ฝัง
- เผยแพร่เมื่อ 9 ก.พ. 2025
- You can find jorbs shirts and hoodies at: www.bonfire.co...
If you want to buy the games we've been playing check out: www.nexus.gg/j...
If you enjoy these videos please consider dropping a like to help promote it to search algorithms. If you'd like to join the community of people who support me making content like this you can do so via:
Discord: / discord
Twitch: / jorbs
Patreon: / jorbs
Twitter: / joinrbs
(Or by Subscribing to TH-cam here :D).
"You could have 99% dodge and then get hit anyway and die in one hit" jorbs misses playing xcom i just know it
Reminds me of what my PoE friend used to say - "Whether you're reducing the incoming damage through armor, or negating some incoming damage through dodge, the most important thing is that you can survive one hit and heal before the next one"
Their trucks are high quality, and the resale value has consistently outpaced other US makes. Their lower end sedans on the other hand, are notoriously prone to several mechanical issues, and have almost no value after 100,000 miles. Overall, the brand is received poorly in US markets despite effective marketing to middle income truck owners.
rough.
Want to get into these comments before a CARAVAN of others really RAM home the quality of this reply, and my reply is NE-ON impossible to see.
In Path of Exile 1 for a long time there were two "dodge" mechanics. Dodge was purely chance based, but a similar effect "Evasion" was pseudo-random and took priority over dodge. If you had 90% evasion and 75% dodge (the cap) then every tenth hit would hit you through your evasion, but you had a 75% chance to dodge that hit and still take no damage. This is obviously very busted as it is basically 97.5% chance to avoid (attack) damage, but it would very often lead to you just randomly exploding when you hit that 2.5% chance to be hit. Most builds that did this put way more investment into Evasion and Dodge than life so you would almost never survive that direct hit.
This was basically impossible to balance, so they completely removed Dodge and replaced it with other defensive layers, one of them being "Spell Suppression". Spell Suppression is a percent chance to take half damage from spell hits, and this can be stacked to 100%. Any build that uses spell suppression basically has to max it out as suddenly getting hit for "double" damage feels really bad and was the thing the Devs were trying to avoid.
I love these PowerPoint videos
I tried to do something like this for my very first character as a Duelist. I did not recommend it afterwards.
@IsonDaya You must become a master of the 6-portal defense
The rationalist shorthand for the generalization of this concept is "The map is not the territory."
Don't confuse the model of the thing for the thing itself. The model necessarily abstracts some elements of the thing away, and those elements may or may not be critical to achieving your goals. This applies in reality perhaps even more than in games.
Came expecting a video about a Silent dodge card, then it turned out to be about a different thing entirely, then it turned out to actually just be about block in sts anyway. Love it
One heuristic I tend to value looking at with proc chances is "chance at two events in a row". It helps highlight chances of a streak. Ex. a 90% chance to dodge is an 81% chance of dodging next two in a row.
Like you said, games are complicated, and I feel like dodge generally tends to be good or bad when there is context sensitive upsides to it, as well as when the opportunity cost for acquiring it is relatively low.
Risk of Rain 2 somewhat infamously has its Tougher Times item for dodge chance which grants +15% per stack but stacks hyperbolically - so you run into immediate diminishing returns even after the first stack.
But Risk of Rain 2 is a chaotic game where you're surrounded by enemies and constantly being dealt chip damage and Tougher Times is a common/T1 item so getting just 2-3 gives you about a 20-30% chance to dodge an attack.
The important context here is that there are attacks in RoR2 that apply status effects, sometimes crippling, sometimes stacking, and a dodged attack nullifies those status effects, so not only does it represent eHP, it also represents some amount of reduced effective DoT, for example.
But critically there are also enemies that will borderline one-shot (exceed your one-shot protection threshold) and a successful dodge there is worth quite a lot. Even if you fail that roll, every successful dodge past that point can be a critical second to let your regen kick in and get you back to safety.
Whereas I don't see these sort of scenarios in Brotato - there seems to be item-based payoffs but I'm not familiar enough with them to give any insight, I'd imagine the opportunity cost for dodge there is just too high for the initial investment when you can instead just get better at defeating enemies before they can hit you.
