Sentinel fixes

edited March 2021 in Sentinel

Just wanted to post some thoughts on Sentinel in its current state following the changes which were necessitated by the y852 games.


Petrify

Petrify strat is far too powerful with lemming/vertigo in its current state.


Previously, your choices for 2 passive pets were:

 - goldenseal cover aff + no fly (butterfly/lemming)

 - petrify aff + no fly (raven/lemming)

 - petrify aff + goldenseal cover aff (raven / butterfly)

Depending on the strat you were pushing and the target's affliction state, different choices were valuable in different ways. E.g. if going for petrify, raven was necessary and a choice would be made between butterfly and lemming based on whether you were indoors, the target's propensity for flying etc.

There was also a cost in changing between strats, as pets are passive when summoned and cant be ordered to attack until the next balance.


The current state is:

 - goldenseal cover aff + petrify aff + no fly (butterfly/lemming) is optimal in all situations except where damage is being pushed on a prone/locked opponent.

Regardless of the strat you're pushing (lock, limb prep, damaging aff stack, petrify), assuming you're using doublestrike to contribute aff pressure, there is now no variation in optimal pet selection.

Suggested fix: Remove vertigo from list of petrify affs, add paranoia to list of petrify affs.


Wrench

I dont understand the rationale behind the most recent change, but must admit I never saw the single leg wrench kill that I think(?) gave rise to it.

The change (extended wrench balance) has stopped sentinels being able to wrench twice on a target who was impaled when they had 2 damaged legs.  Although the writhe time is still extended, it is no longer enough to allow 2 wrenches.  This is a retrograde change as breaking x/leg/leg -> impale/wrench/wrench -> more rubbish -> dismember was a useful damage strategy against people who would focus on parrying head and left both legs open.  This strategy is no longer viable.

Even if the second wrench were possible, the additional time for 2xwrench gives the target more time to squeeze another sip/moss in before the eventual rattle in the dismember chain.

Suggested fix: Increase the additional writhe time for impale with 2 broken legs so that impale + 2xwrench can be accommodated. Shorten the balance of ensnare by 2xchange to wrench so that the total time from impale -> rattle is unchanged.

Comments

  • A Dissertation on Sentinel - Strap in kids, this is going to be a LONG one. I've been formulating my thoughts on the class in response to a number of nerfs that have gone in in recent weeks. It's caused me to reconsider playing the class at all because much of what made the class fun and unique was stripped out. Going to spend some time talking about the impact of the different nerfs and my thoughts on each.


    Trumpet :

    The Problem - Leading up to the games, people complained about trumpet's use as an opener (mostly in groups), which basically allowed it to be used as an instant prone disrupt when combined with enrage raven to undeaf. It probably didn't change this much since we have other class based openers that enable impale such as hydra roar/bind.

    It's intended use was meant to add disrupt to locks to enable truelock in solo play. We have another means to truelock by giving chills, but it doesn't make sense to use it because we can kill with skullbash/wrench instead. It was also a relatively useful anti-air tool since it allowed you to regain balance fast enough to stop someone from flying again most of the time if they had confusion.

    The Nerf - Trumpet no longer gives disrupt (still prones), and gives sensi and a 1s stun against an already prone target.

    Why the nerf causes more problems than it fixes - There is literally no reason to use trumpet against a prone target. None, zero. We can apply sensitivity passively with raven and quickly with handaxe.

    Suggested change - Make it so that trumpet gives addiction and prone when used on an undeaf target. Does nothing additional against prone target, or maybe keep stun, but still it's kind of pointless. Reduce the balance cost to around 2s before nimble to make it a viable alternative to tentacle tattoo. It's current niche is only for unparryable prone against a broken limb. It's about .1s longer than trip without the risk of being parried.


