Ok, so one dev of the 100 is confirmed to play the game? Is this dev himself working on the gear on not? If he is working on something else, is his knowledge being really used?
I would like to see EHG tell us how many of the employees who work on itemization play the game. As let me suggest, that [summon a crab on dodge- replacement] was not made by him, as it felt like something out of a bad joke.
Heâs got Rive set to skip the third strike. And 1/2 to give the Crit buff plus phys pen buff.
Probably needs some melee attack speed on gloves and weapon to get those buffs up (especially the penetration) faster? And Crit chance seems really low if heâs trying to actually use that. I wonder if he wouldnât be better off Multistrike. Could get more smites that way.
Hope he keeps experimenting. Looks like fun.
Not only are they making the game they like to make, as Kulze said, Mike streams every week playing the game.
They nerfed what was overperforming and popular.
If only a dozen people use them, theyâre not a problem for the game. Itâs the same thing you see in PoE: a few players have a pretty strong combo/bug abuse/exploit/whatever build but it only gets nerfed when more people are aware of it and the devs notice it.
Likewise for currency exploits/dupes. Lots of times there are a few players abusing exploits and bugs and duping currency and itâs only fixed when it becomes general knowledge (and thus usage).
If no one had leaked the gold dupe video in LE just to crash the economy, it would probably still not be fixed. Because it wouldnât be much of an issue on the game overall.
The vast majority of those 100 arenât decision makers. The core team plays the game.
We all really need to learn to stop feeding the troll but sometimes it is fun.
Thatâs my favorite eyeroll. 
I aint trolling. I expect people who make the items that players use to be playing the game at 1000 corruption and know what they doing. Otherwise, it makes more sense for us on the forums to make the items. I expect people who make the game, to be competent, and skilled.
Any game like this, with lots of interconnecting systems, will always interact in unpredictable ways. Which is why every single game in history had the same problem when they introduce new stuff. It has nothing to do with competence. Itâs simply a chaotic system and thus impossible to predict, much like the 3 body problem.
So unless itâs a static content game (as in, you donât introduce new items or mechanics to it and just fix the ones in there), OP interactions will always happen. Itâs inevitable. Much like exploits will happen. Theyâre inevitable.
You canât get rid of them anymore than you can get rid of bots.
Fully rid of them no. But if the devs play the game, and they have a member on the team who plays 8 hours a day, and their sole job is to test combinations with different items, and combos, the odds of them finding them goes up specifically. The devs can spawn any item or combo, they dont need to farm for 300 hours to get the item.
Get 1 person per spec, and have them do this each day, and test it. Problem wont be fully solved, but 99% of the outliers will be gone. It aint hard to do. Its called quality control.
Do you realize how many different combos of passives/skills/skill trees/items you can have? If all of their 100 devs spawned a different combination per second, they wouldnât be finished this century.
So no, itâs not as easy as what you make it seem.
Community testing picks up a lot, but there are always stuff that make it through. Itâs inevitable.
Even D3, which had PTR for a week to test the next season (and thus had thousands of people testing it) regularly had a broken combo that had to be quickly fixed. And D3 was a lot simpler than LE.
So yeah, itâs not hard to do. Itâs impossible.
One person per spec, testing things every day, may take a while yes, but they got 4 months in a season to test things. 8 hours a day is a lot, I expect people to work for the entire 8 hours they are paid - unlike some companies where people do 2 hours of work in an 8 hour day.
There usually only 5 skills per spec. Even if you take a two weeks per skill to check - which is a lot, cc have checked skills and builds with mods in weeks, not months. That is 10 weeks, and they got 16 weeks. They got plenty of time.
I expect workers to work, and not talk, watch netflix, make calls, and fall asleep at their desk. If people actually do work, 8 hours is a lot, especially if you can legally force them to work 12 in the last month per cycle release.
No, one of their 79 employees (not devs, that includes accounting and stuff) does a weekly several hours long stream. Several others are known to play the game regularly.
Once more, as often mentioned⌠the initial core team are all diablo-clone players which wanted to make their own game because they have a passion for the genre.
What do you think they would play when theyâre actively making their dream game personally after all? Pacman?
Also as a side-note⌠nobody cares how many of their devs play the game actively. Some might not have time to do it outside of testing, some might simply not be interested in it. The ones at the upper level though are the passion people and thatâs whatâs the major importance. They after all give the ones screwing things up hell because it screws with their own off-time enjoyment.
I know⌠but to be fair some of your arguments clearly feel like you would.
Which the game is not intended for, 300 is the golden area. Still is. And itâs tested to achieve decent results around that area or alternatively they throw in things which just âlook fun for people to mess around withâ and see what happens. Very good choice by the way, PoE did that years ago and it was a blast for people.
Umh⌠you see the product? You saw the implementation of 1.1?
Derive the answer of how skilled they are from that. Itâs self-explanatory.
Well, how long do you think does a single skill take to test?