(safer spaces is objectively better if you have movement speed and enough skill to not immediately get swarmed because it's a guarantee)
I am glad this came out as a semi-response to the comments on the Vampire run.
(And I'm sorry I keep missing you being live but at least i can watch you on here.)
lmao i just saw that comment chain and now jorbs dropped this whole video
which run i am deadly curious
@Wishuponapancake vampire/ big streak #12
I do understand the numbers but with (censored) hours of Xcom I feel VERY diffrent about the increase from 99% to 100%.
It's the difference between needing a backup plan, because you cannot afford to lose everything one in every 100 times you attempt this stunt, and being able to spend your resources optimally _knowing_ what will happen.
I would've liked to see one more slide on an opposite scenario, where dodge is much more valuable than raw HP ("Is HP ever... worthless?"). If the incoming damage isn't small 1 damage hits but instead big 100 damage chunks then going from 1hp to 99hp is a 99x multiplier on eHP, but doesn't actually help you survive a hit at all. Meanwhile 90% dodge 1hp "only" has 10 eHP but is actually likely to survive multiple big hits before perishing.
This example shows that the answer really is "it depends" and that eHP isn't a great metric. "Expected hits to survive" is a much more meaningful metric (especially if you cap it at some reasonable value - being able to survive 100 or 1 billion hits doesn't make a meaningful difference), but it doesn't have a simple formula, especially not if some regeneration / leech effect is in play.
One thing that people don't always consider is how hp stats can interact synergistically with other stats, where dodge chance often cannot. A common instance of this is in Diablo 4 rogue builds. It is often better to forgo dodge chance for max hp past 20-35% give or take, because of the life-steal mechanics. Sure, not getting hit by things feels nice, but the chance to get blown up instantly is still there. If you have high HP values, your health bar often gets chunked down to 25-40%, but it also instantly gets refilled by life steal. This gives consistent gameplay that can be planned around. I can plan, "If I sidestep attacks x, y, and z, then I can face tank and out-heal everything else." Even when life stealing with dodge-chance, the odds of getting blown up by 2-3 hits will always remain and it can become a gambling game when fighting any groups of mobs, or bosses.
This is personal preference I suppose, but I would much rather be able to think to myself "Ahh I just need to learn to sidestep x ability and have better hands/reaction time. There is a clear path for improvement in my gameplay." Rather than, "Ahh unlucky, I got one-hit through my 70% dodge chance. Guess I'll run my head into the wall again and hope I get lucky this time."
Ignoring particular character quirks, in Brotato I value dodge probably the least of the core defensive stats (those being dodge, HP, and Armor). My heuristic for dodge tends to be less structured around the concept of "effective HP", and more around the concept of "defensive layers". If I already have decent HP and some armor, I probably won't choose more of either of those over, say, an epic dodge upgrade (9%). But it is definitely a priority to get the other two stats to decent levels first, and if I have a run without any dodge, that's not abnormal.
Its also interesting because certain factors can make really bad ehp calculations be way better too. Like, if you have something thatll block the next hit every 10 seconds, suddenly theres a threshold of evasion yhat if you reach, youre expected to never die which is WAY lower than ehp would ever show you.
League of legends used to have various items that gave dodge chance. They realised that dodge chance was broken in league, so they removed all forms of dodge except one which gave temporary 100% dodge chance, and in the over a decade since only added one form of dodge, which is again a temporary 100% dodge chance.
نشط: يدخل Jax في Evasion ، وهو موقف دفاعي ، لمدة ثانيتين ، مما يت and so on and so forth
Not part of the point of the video but since you mentioned: dodge is frustrating in competitive games. It doesn't matter if, matematically, it's balanced. It feels terrible to play against and not very impactiful to play with.
@darklorty 100%. It sucks to lose to something random, and it sucks to invest in something and lose because it didn't work in the crucial moments
@@darklorty It's funny how crit is perceived very differently from dodge. It's true that the extremes of the probability curve are very different (always dodge and live forever, or always crit and only increase your damage by the crit multiplier), but the way they introduce a random distribution into combat life expectancy is very similar.
