Users demand 'literal match' option as search engines prioritise relevance

A blog post published on Sid’s Blog has drawn attention to the changing nature of digital search, arguing that modern engines have moved away from literal keyword retrieval towards relevance-based recommendations. The author contends that this shift has progressively reduced user control, replacing the precise search skills of the 2000s with a system that often ignores exact matches in favour of semantic guessing and advertisements.

The critique extends beyond web search to include email clients, online marketplaces, and messaging applications. The author cites specific frustrations, noting that searching for an exact model number of an audio-visual receiver on online marketplaces often returns listings for the same brand rather than the specific item. Similarly, searches on YouTube are described as being interrupted by rows of short-form videos and suggested content, with actual results buried in between.

Gmail is highlighted as a particular point of contention, where pasting an exact invoice number or name into the search bar sometimes returns no results. The author questions whether this is a technical glitch or a quiet change in default settings to "Most relevant". While acknowledging the computational expense of full-text matching across billions of inboxes and attachments, the post suggests that relevance-based defaults may be a natural progression for most users, but not for those seeking precision.

The author describes the current state of search as a "dumpster fire" and acknowledges that the term "en-shittification" might be too cynical or simplistic. However, the core argument remains that there is increasingly no mode for users who know exactly what they are looking for. The post contrasts this with the earlier era when accurate keyword usage and search operators were considered hard-earned skills for information retrieval.

To address this, the author calls for a "literal match only" toggle or option. This feature would kill fuzzy suggestions and ignore semantic guessing, effectively running a reliable search across user data. The post is framed as a direct appeal to product managers, with the author stating they would be happy to pay for better search functionality that helps them get what they need instead of optimising a metric.

The author, who is retired in their mid-30s after a few exits, mentions spinning up a version of their decluttering app specifically for YouTube to cope with the poor search experience. They invite collaboration or contact from those building interesting solutions, noting that their inbox is open.