how it plays out in the moment. (And when you do scrutinize it, the results get really weird sometimes.) The best example of this are the people who have functionally perfected their rotation (or, at least, aren't making any further improvements based on rote action), but do still see fluctuations in DPS output. It's your crit dice rolls being weird. And, yes, there can be other factors, do keep an eye on those.

So is crit chance completely broken? No. It's not. However, as I said, RNG in this game is streaky. People who test know it, people who math (but aren't programmers) don't understand.

Does this matter for this model? Actually, quite a bit.

See, when you crit you deal an extra 50% damage on that tick of damage. Some damage sources can't crit, so they don't matter. However, not all damage sources are equal. A light attack will not deal the same critical damage as a Barricade tick. This becomes very important when trying to "make the crit value real," because you cannot count multiple sources of damage communally.

You can't say, "well, I've got a light attack, I've got some spilled lightning, and I've got a wall of zap out, if one of those ticks my DPS goes up by 16.67%." Depending the source of the crit, the affect of that modifier will be different on your overall damage. When you extract that out into a larger model, where you're looking at things like dropping ults, mixing in light and heavy attacks, multiple DoT sources, and you have a mess that can be modeled, but it's not worth it.

More than that, if you want to, "make the crit real," you need a much longer data set to do that. (Hence, the 15 minutes number earlier, which, ironically lines up with what @T3hasiangod was counting, except he was putting all damage sources into a single slurry, instead of keeping them separate.)

Again, as an overly simplistic example, if you have three damage sources that tick each second, and you take 20 seconds of data, you have three data sets of 20 points each. You do not have one set of 60 data points. To get that 1.2% margin of error he was talking about, you'd need a minimum of 14m50s of data (at two ticks a second, which is only true for some abilities and requires 100% uptime. Not impossible, but not likely in a live scenario.) (Which is why I simply said, "screw it," and called the number at 15m.)

This is made worse if you're looking at Ults that do direct damage, because you can easily end up with a situation where you cannot, fully, "make the crit real," because the fight would need to be comically long to do so.

This isn't saying that crit itself is bad, or that it's unreliable (even though, it can be very fickle at times), simply that trying to project your DPS by estimating what your build and rotation should do, particularly trying to contrast to existing, tested, data, is a bit of a mistake.

If you think Mother's Sorrow should outperform UI on paper, that's great. Nothing wrong with that belief, until you actually try testing it in a live environment. (Assuming you're not intentionally misusing the set to fudge your results, because some people do exactly that in an attempt to say, "hey, look, this guy's actually really bad at the game, this set is terrible and anyone farming it should feel bad.")

Wait a moment -- what are you talking about here?

I was under the impression that:

Set procs can't crit.
Anything else can crit.
Anything that can crit has the same crit chance (assuming it's all magicka-based).
Anything that can crit has the same crit bonus (always at least 50%, plus more from CP in any reasonable PvE build, plus any applicable bonuses from Minor/Major Force or class passives).

What did I get wrong?

T3hasiangod · November 2019

FrancisCrawford wrote: »

starkerealm wrote: »

T3hasiangod wrote: »

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

How simple do you need this?

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables.

Thing about that is, the drop tables are a really good way to see this in action because players do scrutinize them. The overall output is low, and the results draw direct player attention. So, when you're running vMA for a lightning staff, and have to run the damn thing 300 times, that's not normal. Statistically you should have seen one in the first 30 runs. However, the dice roll that determines your drop is loaded. Not, maliciously, not intentionally, but the result isn't truly random.

Now, I'm using the loot table as an example because it's an easy place to see this in operation. However, the same behavior does apply to every other randomly generated event in the game. Ironically, there's a lot of stuff in ESO that isn't nearly as random as it looks.

As for why it doesn't get a lot of attention? For the most part it doesn't matter. If you're testing your builds in game, and I mean honestly testing them, not just saying, "well, I put this together on paper so it should be good," you'll see some of this in action. More frustratingly, some of this is per account. So, if you don't run an alt account (and I don't blame you), you'll never see some of this occur.