@@celestialspark5697 crits are also frustrating tbh. Especially in early game when you have 25% chance to crit and 1 crit could make the difference between winning or losing a fight
Note that the conclusion at 11:34 is only right if you're trying to maximize games won/time until next win. If on the other hand you're having to chain many such games together (win streak/permadeath), then going to infinite eHP actually is infinitely valuable since your winstreak length becomes infinity. But even then, this is assuming that I can have infinite eHP in all games in the streak. If I have a 49% dodge game and a 99% dodge game and I want to streak both, I'll of course buff whichever one I'm the most likely to die in. So, it all depends on what you're trying to maximize.
Discussions like these were the part i most enjoyed about tanking in MMOs, especially around 15 years ago. There were multiple tank types with differing damage reduction strategies (Armor vs Dodge vs Block) that were better or worse for different fights. There was always some calculation required to maximize your average eHP, but that was also just a proxy model for how difficult or stressful it would be for healers to heal you. Even with all that considered, there was still no replacement for timing your abilities well and performing mechanics correctly, but calculating upgrades mathematically was incredibly satisfying before so many tools and tier lists were available.
I clicked because I thought this was about Silent’s Dodge and Roll, I stayed because I love jorbs lessons
Good stuff, Jorbs. I've done a lot of survey work and performance metric development in my career. I've often had to convince clients that they should not care about the average value. This is a great example of that. Most things people actually value are discrete events that happen at a particular known (or well-estimated) point and with a known (or well-estimated) number of opportunities -- death at 0 HP, for example. That is, you don't actually care about the mid-point of the distribution of possible events, but rather the amount of the distribution past a certain point. People so readily jump to the average, working up from the data, that it's actually quite helpful to force them back to working down from the problem itself -- typically what decision are you making and what do you need to know to make it.
It's worth noting that you have to be careful when applying any valuation of dodge chance in a game to check how dodge chance actually stacks in that game. Many games that allow the player to get many sources of dodge chance just treat each source as an independent, separate chance to dodge, meaning you don't get increasing returns by stacking it from different sources (and neither do you get diminishing returns). It's common that people will assume that dodge chance in games is additive as a default.
And yes, you need to apply your decisions about stochastic attributes to a meaningful context. For example, Path of Exile Hardcore players tend to value low-variance builds that are guaranteed to survive a lag spike where enemy attacks hit them, because the cost of low-rolling on something like dodge chance and dying is so extreme. Whereas in a game like a MOBA, where resources are limited, a player may opt for crit chance that gives a high chance to win an encounter over having enough raw damage to be guaranteed to win an encounter because the crit chance comes at a lower gold cost for a given amount of average damage.
Are you telling me if I write a comment explaining something poorly I get a Jorbs powerpoint explaining it better?
I promise I wont use these powers for evil.
Inb4 over explained run explaining the logical flaw in this comment so the loophole only works once
A wise man once said Education is elevation
The game I think about is Battle Brothers, and not just because I think the ugly little men also look like potatoes. Particularly for surviving big hits from late game enemies, there's a lot of discussion about eHP as a concept, but usually framed as how many hits on average it'd take for an enemy to take down one of your men. Having more health and bigger armor is always the obvious way, because to your point, you could go from just *always* surviving 4 hits instead of 3 hits, but dodge chance there really does increase how often you're expecting the worst case scenario, where you had to manage to survive more than 4 hits, to go better. It also helps that you're typically talking about people whose deliberate job is to go stand in front of a big hit, so investing in both literal health and effective health doesn't feel as confused.
All this being said, watching your video is a good reminder to me that as tempting as it can be to skip "hard" stats like health, dying in one hit the moment things go wrong is never a gamble I find worth taking.
Dodge chance is why Armor Class is so valuable in Dungeons & Dragons. AC is the target number enemies have to achieve in order for an attack to be successful. In D&D 5e a Wizard with 10 AC facing an enemy with a (1d20+7) bonus to attack has nearly the same eHP as a Wizard with 12 AC, because their dodge chance only went from 10% to 20%. However, a Fighter with 20 AC has a 60% dodge chance so raising their AC to 22 makes that a 70% dodge chance and thus a much higher eHP boost.
This is doubly true in Pathfinder 2e, where an attack that exceeds your AC by 10 (or is a result of 20 and also a success) deals double damage. This means that higher AC and thus higher dodge chance not only reduces your odds of taking damage, but also reduces how much damage some of those attacks deal. Furthermore, enemy attacks that are 10 or more points below your AC are a critical failure, which can trigger bonuses such as the Swashbuckler gaining their class resource and being able to make a counter-attack against that enemy.
