Epic Rewrites Launcher with Vibe Coding, Performance Disaster Sparks Developer Mockery

Epic Games recently pulled off something with quite a bit of "comedy value": they released a brand-new game launcher, touted as "the fastest ever," but the promotional tagline carried a self-deprecating joke—"Coding is hard, so we just rewrote it with vibe coding." The line itself is absurd enough, but the final product's performance is even more absurd.

What Is Vibe Coding, and Why Is It Being Mocked

"Vibe coding" is a term that has become popular in developer circles over the past two years. Roughly, it means not writing rigorous logic, not debugging, but relying on feelings and AI-generated code to "program by intuition." It originally had a meme-like quality, used to poke fun at people who quickly cobble together prototypes. By putting it into an official promotional tagline, Epic essentially admitted that this rewrite didn't follow a conventional engineering process.

The problem is that Epic is not a small workshop. It's the company behind Fortnite and the Unreal Engine, and the rendering performance of a launcher's list page should be a basic skill. So when users see the launcher stuttering while scrolling and the interface retaining content from the previous screen, mockery is almost inevitable—one video creator directly lashed out: "You made Fortnite, and you can't even display an image list properly?"

The concept of "vibe coding" was proposed by AI researcher Andrej Karpathy in early 2025. Its original meaning was to describe an extreme way of collaborating with AI: the developer barely looks at the AI-generated code, deciding whether to accept it based solely on whether the running result "feels right," and when encountering an error, throws the error message back to the model as-is until the program runs. It is fundamentally different from traditional "AI-assisted programming"—the latter still requires a person to read, review, and understand the generated code, while vibe coding actively abandons the code review step. This approach can indeed greatly speed up one-off scripts, weekend prototypes, and internal demo tools, but the cost is that code maintainability, edge-case handling, and performance characteristics are completely uncontrollable. Karpathy himself also emphasized that it is suitable for "throwaway code," not for production systems that evolve long-term. By putting this insider meme into an official promotional tagline, Epic essentially publicly admitted that the product took a non-production-grade development path, which is one reason the developer community reacted so strongly.

Actual Experience: "Works Gloriously" on Linux, but Scrolling Is a Disaster

According to hands-on testing in the demo video, the launcher's performance on Linux was instead described as "works gloriously," but for a darkly humorous reason: scrolling immediately exposes performance problems, with frame rates dropping visibly.

Even more baffling is a bug jokingly called a "personal favorite": content you just viewed permanently remains in the UI and cannot be removed no matter what. For a high-frequency interactive interface like a launcher, this means basic state management was not done cleanly.

The performance bottlenecks of launcher interfaces like this usually concentrate in two areas: list virtualization and the rendering pipeline. If hundreds or thousands of game entries are all instantiated and kept resident in memory, every frame during scrolling has to re-layout and re-draw textures, causing the frame rate to plummet; the mature approach is to render only the entries visible within the viewport (virtual scrolling) and reuse already-created nodes. The UI residual issue is mostly related to state management and redraw timing: when list items are recycled and reused, if old state is not cleared during data binding, or invalidation redraws are not triggered correctly, the pixels from the previous screen will "stick" to the interface. Neither of these problems is deep technology, but they require clear performance budgets, real-device testing, and regression test cases to guarantee. For a launcher opened by millions of players every day, they are precisely a direct reflection of engineering discipline.

The Steam Contrast: A Performance Patch Within a Day

What really pushed this to a climax was the competitor's reaction. After the news spread, Steam quickly pushed a client performance update that compressed startup time to around 5 seconds. The video creator's assessment was harsh and precise: Epic relies on vibe coding, while Steam relies on "talent, skill, and dedication."

The reason this contrast spreads is not about who was a few seconds faster, but about the standard of judgment it reveals—when a team treats "coding is hard" as a selling point rather than a problem to be solved, users will immediately vote with their feet. The launcher is the entry point players must open every day, and any stutter will be magnified into a brand impression.

The Real Lesson Here

AI-assisted programming lowers the barrier to writing code, but it does not lower the barrier to delivery quality. Vibe coding is suitable for prototypes, experiments, and internal tools. Once it is placed in a formal product facing millions of players, the cost of lacking tests, performance budgets, and state management will immediately become apparent. The essence of Epic's failure this time is not "using AI," but letting engineering discipline give way to marketing rhetoric.

For developers, the value of this case lies in drawing a line: AI can help you write faster, but performance, stability, and edge cases still need humans to take responsibility. A "simple" interface like a launcher is precisely where basic skills are tested the most.

One-line Summary

Epic's vibe-coded launcher rewrite fails, exposing the boundary between AI programming and engineering discipline.

Paragraph Summary

Epic Games released a new game launcher touted as "the fastest ever," but the promotional tagline self-mockingly said "coding is hard, so we rewrote it with vibe coding," which immediately drew mockery from the developer community. Hands-on video showed that the launcher's frame rate dropped noticeably when scrolling on Linux, and viewed content permanently remained on the interface and could not be removed, exposing deficiencies in basic state management and rendering performance. For a company behind Fortnite and the Unreal Engine, this performance was seen as a failure of basic skills. Competitor Steam quickly pushed a performance patch that compressed startup time to around 5 seconds, further amplifying the contrast. The article argues that the core of the controversy is not whether AI was used, but that the team treated "coding is hard" as a marketing selling point rather than a problem to be solved, letting engineering discipline give way to rhetoric. AI lowers the barrier to writing code, but it does not lower the barrier to delivery quality.