Log suspected patterns with dates before concluding — memory overestimates frequency
Record each occurrence of a suspected pattern with date and context in a dedicated log before drawing conclusions, because memory-based frequency assessment systematically overestimates recurrence through the frequency illusion.
Why This Is a Rule
Once you suspect a pattern, the frequency illusion (Baader-Meinhof phenomenon) activates: your brain starts noticing every occurrence that matches the suspected pattern while ignoring non-occurrences. You buy a red car and suddenly see red cars everywhere — they were always there, but your pattern-detection filter is now primed. The result: you massively overestimate how often the pattern actually occurs.
Memory-based frequency assessment is doubly unreliable because it's biased toward vivid, recent, and emotionally charged instances. You "remember" the pattern happening "all the time" when the actual frequency is far lower — or the pattern is genuinely real but you've inflated the count.
A dedicated log with dates and contexts provides the ground truth that memory cannot. Before concluding "this happens constantly," count the log entries. The data often reveals: the pattern is real but less frequent than it feels, or the pattern is frequent but occurs only in specific contexts (which points to the cause), or the pattern doesn't actually recur and the frequency illusion created a false perception.
When This Fires
- When you suspect a pattern is emerging (behavior, reaction, system failure)
- When you're about to conclude "this always happens" based on memory
- During any observation where frequency assessment matters for the conclusion
- When tracking a hypothesis about recurring events
Common Failure Mode
Drawing conclusions from the feeling of frequency rather than from data. "This person always misses deadlines" — but when you check the log, they missed two out of twelve. The feeling of "always" was generated by the two memorable misses, not by the ten unremarkable on-time deliveries. Without the log, the feeling becomes the fact.
The Protocol
When you suspect a pattern: (1) Start a dedicated log — even a note titled "Pattern: [name]." (2) For each occurrence: record date, context, and what specifically happened. (3) Do not draw conclusions until you have 5+ logged entries. (4) When reviewing, count the entries and check the contexts. Is the frequency as high as it felt? Are the contexts revealing a structural cause? (5) Conclusions drawn from logged data are trustworthy. Conclusions drawn from memory-based frequency assessment are not.