That person doesnât âplayâ 8 hours a day, that person at best tests 8 hours a day. In testing you go through the notions a player would go and note down everything which seems to be out of place, explain it in detail to the people in charge of actually fixing it or sending it over to them and then return to testing. Hence to get 8 hours play-time that person will likely work 25 hours for it.
Testing a build entails to see how it works at every step of the progression, so obviously weâll have at least 10 hours - if not more - of testing per build.
Now turn it into actual work hours, how many there are per week, how many week a month has and hence how much gets actually tested by a single person.
Yes, the baseline quality control is also something Iâm complaining about, some things shouldnât be missed and are a gross oversight in how their testing environment is set up. Thatâs something which still needs work clearly.
If you pay em they will do that eagerly⌠I guarantee it 
Ehhh⌠big⌠ehhhhhâŚ
You know as well as I do that you only need to test the things interacting with each other. Some odd things might happen here and there still but the obvious stuff gets at least sorted out.
EHG messed up with those things simply, itâs a very fair point to mention that. They just need to handle the organization for that better and allâs swell.
It isnât⌠it absolutely isnât.
You come in, you need to settle mentally first, you donât get productive before being there for a while.
You need to interrupt your âflowâ regularly to note things down and bring information to other people, takes a few minutes to get âback into itâ then.
You got break times which also affect that.
Itâs not so clear-cut.
Again, this is displaying such a huge ignorance of combinations.
The problem never arose from using a single skill. It arose from a combo of skills and/or a combo of skills with gear.
So letâs start to break it down:
1-Each mastery has access to around 20+ skills. That alone is a total of over 15k possible combinations. Iâm not even going to consider the different layouts each skill can have because that will shoot the number up to astronomic ranges.
2-Each mastery 104 passive points and has access to around 80 different passive nodes. For the sake of simplicity, letâs just say you put an average of 5 points in each. That is a total of 10.100.903.263.463.355.200 combinations possible.
Just doing 2 is impossible, doing 1 and 2 even more so.
And thatâs not even taking into account all the gear permutations, plus the idols.
Since almost all exploits come from players finding out unintended interactions between skills and/or passives and/or gear, itâs definitely impossible to test even the smallest amount of combinations.
Just look at the numbers. Itâs impossible to test every single thing. Even if you cut down 99.99% of the combinations, that still leaves trillions (actually much higher) of possible combinations.
Iâm not excusing them. Some of the 1.0 things should have been caught before launch, like the profane veil interaction.
But itâs also a fact that itâs impossible to predict everything (it IS a chaotic system) and that every single game (that keeps adding stuff, including games that just add expansions/DLCs) suffers from the same problem (unless their game systems are mathematically simple, like an FPS, and even then itâs not unheard of).
Yeah, thatâs the issue though.
If a node has 5 stages then you test the first and last. First to see if it functions right. Last to see if it scales correctly. No need for the in-between.
As for skills, outside of checking for the animations if thereâs no synthesis between them then they donât affect each other, so that reduces overhead massively as well.
The big issue is the different nodes for the skills with different synergizing combinations. As soon as thatâs done though you only need to test the respective uniques actively affecting the function of a skill as well as their general function to see if it does what it tells you to do.
It can be slimmed down tons.
Still a massive undertaking, as all testing is.
You still donât actually grasp the huge size of the combinations. I reduced the problem on the passive tree to assume 5 skill points per node simply to reduce the total number of possible combinations.
Furthermore, testing with uniques isnât enough. PV didnât care about uniques, only minion health.
But even if it were just uniques, thereâs way way more than a million possible combinations with the ones we have now. Even more when you consider sets as well.
Yes, you can reduce a ton. But if you reduce the sun by 1 million, itâs still bigger than earth. Some numbers are just too big to handle.
And, again, most exploits come from things you never even thought of. The immortal exploit in PoEâs scourge came from a unique that barely anyone used or even remembered and one of the new currencies giving negative resistances. That, along with a few other things, gave birth to a build that couldnât die. You could go to Sirus or ubers, go out to the club, come back home, go to bed, and in the morning youâd still be there facetanking them.
In hindsight, most of these exploits seem like theyâre simple and should have been caught easily. But the sheer amount of combinations on a chaotic system actually makes it a fairly common issue.
You can say with 100% certainty that no matter how much you test, some player will find some way to make your interaction work either in a completely unforeseen way or in a much stronger way than you anticipated.
And Iâll even go further and tell you that there are currently exploits in LE, PoE, D4, etc, that no one knows about because no one has thought to try them out yet. But theyâre there.
And this is why you arenât a manager. Iâm not even entirely sure youâve had a job. Itâs also why US companies are shit to work for. You want workers/serfs, not employees.
Iâm not entirely sure which would be worse, him trolling or being serious.
He thinks that developing games & coding & stuff is easy, he has no idea of what itâs actually like nor how long stuff actually takes. So his interpretating of their skill level, based on his âawarenessâ of his own skill level & the complexity of the system/task is way out of whack.