    I'm going to address the butterfly transfix and wrench nerfs together since they both seemed to target single leg kill strats that I think pissed off all of Mhaldor which ultimately led to the slew of classlead submissions that put us in the situation we're in now. Before the changes, if you pre stripped blindness you could impale and transfix together against a valid target. This created scenarios where people complained that you could effectively chain multiple writhes into one another creating a never ending loop of writhes. While i've never seen this situation actually play out, it was very effective and difficult to avoid unless you just had an absolute ton of health, like proficy who had vitality as a runewarden or classes with passive writhe. There were other counters however, such as prioritizing haemophilia and blindness while prone to avoid taking damage due to the bleed and prevent additional transfixes. Even despite my having argued that fact, they came hard for this strategy which was disappointing considering it was still fairly challenging to pull off against people with random leg parry strategies since it relied wholly on which afflictions/defences your enemy had going into the break sequence.

    Wrench - got extended so you can't double wrench (likely unless you break with trip). With the current state of limbs, it may be slightly more reliable to be able to trip break(more below), but it's slightly more reliable with an ability that's already pretty unreliable. In it's current state against a portion of people you cannot single trip break them. Double trip breaking is bad because you give up a lot of momentum and you give your enemy an obvious 'tell' about what limb you're trying to break. In almost all situations, trumpet is a better (my opinion) prone ability after you break with handaxe. Breaking with spear allows them too much time towards healing the level2 break with resto and gives you time to prone them with a small window allowing you to prone them and attack again before they can stand.

    Butterfly Transfix - I understand why this got changed. It was very strong in it's former iteration. People complained for weeks about the ability to impale and transfix on the same balance. Again though, the above curing counters and the inability to enrage while wrenching made it survivable, just not likely on stock priorities since you're probably going to be trying to cure other things than haemo which is what you'd want to be doing to survive.


    Handaxe limb damage - They called it a bug, but basically it's another nerf that fixed a problem, but did not address another problem. I'll explain further. Sentinel was one of a few classes that had 3 different limb damage variants.

    Handaxe was the most potent and fastest, bypasses parry but with no parry bypass. The tradeoff (which may have also been another bug was that you lost some affliction since enrage does not line up with level 3 handaxe speed). Currently, handaxe has been standardized to the spear strike damage amounts as best as I can tell.

    Spear strikes did less than handaxe and were vulnerable to rebounding (which could be bypassed by lemming enrage) but had a bypass of haemophilia (useful only reliable with damage strats since haemo doesn't really help petrify or lock)

    Trip does half of spear damage, does respect a haemophilia bypass (again only really reliable with damage strats).


    Other problems with the class:

    Gaian Axe-Enrage line up issues - like i mentioned before. Enrage timer does not line up with level 3 handaxe, so you get situations in which you're only able to enrage on every other handaxe, which wouldn't be a problem except that it affects situations in which you attack with any other weapon as well. So i could enrage on a handaxe, and have to wait until after a longer spear strike (2.2ish) before i could enrage again. This usually doesn't cause too many problems except that it created some weird scenarios where you'd have to arbitrarily wait another round or three if you ended up on a weird timing.


    Multi-trip shenanigans - This was a bigger problem when juggling 3 different limb damages, now that spear and handaxe have been standardized it's better. It removes situations in which you would have to go through weird prep gymnastics to trip break. Let me explain.

    Target breaks on 6 handaxe hits, but to trip break them you have to handaxe them 5 times, spear strike them once, and then trip break -or- handaxe them 5 times and then trip twice to break. There is no scenario in which breaking with a spear strike is optimal since the time it takes to regain balance botches the kill.

    With the current situation the same target takes 8 handaxes to break, or 7 handaxes and 2 trips. On about half (i think) of targets you will encounter this scenario where you will have to break with a handaxe, prone on next balance, and then proceed to kill, or be in a situation in which you have to trip twice (also not good since you lose aff momentum).