On the other side, crit results get almost zero attention, because we generate so much data. I'll give T3h some flack some times, but truth is, we generate a lot of data towards crit chance. As a result, we're much less likely to actually scrutinize how it plays out in the moment. (And when you do scrutinize it, the results get really weird sometimes.) The best example of this are the people who have functionally perfected their rotation (or, at least, aren't making any further improvements based on rote action), but do still see fluctuations in DPS output. It's your crit dice rolls being weird. And, yes, there can be other factors, do keep an eye on those.

So is crit chance completely broken? No. It's not. However, as I said, RNG in this game is streaky. People who test know it, people who math (but aren't programmers) don't understand.

Does this matter for this model? Actually, quite a bit.

See, when you crit you deal an extra 50% damage on that tick of damage. Some damage sources can't crit, so they don't matter. However, not all damage sources are equal. A light attack will not deal the same critical damage as a Barricade tick. This becomes very important when trying to "make the crit value real," because you cannot count multiple sources of damage communally.

You can't say, "well, I've got a light attack, I've got some spilled lightning, and I've got a wall of zap out, if one of those ticks my DPS goes up by 16.67%." Depending the source of the crit, the affect of that modifier will be different on your overall damage. When you extract that out into a larger model, where you're looking at things like dropping ults, mixing in light and heavy attacks, multiple DoT sources, and you have a mess that can be modeled, but it's not worth it.

More than that, if you want to, "make the crit real," you need a much longer data set to do that. (Hence, the 15 minutes number earlier, which, ironically lines up with what @T3hasiangod was counting, except he was putting all damage sources into a single slurry, instead of keeping them separate.)

Again, as an overly simplistic example, if you have three damage sources that tick each second, and you take 20 seconds of data, you have three data sets of 20 points each. You do not have one set of 60 data points. To get that 1.2% margin of error he was talking about, you'd need a minimum of 14m50s of data (at two ticks a second, which is only true for some abilities and requires 100% uptime. Not impossible, but not likely in a live scenario.) (Which is why I simply said, "screw it," and called the number at 15m.)

This is made worse if you're looking at Ults that do direct damage, because you can easily end up with a situation where you cannot, fully, "make the crit real," because the fight would need to be comically long to do so.

This isn't saying that crit itself is bad, or that it's unreliable (even though, it can be very fickle at times), simply that trying to project your DPS by estimating what your build and rotation should do, particularly trying to contrast to existing, tested, data, is a bit of a mistake.

If you think Mother's Sorrow should outperform UI on paper, that's great. Nothing wrong with that belief, until you actually try testing it in a live environment. (Assuming you're not intentionally misusing the set to fudge your results, because some people do exactly that in an attempt to say, "hey, look, this guy's actually really bad at the game, this set is terrible and anyone farming it should feel bad.")

Wait a moment -- what are you talking about here?

I was under the impression that:
Set procs can't crit.

Anything else can crit.

Anything that can crit has the same crit chance (assuming it's all magicka-based).

Anything that can crit has the same crit bonus (always at least 50%, plus more from CP in any reasonable PvE build, plus any applicable bonuses from Minor/Major Force or class passives).

What did I get wrong?

You are correct.

Literally nothing he said has been proven to be true at all. Everything he said in that post is pure speculation.

starkerealm · November 2019

FrancisCrawford wrote: »

starkerealm wrote: »

T3hasiangod wrote: »

starkerealm wrote: »

Heelie wrote: »

Do you have evidence of a fundamental issue with crit chance in ESO?

If you've tested anything, you've collected it yourself.

Ah, the classic avoidance strategy. You still haven't answered the question you know. If you know our math must be wrong or faulty, then please tell us where we're wrong.

If you can't, then that speaks volumes.

How simple do you need this?

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables.

Thing about that is, the drop tables are a really good way to see this in action because players do scrutinize them. The overall output is low, and the results draw direct player attention. So, when you're running vMA for a lightning staff, and have to run the damn thing 300 times, that's not normal. Statistically you should have seen one in the first 30 runs. However, the dice roll that determines your drop is loaded. Not, maliciously, not intentionally, but the result isn't truly random.

