Amazon is good at so many things. Why is it bad at games?

Why is Amazon having so much trouble making good video games? Because Amazon's internal structure and culture are optimized to generate mass commercial products, not foster creative entertainment, according to people close to and involved with the company's game efforts.

In April, Amazon essentially announced that 2020 would mark the big coming-out party for the company's video game division. After spending hundreds of millions of dollars and almost a decade building an internal game development operation, Amazon said that this year it would launch its first two big games: a team shooter called Crucible and a massively multiplayer fantasy adventure called New World.

Sign up for Protocol newsletters

Things could have gone worse since then, but not by much.

Crucible debuted in May, was widely panned by gamers and critics for simply not being much fun, attracted abysmally few players, and was unceremoniously yanked off the open market and returned to closed beta in June.

As for New World, in March Amazon said it would ship in May. In April, Amazon said it would ship in August. In July, Amazon said it would ship next year.

The recall and delay have left Mike Frazzini, Amazon's vice president for game services and studios, under considerable internal pressure, according to people close to and involved with the company's game efforts. Frazzini joined Amazon in 2004 and has been the company's main cheerleader for game development for almost a decade. He has not yet shipped a successful major game, a fact that has not gone unnoticed by his boss, Andy Jassy, chief executive of Amazon Web Services, who in turn reports to Jeff Bezos.

The third major project that Frazzini's division has been scheduled to deliver this year is an innovative version of Pac-Man that can be played directly on Twitch, the wildly popular streaming service that Amazon acquired in 2014. The new Pac-Man is not a major "AAA" game like Crucible or New World, but it is important to show the way forward for Twitch as a platform for actually playing games rather than simply watching them.

In May, Amazon and Bandai Namco, which controls the Pac-Man franchise, announced that Pac-Man Live Studio would be playable on Twitch in June. It still isn't. The game's official page now simply says, "Coming soon."

So, what's the problem? In recent years Amazon has become a major force in television and film, so we have seen that the company can succeed in generating popular mass entertainment. Why is the company struggling so badly with games?

Discussing the question with people involved with Amazon Games, some common themes emerge.

Games are art, not just a software product

"We're bringing a lot of Amazon practices to making games," Frazzini told me in March. That isn't working because video games are fundamentally a creative endeavor, not the sort of purely quantifiable mass consumer product or service that Amazon knows how to make.

No less than great novels or films, great top-end games cannot be created through user data requirements, A/B testing, behavioral analytics, user surveys and iterative critiques by departments ranging from security to finance. Yet games must jump through all those hoops at Amazon, according to people in a position to know.

That product development sensibility can work for chintzy mobile games that are made to extract as much money as possible from players but does not work in creating multibillion-dollar long-term franchises that generate not just revenue but emotional loyalty. Instead, thinking of games like tech products just leads to watered-down games without a strong point of view or creative direction.

For example, Amazon executives told me that while designing Crucible they solicited private input from hundreds of streamers and esports figures — people who play video games for a living and definitely know fun when they feel it. So how could the company ingest that input and still churn out a mediocre product?

Turns out, the questions Amazon asked the game pros were generally incremental — "Which weapon do you prefer?" "What classes and enemies do you enjoy?" — rather than stepping back and asking, "Does this overall concept work?" That's why Crucible can feel like it was put together with bits and pieces of other successful games, rather than forging a strong vision of its own.

The entire structure of most successful game publishers is built around protecting and insulating the creative people — writers, artists and designers — from the business. Take-Two does not tell Rockstar what the story of the next Grand Theft Auto should be. Mike Morhaime spent decades shielding the creative engine at Blizzard Entertainment from various corporate owners as Blizzard created StarCraft, Warcraft and Diablo — iconic franchises all.

Many precincts of the entertainment business are run by financial professionals, but the successful ones — whether in television, music, film or games — learn to let the creative people create.

"Amazon is run not even by finance guys but by tech guys who instead of putting their creatives outside the bubble and protecting them from the culture, hired them into the bubble and expected them to work within that confine," said one person involved with Amazon's game efforts. "Amazon culture is great for product, horrible for creative endeavors."

At Amazon, games do not get the Hollywood treatment

It is impossible to imagine Jennifer Salke, head of Amazon Studios, issuing her own version of Frazzini's pronouncement: "We're bringing a lot of Amazon practices to making movies."

