A spaced re-attempt of a previously failed problem fails the same way - what does that tell you?

## The second failure is the informative one A first failure tells you almost nothing beyond "there is a gap here". You did not solve it; there are half a dozen possible reasons, and at that moment you cannot distinguish them, because your judgement is contaminated by having just read the solution. A cold re-attempt days or weeks later is a controlled measurement: the specifics have faded, the remedy you applied has had time to work, and the result is a verdict on the remedy rather than on the original attempt. Which is why a repeat failure is disappointing and valuable in the same breath - it is the only cheap way to find out that your fix was aimed at the wrong thing. The wrong response is to conclude that the problem is "hard for me" and lengthen the interval or drop it. That converts a measurement into a verdict about yourself and discards the diagnosis. ## Locate the breakdown before choosing a remedy The useful question is not *did I fail* but *where did the attempt stop*. Three places, three different causes. **You could not name an approach.** Nothing came back - not the technique, not even the family. This is a retrieval failure: whatever you did after the first attempt encoded a solution, not a cue. The remedy is not to re-read the solution a third time; it is to write the recognition trigger explicitly - the property of the input that should make this technique come to mind - and then to meet the same trigger on several different problems, so the association attaches to the property rather than to one story. **You named the approach but could not build it.** You knew it was the right family and stalled on the invariant, the update order, or what the state should hold. That is an understanding gap wearing the costume of a memory gap. The remedy is to derive rather than recall: restate the brute force, name the repeated work, and reconstruct the improvement from that, so the details follow from the reasoning instead of from memory. **You built it and hit the same defect as last time.** The idea arrived, the structure was right, and the same boundary or initialisation went wrong again. That is mechanical, and it responds to a procedure, not to more study: enumerate the degenerate inputs before writing code, run the smallest and largest legal input by hand, check the interval convention explicitly. This is also the diagnosis people most often mis-file as "I don't understand this pattern", which sends them re-reading theory to fix a checklist problem. ## When the repeat attempt succeeds, be suspicious anyway The complementary trap: a re-attempt that succeeds because you remembered the shape of the code rather than because the technique is available. The tell is that you produced the solution without ever re-deriving why it works, or that you cannot state the trigger. Treat that as unresolved and test transfer instead - a different problem the same trigger should fire on. Passing the identical problem again mainly proves that you have seen it twice. ## What the pattern of repeat failures says about the log A senior read goes one level up. If several spaced re-attempts fail, the entries you had marked as resolved were never verified, and the same is probably true of the ones you have not re-attempted yet. That reframes the log: entries move from *reviewed* to *verified* only when a cold attempt succeeds, and until then your sense of how much you have learned is inflated by an unknown factor. Practically, that means keeping a small verification queue rather than trusting a large reviewed pile, and it means resisting the urge to declare a category closed on the strength of a drill block alone. ## Scheduling after a failure The interval should shorten, not lengthen. A failed re-attempt is evidence that the item was not consolidated, so the next contact comes sooner - and, importantly, only after the remedy has changed. Repeating the same interval with the same remedy is running the same experiment and expecting a different reading. Once it succeeds cold, the interval can stretch again, and the item can graduate to a transfer test on a sibling problem rather than another attempt at the same one. ## The cost side Re-attempts compete with new problems for a fixed number of hours, and they feel less productive because you are not adding to any count. The honest framing is that a failed re-attempt just told you that some fraction of your earlier practice did not stick - which is information you were going to receive eventually, either now for the price of forty minutes or later in a room with an interviewer in it.