    Dismember takes too many steps - This was discussed ad nauseum in the ACC and has been made even worse with the new changes. While there were situations before where you could dismember against 1 leg if you had the right afflictions going into the kill sequence. In it's current form it requires 2 limbs to pull off reliably (1 dummy limb, and a leg). Since tertiary transfix was removed this becomes further exacerbated since our only source of transfix is now ensnare. In the current state of gameplay, 95% of engagements are probably going to be group engagements. There are few scenarios where you will find going through the sequence to dismember will be undisrupted enough to be able to pull it off. It's more reliable to focus on damage and ignore the extra steps of truss and impale. Rattle can be useful to delay sips but currently does not stop a tumble, and only respects the owner of the transfix. It's cumbersome and generally unusable even for style points in it's current state since damage would be preferred in just about any conceivable scenario.


    The addition of vertigo to the petrify list - this is problematic, yes. The reason I even suggested this classlead was because i knew that while Sentinels already struggled with arial counterplay, that the changes to trumpet would undoubtedly make things worse. Which they did. Trumpet pre-nerf was a tactically interesting ability since it had the pre-requisite of deafness, and could be really good anti-air against people with confusion. I think maybe removing the disrupt against people on the ground while still affecting air (as opposed to altogether) would have been a better change. As it stands vertigo does stand to make the untargetted petrify route even more potent, something experienced users of the class aren't super interested in.

    Unrelated changes probably motivated by other classes like shaman's low hinder from dizziness motivated the improvement from 20-25%. It does benefit sentinels substantially.


    Traps - they're strong, they're also very unwieldly. They take forever to set up to prep a room, they're single use and don't make sense to retrap in the middle of a fight. There have been a lot of suggestions about how to change them to make them more usable, but without that, they're very niche and often not worth the time they take to lay down.


    I love this class. When I first played it it was super complicated, but once i figured out the limb damage aspect of Sentinel, the possibilities opened up. I feel that the majority of people underappreciate the amount of finesse and coding chops it takes to make this class successful outside of untargetted petrify, which i believe is the LEAST interesting, yet very powerful use of the class. I feel as though some of the nerfs that came were (likely) targetted at me personally because people felt threatened that I would be a bigger contender in the championship games. I say that because i was likely the only or one of the only people utilizing the single leg kill strats that were focused on for nerfing. The limb damage hybrid strats were what made the class very interesting and cool, alongside situational tools like trumpet for anti-air. Gimping these strats and buffing non-limb approaches feel like a huge step back.


    At any rate, Mhaldor made it clear that they wanted to lock me out of even contending, just seeing how had they went after me in groups of 5 and more during the experience event. As salty as I was at the time, weaponizing classleads with the intent of disadvantaging a player or group of players shouldn't have been enabled by the administration consciously or not. I never thought that I would win the staff, and Mhaldor was much better organized in their efforts overall for that to be a possibility, but I can't help but feel like this was something very personal on their part. I'm not one for conspiracy theories but, if Mhaldor has a hate boner for anyone, it's me and it's Venari, mostly driven by Proficy's sore ego and rage to dying to us 1v1. So much so that these 3-4v1 man group ganks have continued even after the championship events have ended.


    TLDR (f.u. Archaeon): Sentinel has great bones but the rage nerfing has to stop cause you're breaking what makes the class really fun and interesting.

  • ArchaeonArchaeon Ur mums house lol

    yeah i agree, but it seems that new release classes have fast ttk's and old ones are getting re-done to have faster ttk's, which makes me think that ttk in general is going to get reduced down. If that is the case, certainly it would be better to test EVERY class on a test server, then release them all at once (hopefully with some RP stuff involved!) as not to cause this massive disparity in between, say, an experienced SNB paladin and some dude who can lolpetrify in 4 balances WHO DIDNT EVEN KNOW HIS TORSO WAS BROKEN DURING A DSB AND ONLY LIVED BECAUSE HE HAD 2 TRUEFAVORS REEEE




    anyways, I digress. Sentinel strong. Suggested change:


    If raven is summoned, give it a 50% chance to poop on the petrified person, which will then heal them.

