Dev Diary: What's in Patch 0.9.3?

Feedback from the Closed Beta showed us that there were game balance issues that needed addressing alongside the addition of new features and content. While the features and content guys worked on hitting their own milestones, we assembled a strike team of three developers to perform a complete rebalance of Path of Exile for the upcoming 0.9.3 patch. In the process of doing this rebalance, it became clear that many fundamental game systems needed changing to take into account feedback. This article explains what we’ve changed so far and why.

• Attributes are now important
• Class identity is promoted
• New passive skill tree layout with keystone passives
• Much better monster difficulty, aided by monster auras
• Single-target attacks are now meaningful
• Combat equations do not care about level anymore
• New damage mitigation equations
• Simpler critical strike system
• Changes to how accuracy is gained
• Support gems have greater impact and modified mana cost scaling
• Implicit mods on base weapons
• Made it far easier to get useful socket combinations on items
• Moving some spell damage onto weapons
• PvP balance
• Unique items with unique flavour and properties
• New skills and support gems
• Changes to the experience system
• New quests in Act Two
• Quest reward revamp
• New Strength/Intelligence item series - sceptres
• Brand new ladder for competition
• Art changes to reduce visual chaos
• Expected timeline for 0.9.3

Attributes are now important

Path of Exile’s core attributes are Strength, Dexterity and Intelligence. Previously, these could only be obtained from the passive skill tree and provided no in-game benefit other than allowing you to meet the requirements of items.

In 0.9.3, attributes contribute to various aspects of your character’s survivability and damage output. Strength grants you bonuses to physical damage and your life total. Dexterity grants you bonuses to evasion and accuracy. Intelligence grants you bonuses to energy shield and mana. These attributes can also now be found on items, in the form of mods like +10 Dexterity. These item mods can be used to chain-equip items that you do not currently meet the requirements for. There will be restrictions to prevent abuse of this functionality.

The results of this change are:

• Characters have an incentive to invest more points in their class’s portion of the passive tree once they have met the requirements of their items.
• Players splashing in other areas of the tree receive additional related bonuses that they may require.
• We now have more item mods, especially ones that directly impact character progression.
• It’s easier to meet requirements of certain items if you want to push your character in a certain direction by stacking the appropriate attribute mods.
• If you find an item that requires 10 Strength more than you currently have, you can use another item from your stash (or buy one from the vendors) to gain the Strength required to equip it rather than being forced to wait two levels.

Class identity is promoted

One of the core foundations of Path of Exile is that any class can use any item or skill if they put in the necessary work to get its requirements. Having said that, we feel that our previous implementation of this system somewhat diluted the class identity of our characters. We investigated several ways of solving this (ranging from strong ‘racial’ skills per class, or maybe class restrictions that were against our initial philosophy), but eventually settled on the concept of starting each character class at a unique place in the (new) passive skill tree.

The result of this decision is that characters are substantially more dependent on their class choice, but still able to play any build they want. If a Marauder wants to take a direct path towards the intelligence section of the tree for some obscure combo build, then he is still welcome to (but suffers the obvious drawback of a lot of wasted opportunity cost).

The new starting layout allows us to push distinct gameplay concepts on each character class at a much earlier level. For example, Duelists could have dual-wielding passives available at much lower levels.

New passive skill tree layout with keystone passives

For a long time, we felt that we could rebalance the game without replacing the passive skill tree. By delaying it, we planned to get this patch out earlier in a hopefully functional way with the old tree still intact. Unfortunately, the amount of changes that require a new passive skill tree layout just became too large, so we’ve had to extend the development time of 0.9.3.

