J.E. Sawyer on Pillars of Eternity Delay

Project director J.E. Sawyer has taken some time on the Obsidian forums to explain the reasons and the timing of the announcement of Pillars of Eternity's delay to 2015, and also took the opportunity to explain what the team is currently working on, which includes, among other things, a restructuring of the ability and talent advancement for all classes, a revision of the current XP rewards system, a review of the game's narrative and some visual polishing of character models and animations.

On why the delay wasn't announced or decided sooner:

I know some people wonder why this was the specific time when we chose to announce our delay. After all, there were big problems in Gamescom/BB build, so why not announce a delay immediately? I think it's a reasonable thing to wonder about, so hopefully this explanation will answer some of your questions.

When you, the individual developer, think there is a timeline problem on a project, it's usually not enough to simply rely on spidey-sense. There are exceptions to this, e.g. if you're working on a small team where everyone has high exposure to almost every aspect of the game. But with a team of 20+ people working on a project at a company of well over 100, gut feelings aren't substantive enough to make immediate course-corrections. What they are good enough to do is start investigating and start planning potential scenarios. When the Backer Beta went out, Adam, Brandon, and I all knew there were major problems, but we needed to quantify those problems in terms of time spent across our team. I.e., how many problems, how long do these problems take to fix, and who has time to fix these problems? We also had found work, which is a general way of bundling those valuable and worthwhile new features and options that backers and internal developers bring up that we think we really should take the time to pursue.

We worked with the OEI owners to quantify all of this work and project it out over the next several months. There's really no point in us hooting and hollering that the sky is falling until we realistically understand how fast the sky is falling and what is required to prevent it from crashing. This took the time between the BB launch and several updates. That gave us burn down rates on bugs, a comprehensive listing and allocation of found work, and time for all of the leads to discuss a realistic timeline to complete the game at the necessary quality level. Of course, we also needed to discuss all of this with Paradox since they are the publisher for PoE and are handling a large number of logistical aspects of completing the game, including physical goods, localization, PR, marketing, and some QA.

We also try to be as general as possible for as long as possible on dates for two reasons a) the closer we get to the end, the more accurate our estimates get and b) nobody likes seeing a ship date shift five times. If we could get all of our estimates right and all of the backed features in and polished exactly on time, that would be ideal. But if I'm forced to pick two of the following three: all promised features, high level of quality, on time -- "on time" is almost always going to be the thing I'd prefer to sacrifice. On many of the projects I've been a part of "level of quality" has been the thing sacrificed, and I've almost never had a say in it.


And on the current work that's being done on the game:

Major things we are working on right now:

* Restructuring class advancement to allow more options, both within the selection of core class Abilities and through the (many) Talents that we implemented based on backer feedback.

* Modifying some class features to be more transparent or just fundamentally work better. Monk wounds (and many of their abilities) were very confusing previously. We've revised them to make what's happening much clearer and easier to use.

* Save/load and other persistence issues. This has been more troublesome than we initially expected and it has caused problems over a much longer period of time than we expected. Save/load issues can be infuriating, so we really need to spend the necessary time on it.

* Memory overhead. Dan Spitzley has already made good progress on this by looking at audio memory issues and inefficient texture use/atlasing.

* Adding in minor bestiary, exploration, lock, and trap XP rewards to increase the regularity of XP rewards across the game.

* Revising how time/turns in the Stronghold work. The turn system (which elapses based on quest progress) solved some problems with time-based stronghold systems in other games (including ones we've made) but it causes some confusion. We're separating "turn"-based resource generation from the random event system, which will be based on the passage of world time.

* Pathfinding. If we could have simply salvaged our previous system, we would have, but that did not appear possible, so we had to fundamentally rebuild it. There are still problems with log-jams between characters attempting to move through the same space.

* Writing/narrative review. On most projects, it's rare to never that we get time to actually review our own writing and consider how major plot moments and information are conveyed in the story. This is also includes Karma Police reviews, where we ensure that the various choices players make have consistent feedback throughout the game.

* Animation feedback and general aesthetics. We just added blended flinches for all hits (ones that don't Interrupt, that is -- Interrupts already have their own animation), which was something we weren't able to do at the beginning of the project. They can be played while a character is in the middle of a reload, attack, spellcast, or even while prone, which helps a lot with feedback.

* Character model/texture quality. It's difficult to balance the needs of a distant camera with the camera views you have in character creation, level up, and inventory, but Dimitri is making his way through the more glaring issues.