  • I'm not going to quote the posts since they're long, but:


    Lyrin:


    The passive selection decision making issue here is a fair concern. I will think about that change some more - good points here.


    Wrench should still be workable to get two off two legs. It was during my testing, so there might be something off here - I'll investigate further.


    Andraste:


    I think a problem here is you are looking at these in the context of the games. These complaints were ending up in my innbox long before the games got close, and not just from Mhaldorians. Its also something that has been on my own radar for a long time.


    Trumpet: stun/disrupt openers are typically just not something I want in the game. I will almost always take an opportunity to phase these out: some people will disagree with that, but that is an active design decision and not one I anticipate changing. That said, I do like the idea of trumpet disrupting a flying target still - that might be a good compromise here in light of the points raised by Lyrin and any changes those necessitate.


    Butterfly transfix: these were almost entirely a team impact decision on my part. I received a lot of one v one logs with them but I didn't find it particularly significant or compelling in this case - a lot of it was down to poor reactions on the transfixee's part. However, putting in a tertiary transfix was a mistake in the context of teams I am amazed has not manifested earlier: that's my error and there was no way this could reallistically stay, given the rise of more and more combat scenarios being team based. Writhes are exceptionally potent team tools and we do need to be very careful with how they're assigned in the modern meta.


    Wrench: I'll not comment on this one further until I can determine why live results are different to my test ones.


    Handaxe: this definitely was a bug. Essentially targeting limbs carried a limb damage penalty but because of how throwing works inline throwing was not applying the same penalty. I won't comment on whether the new standardised value should be raised, I think its too early to say that with any degree of certainty. I am open to discussions on that though - the reason this one waited until after the games ended was because I do think we might need some adjustment.


    I don't want to talk about limb damage too much in terms of difficulty lining up breaks because we're going to talk about a rework in general to make this less of a headache across all classes. That is high on the list now that the games have ended, so I will pass on those points, though I do hear them.


    As for weaponising classleads, I don't think this was the case. You are entitled to your opinion but my bs detector is pretty finely tuned and I found most of the approved classleads put forward to be reasonable: the ones I did not got rejected. I can't really give you more than that in that respect, sometimes people are just going to disagree with my decisions.

  • edited March 2021

    Thanks for reading and commenting Makarios.

    I tested 2xleg impale a couple of times (with queueing) and had some tests regain balance from the first wrench before writhe completed and some not. Balance time randomness might explain why results differ.

    @Andraste Just something on handaxes - I have been using the new, lower values my entire time as sentinel because I gathered limb damage data before the inline version was created and was too lazy to change. I first tried out inline when the alias command limit was introduced, saw they were different and promptly switched back to the lower limb damage.

    I've never really found that the ~20% difference in axe damage has been a problem, and willingly give this up for the benefit of simpler limb prep paths (axe = spear = 2*trip). I think the axe fix is an improvement even if it looks numerically worse! In hindsight mangling limbs probably would have been viable with inline axeing, but I don't think sentinel really needed that.

  • edited March 2021

    Weaponizing classleads is not reality. Things may seem targetted at you if you are the person using them, but they aren't. Typically nerfs happen to a class/situational mechanic that isn't quite intended or out of whack with other metrics of what is acceptable in terms of power. Makarios already detailed why he made his decision, and it's largely the same reason that I supported those leads too. Tertiary transfix is extremely strong and it's not really something you can just 'deal with' in reality. You said prio blind/haemo while prone, and that just never would have worked. You're operating from the assumption that the transfix was the reason for the nerf, when it wasn't. It's what that transfix/prone situation let other classes do, that made it problematic. It opened up unavoidable prone damage sequences by virtue of how the affliction operates. Even more so when it was combined with multiple entangles like impale or entangled from Druid, and you could do it from 0 as an opener. That was never going to stick around.