That said, there is one flaw with eHP projections - attacks that 'overflow' maximum health. For example, let's say you're a squire with 10 HP and a 50% dodge chance - 20 average eHP under normal curcumstances. The knight has 100 HP and a 0% dodge chance and is being attacked by an orc for 50 damage. If you use a taunt ability to redirect the attack at yourself there's a 50% chance you take 0 damage and a 50% chance you take 50 damage... but only lose 10 health. You can't lose more Health than you have, after all. Meanwhile the Knight has a 100% chance of taking 50 damage if you don't use a Taunt. So, in that case your 'temporary' eHP isn't really 20 because you have a 50% chance of negating 50 damage, it's more like 100.
I thought if you go a certain amount below 0, your character dies instead of being knocked out in Pathfinder? Otherwise, great points.
@@asid61Their example in the second part isnt specific to pathfinder, its just showing how ehp can be misleading
Regarding the last example, this is falling for the exact trap stated in the video. Ehp isn't value. Presumably in this setting, both you (the squire) and the knight can both do damage, which is value (assuming you win by killing the orc). The knight has a 0% chance of dying uf hit, you have 50%. By taunting and maximizing ehp, you most likely are decreasing value(chance of losing your dps by dying)
@@kilpatoo8634 That's exactly right! Although obviously there are some situations where context changes things. For example, if the squire and knight are in a game like D&D 5th Edition where typically hitting 0 hit points alone doesn't kill you, just knocks you out, and there are resources like Healing Word that can bring someone back from 0 and into the fight, then the squire 'strategy' can end up being worthwhile. Why tank 50 damage on the Knight that's hard to heal when you can risk the squire dying and then bring them back with, like, 3 HP in a turn or two?
All that to say, eHP is weird. Ultimately I think the best way to maximize eHP is to just kill everything that can possibly harm you, resulting in infinite eHP, because the best status effect you can inflict is Dead.
Brotato is a fascinating example because for most of the game, dodge is much weaker and if you're only trying to win runs, you can basically ignore it.
If your goal is to get as high as possible, then the value flips at some point after 20 and dodge becomes the only useful defensive stat because enemies oneshot you no matter what
Really liking these types of videos from you. I guess I'd call them Powerpoint presentation type videos. I'll admit Jorbs I burnt out on your STS videos at some point, but then the algorithm served me up that MTG ruling video and it got me back into your channel. This (imo) is you at your best.
I think a lot about Goodhart's Law, or the danger of optimizing for a metric of success, instead of optimizing for success. Modeling is all well and good until you realize you've twisted reality into a pretzel trying to fit it to the model, rather than the other way around.
I feel like ive seen two jorbs lectures covering these exact topics. Sometimes it's more important to consider distributions over averages, and be sure that your heuristics are in service to what you actually want
Love the video! The "ch" in stochastic is pronounced as a hard c.
oop i totally studied the language that sound is from too
Dodge always makes me think of Fire Emblem, as there are entries in that series that are not straightforward about hit/dodge chances.
The displayed number shown as the hit/dodge chance is the threshold, but the system rolls 2 rn's and averages them out to determine whether the attack connects. I always wondered if there are other games that pull this stunt.
Great vid! Would have loved just one slide comparing/contrasting with the very similar analysis people like to do on crit chance, especially since this was a shorter than average Jorbs presentation.
Dodge is mostly interesting to me for how its value tends to accelerate, rather than diminish. Generally, getting +10 armor will do less and less for you the more armor you have. However, +5% dodge chance will do more and more for you. This means that Dodge tends to shine in games where you can realistically approach 100% and then have enough of the game left to actually use it, but tends to feel lackluster in games that don't last that long.
this reminds me of the kelly criterion/geometric mean stuff in betting/investing
Wait, is this video secretly about Caves of Qud? IYKYK hehe
A jorbs CoQ playthrough would be so fun
So what I'm getting from this is that 'average eHP' is a pretty _dodgy_ statistic to rely on.
edit: To actually add to the conversation, I think with Brotato people get tripped up by opportunity cost too. Like yeah, in theory high dodge synergies well with high armour, but they don't consider that actually _adding_ dodge to a build demands resources that could have been spent elsewhere.
