melnorme asked: Josh: There seems to be a consensus now that Pillars of Eternity's late game was too easy. Why exactly did it turn out that way? Is it some flaw in the way you playtest games? What steps are you taking to ensure this doesn't repeat itself in future games? Would you say that, in general, professional game development companies have a culture that encourages focusing on the early game, and are bad at polishing late game content?
There are a few reasons. The first is that late game content can be approached by much more varied parties than early game content. I.e., on the player side of the equation, there’s a lot more potentially going on (in terms of party composition, level ranges, ability/spell/talent choices, and gear) than in the early and mid-game. It’s harder to predict for, test, and balance content when you have to account for more.
With speed runs and progression tests, testers are always starting at the beginning and moving forward. If the critical path gets broken, they often need to start again with a new build. This alone means that the closer something is to the beginning of the game (and the crit path), the more times it will be tested and (potentially) reported on. This also applies to the system content that is being used – high-level spells, the most powerful items, etc. Comprehensive test plans can catch functional problems with these things, but their relative value/balance can be difficult to qualify without playing it in the context of late game content.
The last reason is that the late game content was the last content we implemented. The positive effects of this can be seen in things like the area art. Dyrford and Copperlane were some of the first maps we built. Twin Elms’ maps were some of the last. The negative effects can be seen in the things that really need repeated playtesting and iteration to polish, like combat balance.
On The White March, I think we’ve done a better job of more consistently testing each area of the expansion. That said, it’s a lot easier to do that when the content is confined to a relatively small area. In the future, ensuring that we both have equal play day time and equal tester time on late-game content should make the difficulty more reliable/consistent. I also think that making Twin Elms smaller, content-wise, would have mitigated the inherent risk of testing/polishing so much content so close to release. It would have felt better from a pacing perspective as well.
That said, it’s almost always more prudent for a developer to err on the side of early game polish than late game polish. A game with a strong start and a weak finish (barring utterly catastrophic endings) will probably be more well-received than a game with a weak start and a strong ending. Most players (and reviewers) won’t suffer through a bad beginning in the hopes that it gets better.