Now, I'm using the loot table as an example because it's an easy place to see this in operation. However, the same behavior does apply to every other randomly generated event in the game. Ironically, there's a lot of stuff in ESO that isn't nearly as random as it looks.

As for why it doesn't get a lot of attention? For the most part it doesn't matter. If you're testing your builds in game, and I mean honestly testing them, not just saying, "well, I put this together on paper so it should be good," you'll see some of this in action. More frustratingly, some of this is per account. So, if you don't run an alt account (and I don't blame you), you'll never see some of this occur.

On the other side, crit results get almost zero attention, because we generate so much data. I'll give T3h some flack some times, but truth is, we generate a lot of data towards crit chance. As a result, we're much less likely to actually scrutinize how it plays out in the moment. (And when you do scrutinize it, the results get really weird sometimes.) The best example of this are the people who have functionally perfected their rotation (or, at least, aren't making any further improvements based on rote action), but do still see fluctuations in DPS output. It's your crit dice rolls being weird. And, yes, there can be other factors, do keep an eye on those.

So is crit chance completely broken? No. It's not. However, as I said, RNG in this game is streaky. People who test know it, people who math (but aren't programmers) don't understand.

Does this matter for this model? Actually, quite a bit.

See, when you crit you deal an extra 50% damage on that tick of damage. Some damage sources can't crit, so they don't matter. However, not all damage sources are equal. A light attack will not deal the same critical damage as a Barricade tick. This becomes very important when trying to "make the crit value real," because you cannot count multiple sources of damage communally.

You can't say, "well, I've got a light attack, I've got some spilled lightning, and I've got a wall of zap out, if one of those ticks my DPS goes up by 16.67%." Depending the source of the crit, the affect of that modifier will be different on your overall damage. When you extract that out into a larger model, where you're looking at things like dropping ults, mixing in light and heavy attacks, multiple DoT sources, and you have a mess that can be modeled, but it's not worth it.

More than that, if you want to, "make the crit real," you need a much longer data set to do that. (Hence, the 15 minutes number earlier, which, ironically lines up with what @T3hasiangod was counting, except he was putting all damage sources into a single slurry, instead of keeping them separate.)

Again, as an overly simplistic example, if you have three damage sources that tick each second, and you take 20 seconds of data, you have three data sets of 20 points each. You do not have one set of 60 data points. To get that 1.2% margin of error he was talking about, you'd need a minimum of 14m50s of data (at two ticks a second, which is only true for some abilities and requires 100% uptime. Not impossible, but not likely in a live scenario.) (Which is why I simply said, "screw it," and called the number at 15m.)

This is made worse if you're looking at Ults that do direct damage, because you can easily end up with a situation where you cannot, fully, "make the crit real," because the fight would need to be comically long to do so.

This isn't saying that crit itself is bad, or that it's unreliable (even though, it can be very fickle at times), simply that trying to project your DPS by estimating what your build and rotation should do, particularly trying to contrast to existing, tested, data, is a bit of a mistake.

If you think Mother's Sorrow should outperform UI on paper, that's great. Nothing wrong with that belief, until you actually try testing it in a live environment. (Assuming you're not intentionally misusing the set to fudge your results, because some people do exactly that in an attempt to say, "hey, look, this guy's actually really bad at the game, this set is terrible and anyone farming it should feel bad.")

Wait a moment -- what are you talking about here?

I was under the impression that:
Set procs can't crit.

Anything else can crit.

There's a few other exceptions mixed in, but that's more "trivia question territory" than anything to actually worry about.

FrancisCrawford wrote: »

Anything that can crit has the same crit chance (assuming it's all magicka-based).

Anything that can crit has the same crit bonus (always at least 50%, plus more from CP in any reasonable PvE build, plus any applicable bonuses from Minor/Major Force or class passives).

What did I get wrong?

Nothing. You're not incorrect (aside from missing some trivia), but you're kinda missing a critical context for the data being collected and how it's being presented.

When a damage tick crits, that specific damage tick is increased by 50% (+ whatever other modifiers apply to the build's crit severity.) So, for example: if you crit with Wall of Force it doesn't increase your Force Pulse damage. This seems pretty obvious, until you see someone trying to model their crit chance as a flat damage increase across the board to their entire build.

