“The year 2024 is coming to a close in the mobile gaming world with a decline of about 6% in user spending, primarily due to a reduction in disposable income. When diving deeper and comparing Israel to the rest of the world, the decline is even more significant,” said Assaf Vaknin, head of apps at Google, during the Calcalist Gaming Conference in collaboration with Google and Playtika.
However, he noted that the decline is not particularly concerning, as it mainly stems from the hyper-casual games category. According to him, the key point is that “Israel is still responsible for 5% of global downloads.”
2 View gallery
Playtika co-founder and CEO Robert Antokol following the company's Nasdaq IPO
(Photo: Reuters)
“When examining the sources of the significant decline, we see that the main drop is in hyper-casual games but also in racing and simulation games. People download games to relieve stress, so the reduction in downloads in these categories in Israel is understandable—we’ve had enough action this year. On the other hand, there’s been an increase in puzzle and word games, casino games and card and board games. The major events this year include Playtika’s acquisition of SuperPlay, Plarium’s acquisition, Crazy Labs reaching 7 billion installs in recent weeks and VGames raising $142 million last weekend. Despite the successes, the challenges won’t disappear.”
What needs to be done to take the industry to the next level? Vaknin identifies three main opportunities:
Expanding beyond the U.S. and Western Europe: “Israeli companies are very focused on the US and Western Europe but are leaving users and revenue on the table in Asia. Companies that broke into two leading markets like Japan and South Korea, such as Roblox, have leveled up their revenues.”
Hybrid monetization models: “Revenue from both advertising and in-app purchases. I’m happy to see quite a few Israeli companies beginning to adopt this model. With tailored game economies, it’s possible to make a meteoric leap in revenue.”
Leveraging celebrities and building brands: “The impact of such collaborations lies in creating a stronger connection with users. Israeli creativity has room to take this concept further to increase revenue.”
Danny Walawski, a researcher at Google Research, spoke about a paper he published with Yaniv Levyatan, Moav Arar, and Shlomi Prochter, explaining how to turn a model for generating images into a game engine for computer games.
He explained: “DOOM is an iconic game from 1993 that popularized the FPS genre. The plot is very simple—a soldier fighting demons that have taken over a base on Mars. It featured rich and detailed levels, doors you could open, stairs, monsters with different abilities, and an interactive environment where you could shoot barrels or get hurt by acid. What made it possible was a game engine that was considered a milestone in programming and gaming. A game engine refers to the code that runs the game. The engine takes maps of levels and the game’s graphics, manages the game logic and displays everything in 3D,” he said.
“We asked ourselves: what if we ran the game engine entirely on a neural network, end-to-end? That is, all the complex functions of the engine would be managed by the model. The idea is for the model to be created automatically just by watching recordings of the game. Every frame would be generated by the model. We even sent the video to John Carmack, the creator of DOOM’s game engine.”
How can you teach a model to simulate DOOM without writing a single line of code? Walawski explained that instead of a text-to-image model creating images from text prompts, they adapted the model to accept keyboard input and generate the corresponding frames of the game. “If we can run this in a loop very quickly—where the player’s keyboard inputs generate each frame—we can teach the model to predict the visuals automatically, ensuring that what is displayed on-screen matches the keys pressed by the player.”
Get the Ynetnews app on your smartphone: Google Play: https://bit.ly/4eJ37pE | Apple App Store: https://bit.ly/3ZL7iNv
“We needed to provide many hours of gameplay data. The solution was to play the game using a bot. The challenge of automated gameplay has been studied before,” he explained, and they used a bot designed to explore the game as much as possible and reach as many different states. “We optimized the model without writing a single line of code. We published the paper three months ago, and it quickly went viral. We demonstrated that an image-generation model could run a computer game over time without any loss in quality.
“In essence, we transferred an existing game to run on a neural network. What’s exciting about this is that we could use neural networks to generate new behaviors and even entirely new games,” he said.
Walawski outlined additional research directions: “What if we ran this across thousands of games? Could we combine games? Another possibility is taking video scenes from new games—since today, it’s already possible to generate new behaviors solely from images. In the future, we might be able to create entirely new levels this way. There’s an opportunity to develop capabilities that allow us to generate games ourselves,” he said.