As part of a university study on procedural quest generation, in-depth interviews were conducted on a group of industry experts who had past experience at more than 35 indie and AAA companies. Their backgrounds included companies such as Ubisoft, Guerrilla Games, Bioware, Rockstar Games and YAGER. Alongside this a tool was developed, which was finally tested and judged in a qualitative survey by a second group of experts, The article below describes some of the insights gained.
In today's age of ever advancing technology, rising game scope and complexity, as well as a complimentary surge in player expectations, procedural quest generation is becoming an increasingly interesting topic to think about. Recent game titles have shown that a lot is possible with current day tools and knowledge, but also that some problems have yet to be figured out completely. This has sparked a wealth of research into aspects such as generating rich narrative and dialogue, hiding and preventing system-generated patterns, or adding meaning by ensuring that a legitimate impact is being made on a game world and its characters, each time an automatically generated quest is created, undergone and completed.
A thorough and well thought-out automated quest generation system could bring a lot of benefits to some game titles, such as generating an (in theory) endless amount of interesting content, helping bring to life largely underutilized open world gameplay spaces, or saving precious development resources, so they may be used to benefit other aspects of the game. However, despite these potential benefits, the vast majority of larger game companies currently lays a clear focus on hand-crafted quests. Implementations of procedural content seem to remain mostly limited to smaller and experimental games, or to the less significant and repeatable side-quests in larger titles, where complexity is not as important.
A small number of AAA studios can be seen openly taking on the procedural challenge, with some noticeable examples being Bethesda's Radiant Quests in the Elder Scrolls and Fallout series, as well as the Nemesis system used by Monolith in the Middle-Earth games. This only seems to be limited to a handful of examples though, and the question remains why we're not seeing a rise in similar technologies yet. Are there still too many prominent problems to overcome, or are there other reasons?
Image: Fallout 4, one of the games to use Bethesda's Radiant Quests
As part of a study conducted for Breda University of Applied Sciences, a group of experienced developers were questioned about their opinions on procedural quest design in the industry, tools used currently and in the past, and developments that might be wanted in the near future. The participants had current and past experience at more than 35 game companies, with backgrounds in various design principles (level, quest, system, narrative) and programming. Company backgrounds included some prominent developers of (action-) RPGs, including but not limited to Ubisoft, Guerrilla Games, Bioware, ZeniMax, Rockstar Games, and Jagex.
The study found a large divergence in positivity when questioning developers on industry stances on automated quest generation. Although there was a consensus that many indie developers would likely be open to the idea, people were not in full agreement when the same question was posed for AAA companies, shown by answers ranging into the extremes on both ends.
Digging deeper, a range of concerns were found, which largely explains this variety in answers. The following problems likely need to be solved or at least kept in mind in order to get large companies interested in the concept of procedural quest generation for their games:
Discrepancies using procedural next to hand-crafted content: if a game isn't built from the ground up with proceduralism in mind, this might lead to some unnatural feeling moments
Negative player perception when system-generated quests or patterns are recognized; this can break immersion or cause backlash
Overall low quality of procedural quests, caused by lack of interesting narrative and language, faults and patterns in the generation process, or a lack of meaningful impact on a game world and its characters
Investment not always being seen as worth it. It could take months or even years to create a flexible and usable in-house tool. Most studios would rather hire a team of designers to do the work manually instead
Company ideology: some companies take pride in designing and creating their quest content by hand
Genericness: it can prove challenging to make a tool that is usable for multiple types of projects or different studios, as quest mechanics can be very specific to the game being made
Testing and Quality Assurance: Some procedural approaches don't allow enough control over the design process or outcome, making it harder to assure that quests are tested and meet the proper standards
Despite these concerns, a notable finding in the research was that procedurally generated quests were stated to be especially interesting for long standing franchises, where a multitude of games share many of their quest or mission structures and mechanics in common. Ubisoft games were stated to be a good example of this by multiple of the participants. In these kinds of instances, overcoming the many hurdles might be worth the time, as resources put into development of the tools could be easily returned over the significantly long period of time that the franchise might last.
The study found that no generic tools or methods are currently being used in quest creation processes, aside from perhaps Google Docs and similar documentation tools. Some companies were stated to be experimenting with in-house tools at their own pace, but it is still uncertain how quickly that research will be done, or how fast we will see that technology implemented in the actual upcoming game titles.
A prototype quest creation tool was developed with the goal of finding a mix of proceduralism and hand-crafting that is usable by AAA companies. The concepting phase started after the interviews, and was inspired by the extracted data and suggestions. The final tool allowed users to create a quest by placing quest actors (NPCs, Objects, Locations...) on a 2D map, followed by the creating and stringing together of quest actions and manually written descriptions. The procedural aspect comes in the form of gameplay templates, which could be tweaked and loaded to automatically detect active actors on the map and generate a quest from them.
Image: An NPC quest actor in the QuestCreationTool made for this study
A testing round and survey was conducted with a second group of experienced game developers, most of which were different from the first group. The tool was received positively, and although a lot of feedback points were given, most participants agreed that the core concepts were a good foundation to build on if the application were to be worked out further. The procedural templates were rated best among all of the features presented.
As we move into the future, we will likely see more big budget companies try their hand at procedurally generated game quests and missions. There are still a lot of hurdles to overcome before full-fledged main storylines are feasible, but some people are definitely in the process of figuring out the best ways to do this.
For the time being, anyone wishing to develop their own triple-A quest generation tool is recommended to keep in mind the large preference for manual design that currently exists. Suggested is to focus on a solid hand-crafting pipeline first, fine control and tweaking of generated quests second, and all the randomized or system-generated factors third, preferably placing proceduralism in a complimentary, rather than a foundational role. As technology improves and full-on proceduralism becomes more feasible, it is likely that this preference will change, but as of now this seems to be the best way to go about this.
Goran de Ruiter is a Technical Game Designer and recently graduated MSc Game Technology who is looking for new opportunities.
Get in contact and find his portfolio website at: www.gxderuiter.nl