That is because when it comes to film and television, Amazon lets people with deep industry experience run the show and acquire projects being made by outside professionals. Salke was president of NBC Entertainment before joining Amazon two years ago. Her boss, Mike Hopkins, who joined Amazon in February, was previously chief executive of Hulu and chairman of Sony Pictures Television. Frazzini, meanwhile, had no significant game industry experience before joining Amazon.

Perhaps as important, Hopkins — senior vice president for Prime Video and Amazon Studios — reports directly to Bezos while the game operation does not. Rather, Frazzini reports to Jassy, and Twitch, Amazon's most prominent game-related business, reports separately to Jassy. Project Tempo, Amazon's secretive cloud gaming initiative — which could compete with Google's Stadia, Microsoft's xCloud and Nvidia's GeForce Now — also appears to be under construction at least partly within the AWS organization.

By contrast, at Microsoft, the head of gaming, Phil Spencer, reports directly to Satya Nadella.

Bezos has clearly appeared to develop an affinity and appreciation for the distinct culture of Hollywood. That seems to be at least one reason why Amazon is letting its video division function on its own while the game division is shoved into the company's standard operating procedure.

Of course there is a long and distinguished history of technology moguls molding their companies around their own personal entertainment interests. The compact disc's maximum length of 74 minutes was set at least partly by Sony executives determined to listen to Beethoven's Ninth Symphony without interruption. Steve Jobs loved music and hated video games, so the world ended up with iTunes instead of iGames.

The troubles at Amazon Games may at least partially reflect the fact that neither Bezos nor Jassy appear to actually be gamers.

At Amazon, games are partly seen as a tool to sell something else

Bezos and Jassy have pushed the company's games operation to make games that are computationally intensive, according to people close to the division. That is, to make games that use a lot of AWS horsepower.

You see, AWS has a large team dedicated to selling cloud services to the game industry itself. The idea is that Amazon makes games that show off what can be achieved with AWS and other proprietary technologies such as the Lumberyard game engine so other companies will want to buy them. In that sense, Amazon's games aren't supposed to attract just players but enterprise customers as well.

The problem is that this mandate creates exactly the wrong incentives if your fundamental goal is to create games that millions of people will love.

A great game starts with a concept, and then the team asks, "What are the technologies and business models that will bring this concept to life and support it?" At Amazon, by contrast, at least some teams are hearing, "Here are the technologies and business models; find a concept to bring them to life and sell them."

For example, internal AWS accounts are known as Isengard accounts, after the fortress in "The Lord of the Rings" eventually controlled by the corrupted wizard Saruman. When internal game development teams use Isengard accounts, they must pay (via internal accounting) the same rates for AWS compute cycles that outside retail customers pay, according to people close to the teams.

Using vast cloud computing resources has allowed those teams to create innovative prototypes, but the costs are then projected to be so large that the company abandons the projects, according to those people. The problem is that when developing a truly new sort of entertainment (which AWS could certainly enable), it is next to impossible to forecast revenue and demand. By driving the technology and client cost calculations so deeply into the game studios, Amazon makes it more difficult for those studios to focus on making compelling games.

There are bright spots

Amazon rebranded the marketing vector formerly known as Twitch Prime as Prime Gaming on Monday. (Even when his product was called Twitch Prime, Ethan Evans, who runs the operation, reported to Frazzini, not Emmett Shear, Twitch's CEO.) Amazon rebranded the service, which offers third-party games and in-game loot, so it could expand to other game platforms besides Twitch.

Even more far-reaching, Amazon appears close to announcing its first major games publishing deal. Just as the company's video operation mostly acquires and distributes TV and film content made by others, moving into game publishing for external developers would diversify Amazon Games beyond its rocky internal efforts.

And despite the troubles, there are interesting projects in Amazon's internal pipeline. The studio in Irvine that is making New World is also making another online game based on "The Lord of the Rings" franchise, though one wonders how well that studio is managing two huge projects at once. A separate internal team led by acclaimed industry veteran John Smedley is working on an unannounced, eagerly anticipated project in San Diego.

As for the new Pac-Man on Twitch? People close to Amazon's game division say it could be released as soon as next month.

An Amazon spokesperson declined to comment.