    If anything, you should take pride in getting a class nerfed. That means you took something and proved that it was stronger than initially thought, or that something was missed in testing.





    Penwize has cowardly forfeited the challenge to mortal combat issued by Atalkez.
  • I agree with Atalkez. The prone disrupt transfix entangle was wayyyyy too much to have keeping around. Same with the Paranoia/Vertigo pair, that was wayyy too good as well.


    Going forward with these new changes. It's just going to take time to figure out what to change your set ups to. Dismember is still completely possible. Though. I would like to voice my strong distaste for how many steps it takes to get to dismember. Why so many steps 😥

  • @Makarios

    I don't want to talk about limb damage too much in terms of difficulty lining up breaks because we're going to talk about a rework in general to make this less of a headache across all classes. That is high on the list now that the games have ended, so I will pass on those points, though I do hear them.

    You know, all you really need to do to clear up limb damage is release the formulas / base values. While it may be dense mathematically (or not) I am sure it will be coded into a publicly accessible system almost immediately.

    I do think there are multiple ways it could be "simplified" without impacting balance very much, but it really wouldn't be necessary if we just knew how the game works, currently. It's already pretty simple to get a crude formula (not perfect but works within +/- 10 points of damage for each attack anyways, so what would really help is just to release the actual values and formulas across the board so we can take the mystery out of it all.

    At the moment we're still gating reliability of limb damage attacks behind arbitrary mysteries like "Does the game use floor(), ceil() or round()? Does it use it on each operation or just the final one? Does the game use actual max HP or some retro version from before max-hp got reworked?". We don't need to answer those on a per-attack basis, we can just work out approximation formulas that are maybe 99.5% accurate, but what good does it do to require this? As far as I can tell it's just a massive barrier to entry for newer players or even intermediate players who don't have literally weeks of time to spend testing and doing algebra, or buying multiple levels of artefact weapons purely to derive damage stat placement(s) in the formulae.

    I could definitely commit to releasing a publicly available and supported tracking system (on Mudlet) with no restrictions by org or anything like that for applying this in a user-friendly, simple way - if it were released. I think the only drawback to this is that if the formulas are changed, it should be communicated (as in the new formula) instead of the cryptic "limb damage was decreased significantly (for some people, sometimes)" type of update that can throw out weeks of work and make us start all over again each time.

  • Huge disclaimer - I'm a pretty poor combatant. Frankly I aspire to be a midling fighter at this point, but...

    Please don't release any formulas. (1) Half the fun of fighting to someone like me is figuring it out. When I figure out how to make something work, it's a big win moment for me. If the formulas are released it'll become nothing but a numbers/math game. I'm sure it already is to some people, but releasing the formulas would just make that worse. (2) Even if someone releases a script to everyone out there to take advantage of those formulas, it makes the problem of being a math game even worse and creates a handicap to anyone not using mudlet. Release a port for nexus? It's still a disadvantage to people who use something else. Yes, most combatants use mudlet, and between nexus and mudlet the vast majority of players are covered, but it's still not equitable to everyone.

    Please, keep the mystery. Let us midlings and midling hopefuls keep our fun too.

    Argwin

  • Well that's a take I've never hear before.

    Intentionally not knowing things is still an option even if the information is there. Like if you don't want to know what venoms do, just don't read Help Venom - but hardly a good reason to deny the info about what venoms to the rest of the community.

  • ArchaeonArchaeon Ur mums house lol

    Releasing limb formulas would drastically alter combat for knights, monks, bards, etc. Simplification would be better, but outright knowledge would be a terrible choice.

  • edited March 2021

    I agree with both releasing formula and simplifying of limb counting. My only concern with simplifying limb counting is when things are fundamentally changed, it can render certain tactics useless. Some of these tactics are less common, hence might not be taken into account during release of simplified limb counting. I feel like the game has generally be dumbed down a lot, skill cap is coming down drastically, I am sure this benefits the overall population, but again, having a high skill cap can be very attractive to certain segment or level of players. I hope complexity is simplified but continue to allow glaring skill differences that isn't just situational awareness.