Now, technically, you can (kinda) do that, in extremely long fights, (+15m, though really this should be over half an hour if we're looking at light attack damage), but they're lumping every damage tick together, regardless of source, into the data pool and saying it happens much sooner (around the 4 minute mark), which isn't how crit chance works.

So, like I said, you didn't get anything wrong, but you forgot that each damage source crits independently, and does not affect the damage of your other abilities.

Katlefiya · November 2019

starkerealm wrote: »

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables

These perceived "strange anomalies" are no proof that there ist something inherently wrong with ESO's random number generator. Consider a coin that landed heads every time you flipped it. Would you eventually conclude the coin was crooked?
Of course you would! On the other hand, you’d probably find nothing wrong if you saw some other sequence where you can't immediately recognize a pattern. The paradox is that a fair coin is exactly as likely to produce this sequence as it is to produce fifty heads in a row.

ESO being what it is, namely a computer program, it's random number generator is certainly not a perfectly random sequence of tosses with a fair coin. But is it good enough? Maybe. Probably. Who can tell?

The obvious way to check whether a sequence is random is just to throw some statistical tests at it. But no matter how clever you are in devising a statistical test for randomness, an adversary will always be able to find a deterministic sequence that passes your test.

starkerealm · November 2019

Katlefiya wrote: »

starkerealm wrote: »

As it turns out, computers are really bad at creating random numbers. The solution is to generate semi-random "seeds" used to generate a random number. You can use a lot of factors, like the CPU clock, or a rotating variable to generate a seed, and the result will be, generally indistinguishable from a truly random number.

Problem is, ESO doesn't seem to do that. It's very difficult to pin down exactly how the seeds are generated, but they're pulling from some persistent value, somewhere. This is why most players experience very strange anomalies in the drop tables

These perceived "strange anomalies" are no proof that there ist something inherently wrong with ESO's random number generator. Consider a coin that landed heads every time you flipped it. Would you eventually conclude the coin was crooked?
Of course you would! On the other hand, you’d probably find nothing wrong if you saw some other sequence where you can't immediately recognize a pattern. The paradox is that a fair coin is exactly as likely to produce this sequence as it is to produce fifty heads in a row.

Let's chase your thought experiment for a second. The odds of flipping the same coin face 50 times in a row is slightly less than one in one quadrillion. The same as any other specific combination of flipping a coin 50 times.

So, before you get confused, the odds of each coin flip landing in a certain way is 50/50, and I can see the gears turning, "so clearly, the chances of flipping 50 coins and getting the same result will be 50/50," but, that's not true. Flip a coin twice, and there's only a 25% chance that you'll get heads both times. This is because you have four potential outcomes which are all equally likely (00/01/10/11), and you're only interested in one of those. If you said, "get one heads and one tails," then your odds would be 50%, because two of the four potential outcomes satisfy your goals (01/10). If you simply said, "the same result twice," then, again, your odds are 50% (00/11). Though if you're saying, "first flip heads, second flip tails," you're back to 25%.

There's a 50% chance that each individual data point will land where you want it, but you're not interested in that, you're looking for everything to line up. So, you add a third coin, and now there are three data points that need to behave, which means there are now 8 possible outcomes (000/001/010/011/100/101/110/111). If you want 1 heads and two tails, that's a 37.5% chance, because it can occur in three of the eight potential states. However, if you want all heads, then that's now only a 12.5% chance.

I'll spare you a longer and more tedious discussion of this, but if you want a takeaway from this, it's simple, getting the result you want from 50 coin flips is extremely unlikely.

Now, let's climb back out of the weeds.

Katlefiya wrote: »

These perceived "strange anomalies" are no proof that there ist something inherently wrong with ESO's random number generator.

This behavior is extremely well documented in the community. If you believe that the random number generator used to roll crits operates off an entirely unique architecture from the generators used throughout the rest of the game, that is exceedingly unlikely.

Katlefiya wrote: »

ESO being what it is, namely a computer program, it's random number generator is certainly not a perfectly random sequence of tosses with a fair coin. But is it good enough? Maybe. Probably. Who can tell?