My friends and I love to play dominion and we play almost every week. We often try out new sets so we have to evaluate cards quickly. It’s always interesting to see how useful our heuristics are, e.g. what’s the expected value of drawing a card? What’s the expected value of giving your opponents a curse?
Damn you Jorbs! Tricking me into learning again by being so entertaining and interesting!
There's a very interesting comparison between evasion in FTL: Faster Than Light and Into the Breach, two games from the same dev, with two wildly different implementations of dodge.
In FTL, dodge is your first line of defence and you tend to take up to a couple dozen hits per fight. So stacking it high can be very effective, in particular because some of your other defences recover over time, so even one attack missing you can have outsized implications for the rest of your defences in the immediate future.
In Into the Breach, dodge is your very last line of defence, and you can typically only get hit 3 times before you lose, with no in-battle recovery. So stacking grid defence is near useless and very unreliable.
It's a really interesting instance where two games implements a very similar system to such drastically different results.
it's also a big deal that FTL has a fairly large health bar, and that every hit you take can matter a lot. both of those things matter for dodge, because dodging even one hit can be very valuable, but failing to dodge several times in a row still isn't necessarily lethal.
I play for the king a lot, and I usually stack evasion on characters which is effectively dodge. I never thought up of effective HP as a heuristic to show why, but it seemed valuable to me to reduce the impact of variance when it comes to dodging hits. I guess it was a trap all along and instead of using dodge reduce variance, I should’ve invested into other defensive stats.
Alternative title: "Heuristics? Get out of dodge!"
you already tricked me into thinking about heuristics by putting it in the title :D
Excellent video as always, Jorbs. I always enjoy you breaking down topics in this manner. I do think, however, that this could be expanded on in some interesting ways - obviously with the understanding that it needn't be you doing this. I'm well aware that the there is a finite amount of time you wish to spend on topics like this.
The fundamental expansion that I think would be interesting would be to be able to interact graphically/ animatically with this data, because I think it would most easily clarify some of your points.
If a person were able to interact with the variables and see their impact in real time, perhaps it would help those who are less 'statically inclined', as well as exposing new interesting conclusions.
I'll give a few examples:
In the game 'Survive X hits', without dodge (or other variables), the game is entirely binary - 0% chance to win or 100% chance to win.
If we assign some values e.g | survive 50 hits to win | 1 damage per hit | 30 health | we could then animate the graph, 'Number of Hits Taken' and watch how the distribution shifts as dodge increases.
At first we would see a 100% peak at 30, then watch as a tail forms, see the peak's magnitude decrease and shift rightward, and eventually seeing as all but the shallowest tail remain under the 50 hits threshold.
This game could then be tweaked in various ways for various circumstances. For instance, 'Don't take X% Max HP in Y seconds', which might be a reasonable stand-in for some Brotato situations. I can recall you saying something like, "I generally feel in most runs that I can recover if I can survive at least 3 hits". This model would *start* to deal with more of the variables that impact player choice, namely +Armour and +Max HP.
I would be interested to see the boundary conditions that allow you to start using Dodge directly as Effective HP, and the models and animations I am suggesting start to examine this. A fair comparison would be with Minecraft, specifically the Unbreaking Enchantment. Obviously the stakes are different - you don't automatically 'lose', and you can stop using a tool prior to using it's last durability - but adding Unbreaking to a tool gives it a chance of 'dodging' durability damage, up to 75% at Unbreaking 3. At this point (for a Netherite Pickaxe) the game is effectively 'Survive 2,032 Hits' it is fairly reasonable to say "Unbreaking 3 makes your tools last 4 times as long".
If you have a digging project that will destroy 5,000 blocks, you're pretty much statistically in the clear to only need that one pickaxe (though it would be cool to run the numbers on that).
The most prescient way to consider this, imo, is by highlighting your specific circumstances as compared to the average player. As you've said with some of the characters (again, paraphrasing), "I feel like this character should never lose, unless you have a run where you receive absolutely none of [insert common stat here, ie flat ranged damage]". Your goal *isn't* to 'Win Danger 5 runs', as such, but rather minimise losing as many as possible, as you wanted to achieve a rotating winstreak through all of the characters. I don't care if I lose on Jack 10 times in a row before Laser Guns work, whereas that is obviously ridiculous for what you want to achieve.
