Game QA

The Challenges and Illusions of Developing AAA Games —— Part 4

In this article, we'll delve into the challenges and complexities of developing AAA games, focusing on the pursuit of quality and the importance of stage reviews.

Regarding Quality in AAA Games

The pursuit of quality in AAA games is an elusive goal. The definition of game quality varies from person to person; some people appreciate impressive graphics, such as in the Crysis series, while others prefer innovative gameplay, as seen in various indie games. Unfortunately, for AAA games, meeting all these quality requirements is essential.

Achieving sincerity is challenging: The primary condition for AAA games is exceptional production, which means sincerity. Sincerity is not easily attained; it requires a lot of in-depth and meticulous work, often in areas that ordinary players may not easily notice. For example:

The thrilling cutscenes in the Call of Duty series feature large-scale, interactive scenes with advanced AI teammates (often scripted). Every time you play, it feels remarkable, whether as a player or a developer. Having previously participated in the development of the Splinter Cell series, I know how difficult it is to develop interactive cutscenes.


In short, you need:

High-quality animations, are not simply achieved with motion capture but require a lot of experienced animators' meticulous work, and post-adjustment, and heavily rely on the accumulation of industry experience. Effective camera settings that are user-friendly, able to showcase large-scale scenes, and maintain operability so that players can control them well. Innovating control methods is not easy, as they need to be simple and easy to understand for quick player adaptation while also refreshing and original, which is inherently contradictory.

Boundary condition protection for scripted AI is another easily overlooked aspect. Since cutscenes are interactive, the player's actions will significantly influence the AI. As players are unpredictable, scripted AI needs many restrictions and handling in various exceptional situations to perform normally, which is labor-intensive and difficult to cover comprehensively.

Specialized cutscenes require many specially developed features, which are often not universal and can only be used in specific scenes, for one-time use. These cinematic interactive cutscenes also raise higher requirements for tools. Through the serialized development of Call of Duty, many features may still be utilized in sequels, which can be considered a form of investment protection.

Breakthroughs are rare: Players often expect to see technical breakthroughs in AAA games, but these are not easily achieved and require a strong team and a bit of luck. Here are two simple examples:

Ubisoft's Prince of Persia introduced the time-reversal feature, which was astonishing when first seen. This is one of the few significant breakthroughs at the logic level, with related GDC sharing available for reference. Any game at any time has many states, and to implement time reversal at the gameplay level, past states need to be recorded and then replayed. For some games, like Braid and other small games, the number of states in the game may not be too many, and forced recording is possible; for sports games, there are not many states, and recording all rendering commands for replay is also feasible. However, Prince of Persia is a large game with numerous real-time states that are not easy to list entirely. Just look at the size of save files in games with real-time saving features to understand how difficult it is to save a game's complete state at any given moment. Yet, Prince of Persia achieved this and successfully integrated this engine feature into gameplay, making it a fascinating core mechanic.


Ubisoft's Splinter Cell 4, developed by a team in Shanghai, was probably the strongest console game development team in China at the time. At that time, Xbox 360 did not have a development kit, only performance specs. A certain genius developed an interactive smoke effect to be used in the game. In 2005, this was groundbreaking (even today, there aren't many games with interactive smoke effects). However, when they got the Xbox 360 development machine, they were surprised to find that its performance was far lower than expected, making the effect unusable. Of course, luck is also part of strength, so there's no point in blaming the console; it's better to focus on developing usable effects.


AAA games are about spending 80% of the effort on 20% of the features.

In my previous company, we called some particularly memorable scenes "Wow moments," meaning moments when players would exclaim "Wow" during the experience. Wow moments are not easy to create, especially when players today have seen and experienced so much. Just look at the movie industry: 20 years ago, the liquid metal robot in Terminator 2 was amazing, but now people have become numb to any special effects. The development of games is just as fast, and it's now challenging to surprise anyone. When Doom first came out, people were still playing 2D RPGs and were shocked by pseudo-3D games, but in today's age where Doom can run on a watch, would you still have high expectations for an FPS?

For game developers, it is becoming increasingly difficult to exceed players' expectations. You are not smarter than others, not richer than others, and not even more hardworking than others. It is already difficult to surpass expectations with a flash of inspiration, and modern AAA games rely more on overall high production value to succeed.

Regarding Stage Reviews

At certain stages, it is necessary to participate in reviews and accept challenges from the leadership. Reviews are a double-edged sword, both loved and hated.

Due to the need for additional versions for the review, development resources are wasted:

- Some features are needed in the future, but for the current version, compromises have to be made and developed in advance

- Some features can't be completed in time, so temporary versions need to be made for the review, as leaders can't rely solely on imagination for gameplay

Stabilizing a version involves a lot of work, which I describe as road repair, eventually smoothing the road. After the review, restarting the development of new features often feels like digging up a freshly repaired road, and within a few days, the version becomes unstable again.

The extra overtime spent will also cause downtime after the review, as the relaxation of pressure will result in a certain loss of work efficiency.

The review system is lengthy and costly.