The new passive skill tree has different starting locations for each character class. We’ve learned from feedback from previous generations of the tree (as well as analysis of what skills people allocated points to), and have made several observations:

• We need more passive skills than just 750.
• The passives need to be about 50% stronger than they were in the 0.9.2 tree.
• Players don’t like direct bonuses like +30 accuracy - they prefer percentage improvements.
• Many players exhibit narrow passive skill allocation behaviour where they just focus on one direct way of improving damage at the expense of overall character performance.
• We can create more interesting shapes that properly allow diminishing returns and other neat tension-creating passive skill allocation choices.
• We probably shouldn’t exceed 5 or so passives of a related concept in a group, so that players have to travel between multiple groups in different places on the tree to specialise fully in an area.
• Many of the charge related passives were way too powerful and are being moved to unique items and skills.
• We want to have “Keystone” passives that exhibit massive character-changing properties, such as allowing dual wielding of two two-handed weapons as a keystone Strength passive. We’re placing these very build-defining passives in positions that make it difficult to get more than a few on a given character. These passives are not restricted to a specific character class, but generally synergise with strategies of the attributes they are placed in.
• We can have passive skills that don’t allocate any stats but do give additional core attributes (such as +5 Strength). These allow players to more easily specialise in desired attributes and let us connect skill clusters with low-impact nodes occasionally.

By creating a different start location for each character class, we can control the complexity of concepts that players are exposed to early on. For example, a new Marauder won't be confused by seeing evasion and critical strike passives.

Much better monster difficulty, aided by monster auras

Relative monster difficulty is a very difficult thing to get right, and is absolutely crucial to the how fun the game is. Before 0.9.3, our tools for making higher difficulties hard were monster life and monster damage. Unfortunately, high-life monsters feel very grindy to play against, and high-damage monsters instantly kill you in some situations (which exacerbates the problem with evasion as a damage mitigation strategy).

In 0.9.3, we took a step back and tried a different balance philosophy:

• Normal monsters are relatively easy to kill on any difficulty if you’re at the right place on the player power curve. Single target skills should kill them in a few hits, and area of effect spells should be weaker but should feel pleasingly powerful. If the player is behind the curve, this takes a bit longer and if they’re ahead of the curve they feel more powerful and kill the monsters quickly. The goal is that fighting normal monsters (when they are away from a boss) is generally a rewarding experience that helps the player feel powerful.
• Groups of magic monsters are substantially more difficult and must be approached with caution. In 0.9.3, magic monsters can have at most one mod, and these mods are larger than they were previously (making them relatively impactful).
• Rare monsters (which are now capped at 3 mods and do not always spawn with a pack of obligatory magic monsters) are extremely dangerous and are meant to be on par with a PvP fight in terms of difficulty. Certain combinations of rare monster mods are designed to be devastatingly synergistic and may require the player to decide not to engage the monster.
• Monster auras have been introduced, which can spawn on rare monsters only. Other nearby monsters are granted a relatively powerful property which augments their damage, defense or adds a special property (such as reflecting elemental damage back to you, or giving them substantial life regeneration). In addition to making the rare monster encounters difficult, these auras can pose a large threat when there are other normal or magic monsters nearby. Players need to decide whether to focus on killing the rare monster (to end the aura) or to pick off the normal monsters that have been elevated in difficulty by the aura. Monster auras can of course stack to create very difficult fights.

This approach to difficulty gives us the best of both worlds. Against normal monsters, players don’t feel that the high level areas are too grindy or overly dangerous, and can feel powerful as they destroy groups of foes. Against bosses, the difficulty spikes up suddenly (and can be extremely high if there are multiple bosses or dangerous combinations of mods/auras). This means that players need to be powerful to deal with the high level areas, but do not suffer the drawbacks of the artificial difficulty seen in 0.9.2.

Single-target attacks are now meaningful

A design goal of 0.9.3 is that characters should be able to play in high difficulty areas using single-target attack skills if they want to. These skills have been rebalanced to feel meaningful in combat, delivering substantially more damage per enemy than AoE spells. It will often be correct to deliver area damage initially, before switching to single-target skills to efficiently finish off stragglers. The mana costs of these skills have been balanced around this also.

Combat equations do not care about level anymore

Our previous chance-to-hit, damage reduction and critical strike calculations took the player’s level into account. In many cases, levelling up would actually hurt your character’s performance by reducing your effective damage reduction and chance to get critical strikes. The new equations we use do not care about your character level. When you get a stat, it never goes down over time. Sure, some monsters get harder, but if you revisit lower areas, you aren’t at a gameplay disadvantage.