The problem isn't necessarily about dodge increasing your chance at winning then game, but rather the opportunity cost of going for a strategy that has an effect on the deviation of the win data. I hope I've been clear enough in laying out how I see it (I confess that I may have done a pig's ear of it).
If we were to present people with a couple of these graphs and ask them which one to pick I suspect many players would pick the dodge graph. "Look at how many hits I could survive" as the data extends far past the threshold to win, their eyes gleaming with the idea of surviving into infinity.
I could imagine you looking at the graph with a small frown, thinking about the 5% chance of being 2-shot by a hammerhead shark, "I think I'll pick some more armour on level-up".
_________________
I really should could a small app in Godot to do this. It really wouldn't be hard and it'd by nice to have some sliders to play with and be able to get a feel for how these ideas change in real-time.
High dodge is high dopamine value, until the crit lands that ends your run. I like to call that crit, "Murphy's Reminder".
I like the part where Murphy got a crit and Murphied all over the place.
It seems like TH-cam Jorbs has learned how to dodge the camera successfully.
Also, good video! 😊
Given my stats background I had thought of the same concept too, which seems to escape even a lot of the top players in various video games. In the end, the best way to usually look at games is whats the worst scenario possible and how do I improve my chances of surviving it, rather than how much damage can I take on average, which would only make sense if you take consistent rapid but low amount of damage per tick (say 100 damage every 0.01 seconds instead of 100k damage every 10 seconds).
The same is true on the other hand too about your character doing critical hits. Sometimes you may not have the "eDPS" for a fight but through critical hits with some luck you may have a chance to win when an increase to DPS may do nothing. E.g. an enemy has 100 hit points and will kill you in 10 seconds no matter what, then increasing your DPS from 8 to 9 means you still lose 100% of the time, but if instead you get 1% critical hit, it may only increase your DPS from 8 to 8.08 instead of 9, but you have an actual (although tiny) chance to win now.
Healing in video games is also only about two things, do you have the mana/mana regen to heal the entire fight? If yes then the only concern is looking at big damage windows and how to heal them, not your HPS
I've played an MMO in which the faction could have a handful of players transform to have ten times the health pool and hold aggro on raid bosses while receiving support from multiple healers. In this case, increasing your average HPS actually can show through and even result in reducing the number of healers you need to bring.
Yay! A new Jorbs presentation video!
I find it much more effective to instead think of dodge as a miss/invulnerability. Dodges usually avoid status effects as well, because you don't get hit. Most players make the distinction between HP/armor/block tanking and dodge/avoid tanking. A character can be both dodgy and squishy.
Dodge is more of a direct counter to status effects rather than damage.
"100% dodge gives me infinite ehp" mfs when their 10 hp ass dies to blood donation
Dodge has always been a curious stat to me, especially when it comes to tabletop RPGs, where dodge is something that exists innately in the concept of Armor Class or similar. Other sources of damage mitigation used with it are *incredibly* powerful, but by itself, it can only get you so far, and when you get hit, it will hurt. It also doesn’t help you against Saves, which can often be more debilitating and longer lasting than damage, especially in 3.5e, where debuffing spells can temporarily or even permanently reduce ability scores and level. This applies to a wide range of TTRPGs in different ways, and some systems handle it way differently… but that’s the basic idea.
So… yea, Dodge is a useful defensive layer, but like all defences in games, it shouldn’t be your only defensive option. Stacking multiple layers is key for success, and Dodge is only one part of that defense.
your example isn't contrived this is how combat fundamentally works in Old School Runescape - one of the most popular MMOs.
With a variety of different weapons that have different damage maximums, accuracies, and attack cooldowns, calculating damage dealt and received while considering Max HP requires calculating convolutions for damage, distributions, and binomials for how quickly the weapon swings are applied.
It feels like the conclusion here is 'the context matters' but we didn't take the opportunity to assume some contextual variables that are almost always present in games with these mechanics. Yes some were discussed but their impact on the 'goodness' of dodge wasn't really considered and they were only briefly noted as the other option to improving dodge.
1) These games often have methods for enhancing your actual maximum HP.
