← Back to Notes

Cards Worth Keeping

/ Memory Systems Group

A Memory Systems Meetup recap under Chatham House rules

What happens after the honeymoon period with spaced repetition ends?

This call circled around that question: what to do with bad cards, stale decks, life bottlenecks, shifting priorities, and the uneasy feeling that if something isn’t in Anki, it will vanish from our minds forever. Along the way, people shared very concrete tactics for “wisdom decks,” social-skill cards, integrative flashcards, quote collections, and more.

Below is an anonymized summary of the themes, ideas, and experiments that surfaced.

1. Everyone starts with bad cards

One participant opened by confessing that they’re only a couple of months into serious SRS use—and already sitting on a graveyard of discarded cards. Their early mistakes were extremely familiar:

A long-time practitioner pointed out something uncomfortable: many of the cards we think are “good” only feel that way because we wrote them by hand. Writing them creates deep encoding, so early reviews feel smooth. Cards written by AI lack that initial encoding, so they feel worse—even when the underlying card quality is comparable. Over time, both AI-generated and hand-written cards go through the same cycle: some become workhorses, most get deleted.

The shared takeaway: making bad cards (and throwing them away) isn’t a bug of SRS—it’s a permanent feature of the practice.

2. When to delete a card

A recurring question was: When do you actually throw a card away?

A veteran user framed it this way: a single memory is not worth minutes of frustration and a persistent drag on your emotional experience of reviews. Some personal rules of thumb that came up:

One attendee described using tooling to search for related cards: before deleting, they check, “Does this hang together with a larger structure?” If it clearly fits into a well-used cluster, they might rewrite it. If not, it’s likely expendable.

Behind all of this is a deeper mindset shift: forgetting is allowed. The goal is not to turn your brain into an index of everything you ever touched; it’s to keep the small subset that genuinely matters.

3. SRS as a force multiplier for real life, not just trivia

One participant noticed that getting serious about “what deserves a card” forced them to confront their actual bottlenecks in life:

They started building a “wisdom deck”—cards about principles, mindsets, and behaviors that address those bottlenecks. The key twist:

They don’t grade the card based on whether they remembered the text—they grade it based on whether they embodied it that day.

If a card says something like “Approach people despite irrational social fear,” they mark it as “remembered” only if their behavior matched that idea on the day of review. If not, it’s “forgotten,” even if they could parrot the sentence perfectly.

This shifts SRS from a tool for recall to a tool for alignment:

Others echoed this with similar “life cards”: practical investing principles, reminders about gratitude, social habits, and interpersonal courage. These decks were often described as some of the highest-leverage cards people had ever made.

4. Replaying mistakes—and wins

Another attendee uses SRS for post-mortems on social and professional situations. Whenever they handle something poorly—a meeting, a conversation, a negotiation—they create cards that:

The back of the card is often empty; the “answer” is the visualization itself. Review becomes a structured mental rehearsal, similar to how climbers visualize a route before actually climbing it.

To avoid turning their deck into a museum of failures, they pair this with cards about good decisions and wins—a technique inspired by a well-known researcher who keeps cards on “good choices” to revisit over time. The idea is to:

5. Atomic vs integrative cards

The group also spent time on how to handle abstract concepts (math, philosophy, history, business) and grading them well.

A few shared perspectives:

Atomic cards (one clear fact, relationship, or mini-step) are crucial for accurate grading:

But you also need integrative cards—prompts that require putting several atomic pieces together:

People noted that current schedulers aren’t “content-aware”: if you miss only the 3rd item in a 3-item list, the tool can’t automatically shorten the interval for just that one atomic fact. In a perfect world, failing the integrative card would automatically “ping” the specific underlying atom you missed. We’re not there yet, but thinking this way helps clarify card design.

One practical pattern:

  1. Build atomic cards first (each item or concept on its own).
  2. Then add integrative prompts that ask you to recall the full structure or story.
  3. Grade integrative cards harshly (“any small detail wrong = again”), while letting the atomic cards handle the fine-grained spacing.

6. Quotes as portable mental furniture

Toward the end of the call, the conversation turned to memorizing quotes—not just as aesthetic ornaments, but as compact carriers of insight.

One participant has been reading historical autobiographies in the original language and noticed how often their subjects leaned on memorized quotations (often Latin) as a decision-making toolkit and a kind of moral and intellectual scaffolding.

Their current approach:

The sense in the room was that quotes operate on both conscious and unconscious levels, slowly shaping intuitions and “gut feel” over time.

7. Changing life priorities and stale decks

A recurring pain point: what happens when your life changes, but your deck still thinks you’re a full-time student?

One attendee described shifting from “student mode” (linear algebra, computer networking, formal courses) into “trying to build a software business.” Their backlog was full of cards from old priorities, and they found themselves skipping reviews, lowering daily limits, and feeling guilty that SRS “wasn’t exciting anymore.”

Another person told the story of working on their own spaced-repetition app while still in university. At some point, they had days with 120 math cards due—material they weren’t using in their new full-stack/product role. The solution: they bulk-suspended all math cards without even triaging them.

Four years later, they’ve returned to machine learning and are relearning some fundamentals. They don’t regret the mass suspension at all: it prevented four years of heavy daily review for knowledge they might never have needed. Creating fresh cards now is a small price to pay for that freedom.

A few mental models that emerged:

8. Experiments you might steal

Here’s a quick menu of experiments people on the call are currently running:

Behavior-graded wisdom cards

Post-mortem visualization cards

Positive “good decisions” cards

Bulk-suspend by domain when priorities change

Atomic + integrative layering

Quote decks

None of these are “official best practices.” They’re live experiments from people trying to make spaced repetition serve their actual lives rather than the other way around.

Closing thought

If there was a quiet through-line in the conversation, it was this:

Spaced repetition is powerful, but it’s not sacred. The point is not to save every fact—it’s to build a life you actually want, with a mind that’s stocked with the right kinds of things.

The art, then, is less about perfect schedulers and more about taste: knowing what to remember, what to forget, and when to let a card die so something better can take its place.