New damage mitigation equations

There were several problems with the armour system as it stood in 0.9.2:

• At its core, damage reduction and evasion were too similar. Both mitigate a percentage of damage in the long run.
• Due to variance issues, evasion broke down when the player was receiving a lot of damage each hit. They wouldn’t have time to react to drink potions, and if they got unlucky with a steak of failed evades, they would just die in a second.
• Having energy shield be such a large part of a character’s damage mitigation caused players to run far away when it was getting low, due to having no way to recover it in combat.
• The relative amount of armour gained from base items was too high. This meant that characters became very dependent on immediately upgrading to the next available armour piece as soon as it started to drop.

In 0.9.3, the relative importance of armour pieces is slightly lower. It’s still an important game system, but doesn’t dominate other systems such as the actual mods on the items. We also reduced monster damage around the board. At higher levels, evasion characters do not instantly die when they get unlucky against bosses.

In addition, we replaced the armour equation for damage reduction. The percentage of damage reduction now takes into account the size of the actual damage taken. This means that a character with Strength armour can tank a substantial amount of small hits (they’re mitigated a lot), but less large hits (they’re mitigated only slightly). This is in line with how we want the class to play - surrounded by swarming monsters with good survivability. Against high damage monsters, it’s actually desirable to have evasion instead of damage reduction.

These changes create interesting gameplay specialisation for each class:

• Rangers don’t like to be swarmed by foes (but they’re generally at a distance or using melee hit-and-run tactics).
• Marauders are happy to tank an entire group of monsters at once, but should be careful around monsters with high damage output.
• Energy shield is relatively unchanged, but due to changes in the relative importance of armour pieces, they have more life (relative to their armour) to fall back on when the shield is depleted.

Simpler critical strike system

The previous critical strike system (where it was an arbitrary rating that translated into an estimated chance-to-crit) was confusing and generally problematic. The new system involves skills and weapons having a specific chance-to-crit (for example, 5% or 7%), with the character’s critical strike chance gains coming from percentage improvements to this value. For example, if you have +50% critical strike chance from your passives and +50% from your items, then your base 7% would be doubled to 14%. This is much easier to understand and thankfully allows us to remove a confusing rating value from the game.

Changes to how accuracy is gained

Because accuracy is a rating, you need to get an increasingly large amount of it to hit monsters as you level up (because their evasion also escalates as the monster level increases). Although this core system is fine, the way we were giving it out before was pretty bad. Due to needing several thousand accuracy at higher levels, we had to give it out in large chunks on item mods, passive skills and the accuracy support gem. Players who didn’t know they needed it missed way too much, and players who wanted to excel at accuracy found it easy to get way too much of it.

After thinking hard about what we wanted to achieve with our accuracy/to-hit system, we decided that we want people to generally be able to hit somewhat often, providing that they’re using up to date gear and haven’t completely shunned accuracy. We want them to start at around the right ballpark value, so that we can give out smaller accuracy increases that cause a tangible increase in perceived chance to hit. Because of this, we now grant a lot more accuracy per level and include accuracy in the base properties of many dexterity weapons. This creates the situation that we wanted: characters don’t get frustrated by combat and do feel that they get useful benefits by investing in accuracy

Support gems have greater impact and modified mana cost scaling

Our intention is that players do not augment each active skill with every single possible support gem. We want people to pick a small subset and make that part of their build. To achieve this, we’re making the support gems substantially more powerful and are increasing their mana cost to compensate. In the late game, we want players to find that their skills supported by 1-2 support gems are spammable, and ones supported by 3-4 support gems are very powerful and consume a substantial amount of mana. By freeing up their sockets like this, we encourage players to have several skills used in parallel, to increase available tactical options.

We have also changed the support gems to just scale the mana cost of the affected active skill by a fixed amount per type of support gem(for example, +50%) rather than exponentially increasing the scale (in 0.9.2, the scaling value increased as the support gem levelled up). This should address late game mana consumption issues.

