Path of Exile - How We're Developing Our Next Expansion Differently

How We're Developing Our Next Expansion Differently

Posted: 20 Oct 2020 01:12 PM PDT

This year has been tough for our team and has thrown a lot of unexpected challenges at us. This has caused us to adjust how we're developing Path of Exile, which will affect what's happening with our December expansion. We want to let you know what to expect and this post covers that in more detail.

From Path of Exile's release in 2013 until late 2015, we struggled to grow the community and were getting worried as the game's popularity started to slowly decline. We tried releases of many different sizes and cadences, before eventually settling into a 13-week cycle with the launch of Talisman in December 2015. Since then, we have developed 19 leagues with this cadence and had a lot of success with it. Path of Exile grew exponentially and allowed us to put even more content into each expansion to meet the expectations of our growing community. Chris even presented a GDC Talk on this process, which was very well-received within the gamedev industry. He still receives mail every week from developers at other studios who feel that the talk was of great value for their teams. Things were going well and we thought we knew exactly what we were doing.

Then 2020 hit and exposed just how vulnerable our development process was to unexpected events. To some extent, we were lucky that a black swan event (such as a key team member leaving) hadn't caused similar disruption to our schedule before this. We want to preface this by saying that the government-mandated lockdowns were not the root cause of the issues, but they had a significant impact and added to an already high-pressure situation. Due to the way we've been developing expansions, we had almost no wiggle room to manage the additional overheads of lockdown. Even under normal circumstances, some expansions were coming in quite close to the wire. There is a reasonable chance that we may experience another lockdown, or some other unforeseen event that adds extra pressure and we need to create a development plan that has enough breathing room to allow that to happen. After two lockdowns, we delayed Heist's release by a week and it was still not enough to mitigate the combination of constrained resources and ambitious development scope, as Heist was by far the highest-content league in PoE's history. (Adding to this pressure, our country's borders are closed which means our international hiring is frozen for the foreseeable future).

Which leads to the next issue - regardless of how difficult pandemic pressures make development, it's genuinely hard to scope out how long a Path of Exile expansion will take to develop. Some systems that appear easy to create end up taking several iterations to get right. Conversely, some things that felt like they'd be really hard just come together quickly and work the first time. Usually these over- and under-estimates average out during the development of an expansion, but sometimes you get ones that are developed a lot faster (Legion) or slower (Delve) than usual. If you categorise Path of Exile releases into the "good" and "bad" ones, you see a clear pattern of times when development took less (or more) time than expected. This shows that correct scoping and risk mitigation is critical to ensuring a good Path of Exile launch.

Another important topic to discuss is that of Feature Creep. This is when the featureset of a piece of software gradually increases over time as developers think of more cool stuff to add, eventually causing production problems. This is a somewhat common problem in software development (for example, there's a boss in Diablo II called Creeping Feature as a nod to this, over 20 years ago). While Feature Creep sounds like a terrible thing, it can often be great for making a game feel special. A lot of the stuff that makes Path of Exile special was added because a developer thought of something cool and worked hard to squeeze it in a specific release. While Feature Creep can wreak havoc on a schedule (and hence the overall quality of an expansion at launch), it's also important to make sure that developers have a way to still add those special touches that make the game feel like it has endless stuff to discover. We feel that this is best done in the planning phase rather than late in development when such changes can affect the quality of release.

Late in Heist's development cycle, we had a serious internal discussion about how we could restructure our development process so that subsequent expansions are less risky. This discussion resulted in an experiment that we decided to carry out for the next three month cycle.

We have defined a very specific scope for December's 3.13 expansion. It contains everything that a large Path of Exile expansion needs, but no more. Chris is personally handling the production of this expansion to make sure that no work creeps in that isn't in the planned scope. The schedule that we will hopefully achieve with this approach will likely have everything quite playable and ready for gameplay iteration before our marketing deadline, and in a very stable and polished state by the time it is released.

The positive consequences of this experiment are clear: if it succeeds, we'll be able to deliver 3.13 on-time, with a strong stable launch, plenty of gameplay iteration and solid testing of features. If this experiment works as we expect it to, we'll be able to continue using it for future expansions which will allow us to continue with our 13-week expansion cycle, which we strongly feel is best for the continued growth and long-term health of Path of Exile in the period before Path of Exile 2 is released.

This experiment comes with some side effects, however. You'll definitely notice that the patch notes are much, much shorter than they usually are. That's because we're focusing on getting the most important changes done, and doing them well. We're aiming to try to fit the patch notes on just a few pages, if we can manage it. This does mean that we have had to be careful to pick our battles though - the balance changes we are doing have been carefully chosen to have the largest impact and fix real problems. It's also likely that we'll front-load the announcement to have more of the expansion's contents revealed at once, reducing the number of small teasers we post in the weeks following announcement.

Our goal is that 3.13 takes 50% of the overall development hours of Heist (which means going from a situation with overtime to a situation with testing time), and yet feels like a large December expansion. If you're interested, it's an Atlas expansion (like War or Conquerors) with an in-area combat league and a few other bits and pieces. We'll also be announcing it in a slightly different way than we usually do. Stay tuned!