    We want to be able to continue to say "OMG, Lyrin is good!" and not "OMG, sentinel is oppresive!".

  • ArchaeonArchaeon Ur mums house lol

    Lyrin sucks tbh. Total scrub

  • edited March 2021

    I give Lyrin merit to the content is this thread that he has created. He has come from great understanding of Sentinel and at the same time remain impartial with his suggestions. I find it productive and helpful to the game.

    Then again, I am also trying to illustrate that i find it more preferable to see credit given to great players of a class, than seeing general sentiment and trend crediting a class's advantage, than acknowledging a great player.

  • ArchaeonArchaeon Ur mums house lol

    I think the general problem is there's a gap between new and old. A big one. New classes greatly outpace the old classes. New also counts as reworked.

  • Hello everyone that's following this, or ends up reading this. The latest class change as of me posting this to Sentinel was reverting Vertigo back to Paranoia. This is my response to that.

    Is the effect of vertigo the only reason for the change, or is the speed in which it hits that is effecting the change? I did a 5 minute test with my raven and lemming both passively hitting someone. This test ended with Paranoia hitting 21 times in 5 minutes, and Vertigo hitting 46 times in the same 5 minutes. My though on this is, if the change from Vertigo to Paranoia was simply the effect of Vertigo on people, can we bring the passive affliction speed of the raven inline to the passive affliction speed of the lemming?

    5 minute test - Vertigo: 46 / Paranoia: 21

    Thoughts? Comments? Complaints? What do you guys think?

  • Hello everyone that's following this, or ends up reading this. The latest class change as of me posting this to Sentinel was reverting Vertigo back to Paranoia. This is my response to that.

    Is the effect of vertigo the only reason for the change, or is the speed in which it hits that is effecting the change? I did a 5 minute test with my raven and lemming both passively hitting someone. This test ended with Paranoia hitting 21 times in 5 minutes, and Vertigo hitting 46 times in the same 5 minutes. My though on this is, if the change from Vertigo to Paranoia was simply the effect of Vertigo on people, can we bring the passive affliction speed of the raven inline to the passive affliction speed of the lemming?

    5 minute test - Vertigo: 46 / Paranoia: 21

    Thoughts? Comments? Complaints? What do you guys think?

  • edited March 2021

    I'm fairly sure all pets have the same affliction "speed", but its not exactly a consistent thing.

    They act along the lines of "Every few seconds, they have a chance to give an affliction. Any time they don't give an affliction on this tick, the chance they will next tick goes up. Until they give an affliction, which it resets back down to the original chance."

    I think you'd need a very large pool to accurately measure different animals, and you'd want to measure average time between ticks.

    It's worth mentioning that the previous enrage change, making it no longer dismiss animals, increased the passive aff speed by a good chunk over what it used to be. Dismissing and resummoning an animal would reset it to the original fire chance, but currently both passives get to go at full speed.

  • I bring data and graph!

    First, pretty graph:

    Figure 1. cumulative probability graph of the time between procs for lemming (brown) versus raven (black), based on a dataset of 100 procs each.

    Method:

    I ordered my lemming to attack me, capturing proc times using a trigger that recorded the Lua os.clock() output each time I got vertigo, until the record had 101 numbers (first number was when the lemming was ordered to attack). I repeated this same exact protocol, except for raven next. I exported the data to R for analysis.

    R analysis:

    I subtracted consecutive proc times to obtain interval lengths. Then I constructed empirical cumulative distribution functions (ECDF) for each animal using R's ecdf() function. Finally, I plotted the two alongside each other (see Figure 1, above).

