X Nears Completion of Search Code Rewrite

X is close to launching a rebuilt search system, according to a post from Head of Product Nikita Bier. The overhaul addresses growing complaints about unreliable results, which users have described as increasingly unusable amid spam and overload.

Bier explained that the legacy search infrastructure has been “getting hammered by AI agents and has been choking at scale.” The team is finalizing a full rewrite of the original Twitter-era search code, combined with upgraded bot detection. This follows recent efforts to curb junk content, including improved automated spam filters and measures against AI-generated replies flooding conversations.

The small engineering team – reportedly just 30 people handling day-to-day operations – has been prioritizing fixes for spam, AI “slop,” and system stability. Bier noted on February 20 that there’s no single “silver bullet” against rising AI spam, a challenge affecting all social platforms as generative tools proliferate.

The issue extends beyond user experience. Platforms like X rely on user-generated content to train AI models (e.g., xAI uses X posts for Grok). Ingesting more AI-created material risks degrading data quality, creating a feedback loop of less informative outputs. Meta faces similar pressures, balancing promotion of its own AI tools (to justify massive investments) against overuse that annoys users and strains infrastructure.

The rewrite is part of broader mitigation: enhancing bot filters, reducing AI reply spam, and rebuilding core systems for resilience. While too late to reverse AI integration even if that was the intent, X aims to reform internally to limit negative impacts. With users frustrated by degraded search and spam, the upcoming update could restore reliability and trust in discovery – if it’s successful, of course.