My Advice After Deleting 6.6 Million Reviews in Anki

Deleting 6.6 million “ghost reviews” in Anki is the cautionary centerpiece: nearly all of the time spent studying ended up tied to flashcards that no longer exist—or no longer matter. After checking Anki’s stats, the creator finds a history of 6.6 million reviews tied to flashcards that were later deleted, with those removed items representing about 93% of the collection. Only around 5% remains in a main profile, while a separate phone profile keeps roughly 100,000 cards. The practical implication is stark: the study time wasn’t just inefficient; it was largely evaporated, leaving behind a record of effort that can’t be used to build real-world knowledge.

The waste shows up not only in review counts but also in time. The creator estimates more than 3,300 hours spent on reviews alone, and argues that creation time must be added too—especially because many cards were manually built, including low-quality, low-effort exam prep cards created over years. The problem wasn’t simply “too many cards.” It was the workflow: using Anki as a cramming engine, then deleting decks after exams once grades were secured. For safety, decks were sometimes exported to files, but once deleted, the knowledge and review history effectively disappeared from ongoing learning.

A second driver was the mismatch between memorization and applicability. Early on, the creator used shared decks and also generated cards using programming, then later shifted toward learning vocabulary in foreign languages—often obscure terms. Over time, those older cards became “worthless” because the knowledge couldn’t be applied in real life. The result is a long-term trap: flashcards can produce impressive recall during a “honeymoon period,” but if the content isn’t relevant to future use, the system becomes a time sink.

The advice that follows is framed as an optimization principle. More reviews do not automatically mean more knowledge; the creator calls that assumption backwards. Success should mean the minimum number of reviews needed to retain knowledge that is actually recallable, usable in real situations, and “cheap” in time cost. Instead of asking “Can I memorize this quickly?” the creator urges learners to apply a threshold test: if a card’s information won’t be applicable later, it shouldn’t stay in Anki. In school or medical contexts where cramming is unavoidable, the goal should be “smart cramming”—the least repetitions possible—rather than building a massive deck that later becomes dead weight.

The message lands as a personal audit: after nearly 16 years of mistakes, the creator treats the deleted decks as evidence. The takeaway is to continuously question each card’s real-world value before it accumulates into millions of reviews that can’t pay off later.