2) Enemies in these games almost never hit for exactly 1 damage.
Both of these can be taken into account by thinking about enemy damage per attack as a % of your actual maximum HP. Then the better heuristic to compare would be average number of hits until death. For the value of increasing your maximum HP, the average hits until death (aHUD) is a very straightforward graph that is flat as maximum HP increases until exactly the point where you survive the next additional hit. For dodge it's a smooth curve. From here we can see that for a given enemy attack strength (relative to a % of your maximum HP), the value of the next 1% dodge or next increment of health (this comparison would need the context of the cost of these attributes in a specific game- i.e. does 10% dodge cost the same as +10 max HP).
I do expect max HP to generally have consistent value in line with its graph and dodges value to vary though I'm not sure in what way (I would assume it would scale somewhat similar to the way eHP did since going from 99% dodge to 100% dodge does make you aHUD go from a finite value to infinity). This leads me to believe that the optimal strategy in games like this would be to get enough maximum HP to ensure sufficiently small % of your total health is taken per hit, then later focus on dodge - of course that's a no brainer.
TL;DR More stats good
I thought you were gonna talk about the variance. With high dodge but small hp, you'll be really stressed even in easy fights but you might be able to win really hard fights that you shouldnt. With high hp and low dodge, you'll know how a fight will result in and have a further planning horizon. Discretization is also an important factor. If a boss has a really big attack average effective hp might not carry much meaning.
if there's one thing i would want to teach people who want to play strategy games, and also would want them to take it from games and apply it to the rest of their life, it's "get clear on what these decisions are trying to accomplish, and only then start deciding whether the decisions are good or bad."
not all of life (or games) is making decisions! but insofar as any of it is: thinking critically about what motivates your decisions will empower you. and if you're just not in a position to know what constitutes a good decision, develop skill and tolerance around that. maybe you don't always have to know, but even appreciating such a thing is in itself a good critical stance. everyone can do this.
I learned this lesson in diablo 3 since monk used to have very little damage mitigation other than dodge, so even though you could get similar average expected hit points to other characters, you'd just die randomly. In games like diablo it's a lot more valuable to know that getting hit 3 or 4 times wont kill you then needing to roll at least one successful dodge every 4 times you're attacked.
With respect to marginal value-adds of dodge chance, a big difference between poker and brotato/diablo/(insert your favorite action rpg) is in the latter you're choosing which hand to attempt to draw and win. E.g. buying a steel helm is like trying to draw a straight flush -- it might rarely protect you from a lethal headshot, but otherwise you draw high card and only provide minor damage mitigation. Or buying platemail is like trying to draw a pair -- it could provide consistent damage reduction, but fail to protect against the lethal headshot.
Whereas in poker (texas holdem), you're deciding how many chips to bet and need to consider Kelly Criterion in addition to expected value, since the goal generally is not only to make money, but also never go broke.
Dodge scales best with tardigrades. The stochastic element of dodge probability will trend towards 0 as you approach larger numbers of tardigrades. (Duh, inf tardigrades makes you unbeatable anyway) But dodge gets you there quicker and can be built up from the start of the game and offers some interesting combo items.
had to double check because you had me wondering whether I've been pronouncing stochastic wrong my whole life - but the online dictionaries seem to agree with me so I'll share. the ch in stochastic is pronounced as a hard k sound.
I did have to double check if it was a regional thing, or just some rogue stats professor is out there spreading chaos pronunciations.
@@rcuhljri used to pronouce chaos as /tʃ/owse and still occassionally do
Early in the vid Jorbs says "What is dodge? What are we even talking about?" and my brain immediately went "no, that's not how it's pronounced'... DOGE is destroying the country, not 'dodge'. But that's not what we're talking about.
I had never heard of average ehp, but i googled the evasion (dodge) cap in a game i'm playing. reading the first post about it, they mentioned average ehp. huh, the more you know.
now i just wanna see jorbs play spirit island with ben and wheeler from loading ready run
eHP is strictly worse than AtK/TtK (attacks to kill/time to kill) as a heuristic, because it fails to include literally every other factor of actually surviving, which is the bottom line.
It's the reason why say in SCP:SL, revolver and shotgun are terrifying weapons whereas most of the damage auras are comparatively much more manageable (those damage auras still deal lots and lots of damage! but they don't do it in literally 2 taps and you can try healing in between, which you can't against shotguns and revolvers).
