Show HN: Got sick and tired of AI in search engines, so I made my own (glass.8ball.space)

Project Glass is a new, privacy-first search engine explicitly built to exclude AI/LLM assistance: the creator published a simple, public ranking algorithm and offers per-result transparency so users can see what factors produced each hit. Results are ranked using a weighted mix of title match (0.90), snippet/body match (0.42), click rate on a log scale (3.00), and recency (1.00) with a decay half-life of 365 days. The engine only extracts page title and a small snippet, refuses paid placement, does not store personal search history, IPs, or tracking cookies, and uses a one-time token per result to guard server-side click tallies against simple scripting. For the AI/ML community this is a deliberate counterpoint to opaque, AI-heavy search: it emphasizes auditability, reproducibility and privacy, offering a useful baseline for evaluating the tradeoffs of non-ML ranking. Technically, its strengths are predictability and low data collection; its limits include potentially weaker semantic relevance without AI, sensitivity to click-based manipulation despite tokens, and scalability/anti-fraud challenges for a click-driven system. As a “call to action,” Project Glass underscores that useful search can be built without hidden ML models — but also highlights why many services still rely on more complex, often nontransparent algorithms.