Pretty much anyone with a background in programming.

This is one of those moments when, if you have the background, you can instantly recognize what's happening. Climb out of the segment of the community overly obsessed with charting out their builds in Excel, and you'll quickly learn that what I've said about RNG in ESO isn't even mildly controversial.

Katlefiya wrote: »

The obvious way to check whether a sequence is random is just to throw some statistical tests at it. But no matter how clever you are in devising a statistical test for randomness, an adversary will always be able to find a deterministic sequence that passes your test.

This makes it sound like you believe that the game is sentient and is intentionally trying to confound your analysis.

T3hasiangod · November 2019

I still see no evidence of this loaded crit chance or different RNG for crit that you are claiming exists.

Tons of end-game PvE players who have tested builds and sets for probably tens of thousands of hours combined have not noticed this.

What makes you think you're so special?

None of what you have said has been proven with factual evidence. Nobody except you seems to prescribe to this idea that each ability essentially has its own RNG counter.

Unless you are the coder of the crit RNG used in this game, everything you have said is pure speculation and should be treated as such.

starkerealm · November 2019

T3hasiangod wrote: »

I still see no evidence of this loaded crit chance or different RNG for crit that you are claiming exists.

That's because it doesn't happen in Excel.

Heelie · November 2019

starkerealm wrote: »

T3hasiangod wrote: »

I still see no evidence of this loaded crit chance or different RNG for crit that you are claiming exists.

That's because it doesn't happen in Excel.

I still ask the question, why do we see no evidence of this in our parses? And also why would make make a difference if stacking crit or not?

Katlefiya · November 2019

starkerealm wrote: »

I'll spare you a longer and more tedious discussion of this, but if you want a takeaway from this, it's simple, getting the result you want from 50 coin flips is extremely unlikely.

Oh dear, are you really trying to lecture me on the probabillity of coin tosses? Of course it is extremly unlikely, I never said otherwise. But my point was something else: that a fair coin is exactly as likely to produce this sequence as it is to produce any other possible sequence. Perceived "anomalies" in itself are no proof that ESO's RNG is comparatively bad.

Can you ever be reasonably sure that something is not random, in the same sense you can be reasonably sure something is random? Even if a sequence looked random, how could you ever rule out the possibility that it had a hidden deterministic pattern?

starkerealm wrote: »

Katlefiya wrote: »

The obvious way to check whether a sequence is random is just to throw some statistical tests at it. But no matter how clever you are in devising a statistical test for randomness, an adversary will always be able to find a deterministic sequence that passes your test.

This makes it sound like you believe that the game is sentient and is intentionally trying to confound your analysis.

Don't be silly, I was not hinting at the game being the adversary, but that any attempt to check for randomness is difficult if not impossible. You might have heard of the Kolmogorov complexity, which is the measure of the inherent randomness of a sequence. Ironically, the Kolmogorov complexity is impossible to compute.

starkerealm · November 2019

Katlefiya wrote: »

starkerealm wrote: »

I'll spare you a longer and more tedious discussion of this, but if you want a takeaway from this, it's simple, getting the result you want from 50 coin flips is extremely unlikely.

Oh dear, are you really trying to lecture me on the probabillity of coin tosses?

Of course not, I'm actually explaining it to you. If you didn't want to have your ill fitting analogy dissected, maybe you should have found a better one.

Katlefiya wrote: »

Perceived "anomalies" in itself are no proof that ESO's RNG is comparatively bad.

I did not say, "bad," I said, "streaky." Moreover, I said all ESO RNG is streaky. Which, if you play, isn't really new information.

Katlefiya wrote: »

starkerealm wrote: »

Katlefiya wrote: »

The obvious way to check whether a sequence is random is just to throw some statistical tests at it. But no matter how clever you are in devising a statistical test for randomness, an adversary will always be able to find a deterministic sequence that passes your test.

This makes it sound like you believe that the game is sentient and is intentionally trying to confound your analysis.