On time for a lecture from Prof Jorbs
It's not EHP, you should think about dodge the same way as drop rate: x% drop rate means 50% chance of getting it after y tries.
In strategy games it's probably x% dodge means 20% chance of losing after getting hit y times, assuming you can tolerate 20% error rate. Then you either solve for x or invalidate the strategy based on the required y being too high.
stoTSAstic! stoCHASTISE me baby! (Love you Jorbs!
All odds in game is 50:50. It happens or not.
The problem with optimizing heuristics can become really large in machine learning -- which will optimize whatever you tell it to, and not necessarily what you actually care about. For advertising, click-through rate is not the same as return on investment or profit. For us humans, it's very difficult to have heuristics that properly represent "the good of society"
And then, undeterred by all the probabilities, murphy's law hits you and denies any and every calculation made so far just in spite of having to abide by logic.
:D To some extent, kind of feels like the base rate fallacy where thoughts aren't spent on "what is the usual case" but "uhhh, nice, big shiny number" and "i think i know what this number means because there is always a simple clear and wrong solution", as in forgetting to pay attention to "what game am i playing" vs "what game do i think i'm playing" and "what would i actually have to do to win instead of just going uhhh, nice, big shiny number"
I just think it's more often than not a.. somewhat bad strategy to apply the rules of uno / black jack to poker just because both / all three have numbered cards with some colors
Cool video. Just FYI, stochastic and stochastically are pronounced with a hard "c" or "ck" sound, not "ch".
Intuitively, I would think that 'Average EHP' would scale with incoming damage, no?
I feel like (Incoming damage per hit)*(chance to get hit) should get counted somewhere
I only ever get 1% dodge chance cuz im a skilled enough player to dodge every time with that
hey jorbs love you appreciate you could you please use black slides
I love the anti-heuristic videos. Ive always dislike average as a sole statistic for evaluating. I played a lot of poe1, and dodge was a good damage reduction mechanic, but it had a ton of variance and not survivability against slam attacks. I always just accepted slam=dead. For smaller but multiple attacks i would think about the variance and what recovery mechanics it would take to decrease the variance.
Or another way i think about why average isn't a good metric is the lottery. With a sufficiently sized jackpot, the expected value is positive, but the probability of winning more money than spent hasn't increased.
I don't think heuristics are a completely bad way of reducing complexity, but more dimensionalities can certainly improve the understanding and effectiveness of the heuristic.
Consider me tricked Jorbs!
Opinion rejected. Doge is always very bad and Elon Musk is stinky.
You really have to give Donald Trump credit. I've never seen the Canadian left and right more united on any issue.
The master negotiator sacrificed economic stability, the goodwill of America's closest trading partner, and united political factions to oppose him. In exchange for that hefty price? A meaningless committee, an equally meaningless bureaucrat, and the exact same deal that was already negotiated a month ago. Truly a master!
@@CaptainScarfish So predictable that he'd huff and puff, then completely fail.
The main problem with your main point is that people are often too confident in then skill than reality so the scenario of being too good to be hit is irrelavent to the vast majority of players. Even if you are that good people still have periods of distraction or just mess up occasionally as part of human error.
doge > dodge
eHP for me now after playing PoE and World of Warcraft for so long is just a pass/fail mechanic, so Dodge is not really a thing outside of 100% for single hits or high for multi hits, but i find really cool to abstract those concepts. Generally in games balanced around 'difficulty and live service' dodge becomes a mini-game of can you get 100% because of a skill or perk? go for it. If not is trash, i wish we could become more simple like in single player games, that you can think more about it.
No we need fo calculate dodge % block % parry%...
This video is why I hate playing Ghost in Brotato lol like sure 90% dodge oooo wowee sounds good on paper but man that -100 armor feels way more important most of the time
i prioritize building max hp over getting to 90% dodge on ghost!
@@Jorbs I had what felt like a strong start and then wanted to table flip because the walrus one shot me from max health. Will aim for more max HP! Hoping to get consistent enough to winstreak as well and haven't quite solved all of the characters, but I'm enjoying the journey and enjoying the differing approach you and I have taken to some of them.
Ooo tanksims
xcom