Although it has many drawbacks, it can help the team clarify their direction and thoughts, help leaders see the potential of the version, and allow for broader communication within a larger scope. Different teams and people with different functional attributes can engage in more cross-border exchanges. Until a better way to control project quality is found, a standardized review system is still helpful.

You attend the review with the latest version of the development, facing the most stringent criticism from the bosses. Since this AAA game is so important within the company, leaders from various departments, both known and unknown, attend the review meeting, trying to witness the birth of a miracle. The review begins, and the leaders seem thoughtful, casually glancing at the PowerPoint presentation, their glasses reflecting a mysterious light, just like your nervous heart. After going through the PowerPoint, they watch the version and then launch a collective attack on you:

Recon, refers to scouting and looking for weak links, which is a necessary prerequisite for launching an attack. The leaders patiently look for weaknesses in your project, carefully testing your bottom line. Is this feature not completed yet? Can it be done in time? Have players' tastes changed? Is it difficult to commercialize this direction? Is the frame rate a bit low? How can it compete with rival products? Why haven't the features mentioned in the last review been implemented yet? Why aren't the fonts aligned on that PowerPoint slide?

Be very careful in dealing with Recon, as it does not fully describe its cruelty. The Chinese term "pulling threads" is more accurate, as it refers to pulling out more threads when finding one on a sweater. In this scenario, leaders work together to patiently pull on a thread and then seize the opportunity to pull out even more threads, exposing more problems.

To deal with Recon, you must cut through the mess quickly. For example, if the issue is "the fonts on that PowerPoint slide are not aligned," it could lead to discussions about work attitude, team initiative, or even inadequate art design. Do not get caught up in the details; decisively tell the leaders that it was a temporary worker who made the slide and has been let go. As for the low frame rate, you can have the developers multiply the displayed FPS by 5, so when the leader brings it up, you can say, "Look here, our frame rate is already 180, boss, your eyes must be tired from working so hard, I truly feel for you," and quickly change the subject while the boss wipes away tears.

If you don't defend against the attacks and get entangled in multiple threads, you'll only be trampled upon. Your version is riddled with flaws, your progress is a disaster, the gameplay lacks originality, and the so-called innovation is just a joke. Can you live up to the work we've entrusted to you? Can you live up to your team? You have failed the nation's training.

At last, the fatal weakness is discovered, and the bosses' pent-up anger erupts, unleashing their ultimate attacks.

Assume a defensive stance, take notes diligently, and accept that you have no way to fight back. Strive to minimize the damage, humbly accept all opinions, and discern which ones are valuable and which ones aren't. In the future, seek opportunities for in-depth communication with each boss individually. Since there are so many bosses present, those who can be leaders must have high combat power, and arguing directly is a surefire path to defeat. Preserve some lifeblood and return, perhaps next time you can launch an ultimate attack.

In some cases, the leader won't trample you because they haven't found your weakness, but their overall impression is off, so they'll still circle you, trying to unravel another thread. Generally, an overall negative impression is the most challenging to address since intuition is often accurate, and the game isn't enjoyable even if everything is done correctly. If they can't find any flaws after an extensive search, the leader can only symbolically step on you a few times. This situation is perilous; if there are no weaknesses to improve, there's no hope. If you can't identify the problem yourself and even the wise leader can't find your weakness, then you're truly hopeless.

Although you've disappointed the leader once more, they still hold high expectations for you and your project. Next, they'll provide support to help you grow.

You can refer to the second article in this series: The Challenges and Illusions of Developing AAA Games —— Part 2, where they'll assign you a TD (Tech Director) or CD (Creative Director), incorporate a distinguished foreigner into your team for guidance, and establish a stricter milestone plan to foster growth.

Only by employing both the "carrot and stick" can the art of management be displayed, so the leader will do their utmost to assist you, and you have no option but to devote yourself to the art once again.

Whether a project can successfully pass each review is crucial to its ongoing success. Based on previous experience, successful projects generally surpass leaders' expectations in each review. This is easier said than done, which is why we witness so many failed projects.

To be continued - next time, we'll discuss the grand finale!

Latest Posts
1Navigating the Road to Success in Designing Distributed Systems This article discusses the author's various endeavors in the realm of mobile gaming to implement distributed systems and outlines the definition of services, the development of the overall framework, and the process of internal service segmentation during these efforts.
2A Case Study on Debugging High-Concurrency Bugs in a Multi-Threaded Environment The article covers the debugging process for core dump issues, memory leaks, and performance hotspots, as well as the use of various tools such as GDB, Valgrind, AddressSanitizer, and Perf.
3A Comprehensive Guide to Using Fiddler for Mobile Data Packet Capture In this article, we will primarily focus on how to use Fiddler to capture data packets from mobile devices.
4Android Performance Optimization: Best Practices and Tools This article summarizes the best practices and tools for optimizing Android app performance, covering topics such as render performance, understanding overdraw, VSYNC, GPU rendering, memory management, and battery optimization.
5A Comprehensive Guide to Android NDK Development with Android Studio This guide provides a step-by-step tutorial on how to set up and use the NDK in Android Studio, covering everything from environment configuration to writing and compiling native code.