Implicit mods on base weapons

The base weapons in previous versions felt too similar. A level 60 axe had identical damage output per second to any other weapon type (with some scaling for ranged/two-hand weapons). Needless to say, this was too bland and didn’t help promote any relative strengths and weaknesses of various weapon types.

In 0.9.3, all weapon types have implicit mods that are specific to that class of weapon. For example, swords are more accurate, axes have higher damage, mauls stun for longer and staves allow you to block attacks occasionally.

Made it far easier to get useful socket combinations on items

A major problem with items in 0.9.2 is that it’s really hard to get good socket arrangements. It’s very unlikely for them to either spawn with good sockets or to be successfully upgraded with the Jeweler’s Orb currency items.

To resolve this, we have made the following changes:

• Two-socket items are basically the norm now, after level 8.
• It’s very likely for an item to have two linked sockets.
• It’s quite unlikely for items to spawn with more than two linked sockets. Although it scales all the way up to six linked sockets, we want it to still be extremely hard to get items that have more than two sockets connected
• Two-handed weapons can now have up to six sockets, rarely.
• Jeweler’s Orb is being retired due to being too weak for its relative (complexity-driven) rarity. It’s getting replaced by three new currency items which interact with sockets. We are still holding tightly to our philosophy that currency items must modify items randomly rather than converging on perfection. The improvement here is that the new variety of currency items will allow you to choose exactly what aspect of the socket arrangement you are unhappy with (colours, links, quantities).

This means that the value of high-end well-socketed items is preserved, but it’s much more common to get items that have at least a partially useful socket arrangement of two linked sockets. People will still want to reroll to try to get lucky with three or more linked sockets.

Moving some spell damage onto weapons

To help reduce the effectiveness of spellcasters in short-duration rushing leagues, spells have been rebalanced so that a certain degree of spell damage from items is assumed for spells to do maximum damage in the long term. Like melee characters, spellcasters now have to worry about finding appropriate +% damage mods on their weapon. This will hopefully push the new high-spell-damage wands/sceptres/staves/daggers as valuable items to find/reroll.

PvP balance

When testing, we found that our old balance values did not work for PvP situations. Witches and Duelists had an incredible advantage for many reasons and we just weren’t able to solve it in a fair way. The new balance takes into the various PvP situations that we anticipate, so there isn’t too much discrepancy between classes. There are still strong strategies that work well in PvP that causes some classes to be better suited to fighting others (for example, critical hits are great against Marauders because of the changes to the damage reduction system), but these are intentional and do not result in one class being worse against a random opponent. The player’s build is more generally important than the particular character class (though certain classes excel at certain builds).

Unique items with unique flavour and properties

Although we have a design philosophy that unique items should not be the best items in the game from a power point of view (this is reserved for the absolute best magic/rare items), it’s still critically important that we have a wide range of thematically awesome and flavourful unique items that offer special properties that are not available on other items. For 0.9.3, we have put much effort into implementing unique-specific mods that allow us to add a wide variety of interesting uniques to this patch. We are trying to design these around specific niche builds, so that whenever you find a unique item, you feel like you want to design a character around it. We expect to add a decent number of uniques in 0.9.3 itself, and then add more in most subsequent small patches.

New skills and support gems

There are currently 17 new skills designed for addition to the game relatively soon. Some of these will make it into the initial 0.9.3 release and others will be patched in with minor maintenance patches every few days as they become available. We made sure to plan these skills around areas that are currently lacking in the game. For example, there’s an early single-target Strength skill, several new bow skills and a few fire spells that are being given priority. It has been a while since we last added new support gems, and we feel like this is one area of the game that has a way to go before it’ll live up to the promises that the system made. We’re including several new support gems in 0.9.3 and have some pretty complex plans for dramatic skill changing supports that will be added when they’re ready.

Changes to the experience system

Modifications are being made to the experience system to allow players to play a few areas above their level without penalty (while still being penalised for being abusively powerlevelled), and to reinforce design goals of diminishing returns during high-level play.

We’re also rebalancing the amount of experience required for each level gained, the relative monster levels of all areas in the game and the experience curve of skill/support gems.