    It is pretty clear from the graph that the two animals have different proc rates. However, just in case anyone wanted to argue the point, I also ran a two-sample Kolmogorov-Smirnov test, which rejected the null hypothesis that the two sets of data were drawn from a common distribution in favour of the alternative hypothesis that the two sets of data were not drawn from a common distribution (p = 2.25 * 10^-7).

    -------

    Technical stuff aside, here's how you read the graph. If you want to know what percent of time a proc will come before a certain length of time has passed, you go to that time along the x-axis and go up until you hit the curve, then read the value on the y-axis. For example, the lemming will proc under 20 seconds close to 100% of the time, but the raven will only proc under 20 seconds ~80% of the time.

    Conversely, if you want to see what length of time you need to allow for the animal to have at least a given probability of procing, you go to that probability along the y-axis and go horizontally until you hit the curve, then go straight down and read the value on the x-axis. For example, 50% of the time the lemming will proc in about ~5 seconds, but for raven that time is ~10 seconds (this value is called the "median").

    Here are some of the summary statistics:

    Lemming:

    Min. 1st Qu. Median   Mean 3rd Qu.   Max.

     0.177  2.224  4.457  6.451  8.542 23.508

    Raven:

    Min. 1st Qu. Median   Mean 3rd Qu.   Max.

     1.237  5.624  9.671 13.487 19.344 62.159


    Attached: raw interval data in .csv format (I know it says .txt, I had to change the extension because the forums doesn't allow .csv files).


  • Wow love it.

    Dont suppose you could do this for all the other pets? :D


  • It's really hard to see wolf because it overlaps pretty much perfectly with fox. Those two are the completely vertical lines near ~7.5 seconds mark, meaning they hit every 7.5 seconds, no randomness involved.

    I'll post the summary statistics another day. Looking at how the butterfly and badger curves overlap at some of the bends, I think I may need more data . More AFK time for me in the future, I guess.

  • edited March 2021

    Quoting @Venari: Is the effect of vertigo the only reason for the change, or is the speed in which it hits that is effecting the change?

    The reason for the previously suggested change was to account for changes in Sentinel anti-air with the removal of disrupt from trumpet. The recent reversion of allowing trumpet to still carry disrupt against flyers/trees removed some of the necessity of vertigo needing to be in the stack. I suggested that change because it doesn't bind us to specifically use lemming on passive and provides us with more interesting incidental (nonpassive) arial counterplay that rewards reaction and carries across approaches (petrify, lock, prepping). You just have to pre-strip deafness earlier in your sequence to ensure that you have access to trumpet when you need it. As I currently do it, I strip deafness early in my stack anyway and then apply sensitivity later if the approach calls for it. The condition of needing undeaf (to me) seems like a reasonable condition for the benefit of arial disrupt since you can't strip deafness from the ground.

    I can see it currently serving 2 purposes. 1. On the ground against an opponent trying to use fly to escape. -Or- 2. As a way to ground a flying opponent when you are also flying or after you've tracked to them. (It might be an interesting chase strategy against people who are flying through the wilderness). The recovery time associated still allows for an enemy to concentrate and move away on the ground in most situations that traps aren't present and since most of the time when we engage from above via track, we haven't already set traps below.

    The DD done by Rollanz above does raise the question about whether the passive animal attacks should be standardized, since the swapping in and out of vertigo and paranoia uncovered a difference in the passive affliction speed. I won't speak to the balance of it (because I haven't had a lot of time to test), but it does make sticking petrify more challenging. Whether that increased challenge is in line with what the intended class balance is meant to be is up to ACC and the admins to determine and possibly still up for some debate.

  • I'd be rather shocked if the increased difficulty wasn't part of the reasoning for the change - you are not meant to be able to pressure one kill route and succeed regardless of your opponents priorities, but with vertigo afflicting so quickly passively even with the opponent having impatience prioritised over paralysis with shield spam it was possible to petrify - this really should never have been the case.