Don't be silly, I was not hinting at the game being the adversary, but that any attempt to check for randomness is difficult if not impossible. You might have heard of the Kolmogorov complexity, which is the measure of the inherent randomness of a sequence. Ironically, the Kolmogorov complexity is impossible to compute.

I think we're talking about something completely different here. The Kolmogorov Complexity I know has to do with the amount of code you need to generate an object, and how the code that goes into it will (almost always) be larger than the object itself. Which, I mean, it's true, but not particularly relevant to the discussion at hand. I suspect you're thinking about something else.

starkerealm · November 2019

Heelie wrote: »

starkerealm wrote: »

T3hasiangod wrote: »

I still see no evidence of this loaded crit chance or different RNG for crit that you are claiming exists.

That's because it doesn't happen in Excel.

I still ask the question, why do we see no evidence of this in our parses?

In your case? Because you're a healer main. Probably the same answer for Asian. As a tank main, crits don't even exist in his world.

Heelie wrote: »

And also why would make make a difference if stacking crit or not?

It's more accurate to say it matters when analyzing and prototyping builds. If you're going to compare two builds, you need to do it in game. You also need to actually parse those builds out.

It also becomes significant when people try to use crit chance as a fixed value. Simply going DPS=PrecritDPS+(PreCritDPS*(CritChance*CritSeverity)) doesn't work. It won't reflect what will happen in game.

I get why people want to use calculation models. It's easier to say, "well, I have these tools, so I can take player skill out of the equation." Because you're never going to be able to control for that. It also becomes less threatening for some, genuinely trash DPS, to present their DPS builds, while being able to hide behind, "this is what the build can do," without having to demonstrate that in game (or even being able to.) "But, no you don't understand, I crunched the numbers."

It is also attractive to be able to say, "I can force crit chance to be a fixed frequency." It makes modeling things a lot easier. The problem is, when a simple AoE tick crits, it will be less damage than when your ult crits. So, you can't really take the sum of your DPS and then multiply it by your crit chance, because it's including a mix of different sources, each of which will crit with a distinct value.

Here's the part where it gets annoying, if your crit starts weighing heavily on one damage source over another (and this can happen), RIP modeling this mess.

So, yes, stack crit, it's good. I run Stam, I'd need to be a huge hypocrite, and a moron, to say avoid this. But, when you're looking at what your crit will do, you need to check it personally. You can't simply count on, "making it real," because it will never, fully, cooperate.

Bobby_V_Rockit · November 2019

Enough maaaaaaaths lol

FrancisCrawford · November 2019

What exactly is being claimed to be nonrandom in practice here?

If I have a tooltip crit chance of 60%, for example, I would expect each skill or basic attack kind I have to crit 60% or so of the time, when averaged across multiple fights, parses, whatever, with the likelihood of a large percentage deviation from the expectation decreasing as the number of trials goes up. I'd further expect the likelihood of "hot" or "cold" streaks to be more or less in line with the usual binomial distribution math.

If ESO's "RNG" delivers something close to that, it's doing well. If it doesn't, it's doing badly.

I gather it's being accused of doing badly -- in what way? Actual long-term deviation from the putative average? Too many streaks?

starkerealm · November 2019

FrancisCrawford wrote: »

What exactly is being claimed to be nonrandom in practice here?

The issue isn't about something being nonrandom, simply there are issues which arise from attempting to model data outside of the game.

T3hasiangod · November 2019

FrancisCrawford wrote: »

What exactly is being claimed to be nonrandom in practice here?

If I have a tooltip crit chance of 60%, for example, I would expect each skill or basic attack kind I have to crit 60% or so of the time, when averaged across multiple fights, parses, whatever, with the likelihood of a large percentage deviation from the expectation decreasing as the number of trials goes up. I'd further expect the likelihood of "hot" or "cold" streaks to be more or less in line with the usual binomial distribution math.

If ESO's "RNG" delivers something close to that, it's doing well. If it doesn't, it's doing badly.

I gather it's being accused of doing badly -- in what way? Actual long-term deviation from the putative average? Too many streaks?

The claim is that you can't model crit chance over extended parses by using a binomial distribution, despite the fact that crit chance is a binomial function.

Because of reasons that have yet to be explained or expanded upon.