New quests in Act Two

As you probably know, our current version of Act Two is missing a lot of quests. 0.9.3 includes much more finished versions of several major new quests, such as ones in the Chamber of Sins, Weaver’s Chambers and the major bandit quest chains.

Quest reward revamp

In addition to adding new quests, we’ve used this opportunity to properly sort out quest rewards. We’re not giving currency items as rewards any more (because of obvious abuse issues), but have standardised on a rule that quests that block your progress through an act reward you with a skill gem or support gem. Side quests reward you with either a passive skill point, or a skill respec point (not enabled in 0.9.3). The bandit quests reward you with a different permanent passive bonus depending on which bandit you sided with. Act-boss quests do not reward you with skill gems any more because most people forget to go back to the previous act to collect them. For now they reward you with access to the next act, and later (once we’ve solved abuse issues), they’ll drop really good items the first time you kill them.

We've planned the skill gems that you get from quest rewards so that they are given out in more sensible patterns for each class and with appropriate combinations of AoE and single-target damage gems offered at the same time.

There are no longer 50 gems to choose from for later quests. We limited the options to create more opportunity cost and so that we don’t confuse the players with too many gems to chose from.

New Strength/Intelligence item series - sceptres

We've added a series of one handed Strength/Intelligence weapons - sceptres. These are useful for Templars who want to use shields and help us fill a gap in our weapon layout that hadn’t been addressed before. This is also a good opportunity to mention that we’ll continue to use British/New Zealand spelling for words such as sceptres until we support localisation. At that point we’ll allow players to default to American English, so that they get the expected “Scepters” spelling.

Brand new ladder for competition

In order to get clean feedback on the balance of 0.9.3, we’re moving all old characters to their own “Legacy” leagues (one for hardcore, one for non-hardcore) where they can continue to be played. Players may make new characters in the Legacy leagues if they would like to.

We only want to hear balance feedback from the new Normal/Hardcore league. The purpose of the Legacy leagues is so that people can still play the characters they have become attached to (as opposed to us just deleting all the characters). Because many of the items will now be dramatically overpowered, we have to be very careful when dealing with balance feedback from the Legacy leagues.

Legacy characters will of course get a full passive skill refund because of the new layout. Their passives will need to be allocated again before they can use their items. They will find that many of their items cannot be equipped because requirements have changed. Some gems have changed colour. Many mods will have been rerolled or removed. In general, the Legacy league will be a bit of a mess for a while, but will at least allow people to keep their old characters through the changes we made.

The new leagues will have an entirely new economy, disjoint from the Legacy league. It’s a new opportunity to race up the ladder and prove that you’re faster at levelling than the other testers!

Art changes to reduce visual chaos

Some people found it hard to see what was going on in combat in previous versions. In extreme cases, people would complain of slight headaches from the visual chaos that would occur with intense combat in some areas. We’ve made many art changes that should make it easier to see what’s going on in combat and to reduce eye strain. Characters, monsters and items now stand out from the environment a lot more.

Expected timeline for 0.9.3

This is a very large patch, and although it’s coming together really well, it’s going to be at least another two weeks (which places us in mid-October) before it’s available on the Beta realm (due to the massive amount of Alpha testing needed before it gets exposed to thousands of players). Beta testers with a good history of consistent useful feedback may be selected into the Alpha to help with this testing. Once 0.9.3 is deployed, we will scale up the scope of the Beta in order to involve more community members in testing. Please note that this article doesn't represent the complete set of patch notes for 0.9.3 - there are dozens of other changes that are not described here.

It is unfortunate that there is a large period without a major update on the Beta realm, but the work is substantially faster when we are not polishing and doing quality assurance for two days of each week. We will continue to patch problems in the 0.9.2 series as they are identified. The 0.9.2g patch already has some fixes prepared, for example.

We’re very excited about the upcoming 0.9.3 patch, and feel that it’ll go a long way towards making Path of Exile a lot more fun to play at high levels. Combined with the continued content and feature progress, we’re definitely on target to enter Open Beta in early 2012. If you’re a new reader who wants to sign up to beta, please create an account on our site.