Youâll just confuse the poor dear if you use decimal points instead of commas for thousand separators. Iâm not even sure heâs heard of the country you come from, he probably things France is a town in the US (& to be fair, it probably is).
Iâd imagine that working out what to cut out is probably not an entirely trivial task (to actually do it, not just theorise it)?
There isnât one, but please stop saying Iâm French 
Sorry, I forgot you were Swiss (French Swiss, obviously!).
At the very least you could say Iâm French Canadian, at least those are friendly people as well. And there is a very large portuguese population in Canada.
Yeah⌠but I donât take that hyperbole there serious, the number is a mess and the example also was one.
Once more, you combine several stages of testing together commonly for such stuff and solely test for synergistic situations.
For example a node which gives 5% attack speed is tested if it does, then if it scales⌠then never touched again since after all⌠it functions! It has no synergy. All the possible combinations for it are hence â0â and unless the overall code implementation of letting the effects be added together is messed up has no realistic faults.
If you go by those general rules youâll miss some âoddâ bugs but not the obvious ones which youâll get guaranteed flak for (like a node scaling 10 times more then it should). Also⌠you reduce the magnitude of your needed testings related to your number by⌠a quadrillion⌠which is kinda a big deal
And yeah, youâre right, itâs still a lot, but it makes the difference between a realistic testing environment and just giving up since it means jack to even try it in the first place
And as for the âodd exploitsâ⌠that was never under active scrutiny in the first place, thatâs a given. Anyone complaining about them existing and not being found can be simply educated about the situation and if it keeps up ignored.
On the contrary itâs more then acceptable if someone points out that a singular node is broken or a direct synergy is not functioning. Those are the first-layer tests, those are basic, those shouldnât exist⌠and if they exist then at least only 1 which can be explained by âwhoops, forgot about that one, sorry!â which is dealt with the following: Hotfix in a hour, done, never heard again.
Ah, but that is where you are wrong: because sinergies emerge where you least expect them.
For example, as I said before, the immortal Scourge bug in PoE. Iâm sure they tested the new currency, saw that it decreased resistances as expected and then, according to you, never needed to touch it at all. However, it interacted with an item no one used and barely anyone remembered in such a way that you now couldnât be killed even if you tried.
So, using your example: you have a passive node that give you attack speed. You then have to test it with every combination of everything that also uses/mentions/interacts with attack speed. Because you can easily break the game if you donât pay attention to how everything stacks.
And if you then test that attack speed along with a unique that also cares about, letâs say crit chance, you also have to test those combinations.
The number I gave was not hyperbole. It was actually a reduction of the real number. But if you actually want real numbers, Iâll give you a better example:
There are 19 unique helmets, 21 unique chests, 25 unique gloves, 14 unique belts, 23 unique boots. Without even caring about any other gear slot or any other rarity, that alone is over 3 million possible combinations. And that would have to be tested against every single build to make sure there arenât any exploits happening.
Even if you only considered established builds and donât try to find new exploit combinations (which will happen), thatâs over 500 million combinations to test. And as I said, this doesnât even take into account sets, rares, etc. And the PV exploit was done with just rare affixes.
Really, youâre trying to reduce a chaotic system (a hugely chaotic system) to a simple and predictable one and it canât be done. Even the 3 body problem, which only has 3 moving parts, is unsolvable.
Yeah, they absolutely do! But we donât have the complex massively layered synergies like PoE has.
Also not the intricate testing environment which PoE does.
Still⌠you generally make a database with the specific values and interactions, basically helping the work of a tester for those things. Value-wise you can automatically check. Add the new thing you add into the equation, enforce limits, let the debug program check if it breaks any specific limits which it shouldnât break. Not manually done in the first place⌠at least when that database and debug program is setup for it.
Until then? No chance in hell you will catch something like that in the first place, why even bother outside of a quick general test? Not worth it.
Since your example is an item nobody touches⌠thatâs exactly the place when that debug program would catch such things which the automation attached to it, you use tools to improve the work of humans after all since both do different things well.
Which you shouldâve in the database and properly tagged. Which equipment spot and class? Hence which other items and which skills can it even interact with? That reduces overhead. You wanna reduce overhead. You donât need to check a primalist chest piece for a acolyte build, would be senseless. Combinations which simply canât happen.
Like this you wittle situations down. You only set a flag for checking cross-usage of affixes if a item influences another type, then have the limiters⌠and then the âodd ones outâ are solely left, something which has unintended interactions you didnât think would even happen and despite proper tagging doesnât raise any automated flags in the first place⌠which are generally few.
So no, the actual tested interactions are mostly âdoes it generally functionâ and 'do I see something which has been entirely overlooked but feels âwonkyâ â by then.
It can be done, it depends on the testing environment. Testing environments take time and resources to set up, especially time⌠and are often simply not done to ridiculous degrees as someone with OCD like me would do. Not worth the effort, just fix a issue when it comes up, more important stuff to handle before.
Impossible? Nah⌠definitely possible, good to do earlier then later since long-term it saves time, short-term youâll likely need more time and hence canât do it.