  • edited March 2021

    I support looking into the proc rate for petrify aff pool. Personally I think the proc rate that ranges from 2-20 (lemming) is too quick at 2s and not helpful at 20s. I would much prefer a proc rate of 3-9s, or even 5-9s. Predictability like this allows a lot of planning on both the offense and the defense side.

  • edited March 2021

    @Dochita, I believe the animal has a set chance to proc at the end of each tick (probability and length of tick being specific to the animal). There isn't an actual upper limit on how long it can take to proc, lemmings and wolves aside.

    Brown is lemming ECDF, as I've already presented in previous graphs. Black is the ECDF of a randomly generated dataset based on this model:

    1) randomly pick 1000 numbers from the exponential distribution with rate parameter 1/5.25

    2) Round the numbers to the next 2.15 (that is, push them to the end of a "tick")

    3) add a tiny offset to each number (randomly picked from a normal distribution with mean of 0 and standard deviation of 0.01) to make the result plottable

    The graph suggests I'm fairly close to the truth. If so, that means a couple of interesting facts from exponential distribution applies to the distribution of proc interval lengths:

    1) The distribution is memoryless. How long it's been since the last proc has no bearing on the chance of the animal procing during the next tick

    2) The variance of the interval lengths increases as the proc rate decreases (or as the mean interval length increases if you want to look at it that way). The proc intervals for ravens would be expected to has greater spread than that of lemmings. This is borne out by my sample. The sample variance for the raven interval lengths is 137.16, compared to 22.52 for the lemming.

  • ArchaeonArchaeon Ur mums house lol

    tldr, math nerd

  • sigh

    tl;dr: the fox and wolf always proc at their next tick. The other animals have a constant chance to proc during at the next time tick - this chance never changes.

    Because of how it works, decreasing the proc chance at a tick and/or lengthening the time between ticks makes the time of next tick more unpredictable as a secondary effect. This is not obvious to people who aren't a "math nerd*", as Archaeon puts it.

    * I personally prefer the term "statistician"

  • Updated plot for the raven proc rate update. Note that I had to double the width of the x-axis in this version.

    Summary stats for old raven:

     Min. 1st Qu. Median   Mean 3rd Qu.   Max.

     1.237  5.624  9.671 13.487 19.344 62.159

    Summary stats for new raven:

     Min. 1st Qu. Median   Mean 3rd Qu.   Max.

     6.156  7.107 19.398 26.501 32.672 124.575

    19 seconds median means there's a 50% chance that the raven doesn't do anything in the next 19 seconds. The good news is that there's a 25% chance it proc within the first 7.1 seconds. The bad news is there's a 25% chance it doesn't proc at all for 32.6 seconds. Further along the long tail, there's a 20% chance it doesn't proc in the first 39.2 seconds and 10% chance it doesn't fire in the first 53.3 seconds.

    ====

    In terms of petrify afflictions, sentinels now have:

    • Doublestrike (slow attack, ~2.45 seconds with a level 2 spear, 2.74 with non-artefact spear) for one venom + impatience (or epilepsy if they already have impatience)
    • Enrage wolf (hallucinations, tertiary abilities that can be used every attack or every other attack depending on whether I'm throw a fast axe)
    • Raven (paranoia, 19 seconds median time to proc, see above)
    • Everything else depends on venom
    • No passive/tertiary goldenseal stack

    I'm pretty sure this means petrify is next to impossible to pull off solo with a non-artefact spear. It may be feasible with artefacts.

  • edited April 2021

    Definitely ideal to pursue the prep path now with haem pressure and threaten petrify if they prio swap haem. Bleed builds up quite fast with procs and enrages so the pressure is annoyingly good. Barb clots away mana making focus real precious. If they don't swap single leg wrench end game is quick to get too. 2c.


    The difference between arty and non arty spears are too great, compared to other class arties. Hopefully this will get some adjustments.

Sign In or Register to comment.