Ludwig Koch and his method
Ask anywhere how to learn Morse code and one name comes back: the Koch method. Nearly every serious trainer of the last thirty years is built on it. Which makes it worth asking a question that is asked surprisingly rarely: what did Koch actually prove — and what do we merely believe he proved?
We went through the sources carefully before building our own trainer. Here is the honest picture.
What Koch proposed
Ludwig Koch was a German psychologist who, in 1936, published a dissertation on the working psychology of Morse reception. His approach inverted the folk wisdom of his day. Instead of learning the whole alphabet slowly and speeding up — the path that reliably produces a painful plateau around ten words per minute — Koch had learners start with just two characters at full target speed. Only when a learner copied the current set with about 90 percent accuracy was the next character added.
The two core insights have aged remarkably well. First: character speed and overall speed are different things, and the character's sound shape should be learned once, correctly, at its final tempo. Second: growth should be gated by demonstrated mastery, not by a calendar.
What the record actually supports
Here the picture gets thinner than most retellings admit. Koch's dissertation is not available online, and nearly everything commonly cited about it — including the famous claim that his students reached about 12 WPM in under 14 hours of training — comes through secondhand accounts, most influentially a 1997 article by Dave Finley. Basic facts a modern reader would want, like the number of test subjects or a control group, are simply not on the public record. There are also hints the subjects were selected for aptitude.
Just as striking: we could find no independent replication of Koch's results in the decades since — not even from the method's advocates. The most rigorous Morse training research on record comes from elsewhere entirely: the psychologist Fred Keller's work for the US military in the 1940s. His successful "code-voice" method centered on something else — immediate feedback after every single character — and used neither Koch's incremental character introduction nor a 90 percent threshold. Keller later argued that the dreaded learning plateau is largely an artifact of poor training design, not a law of nature.
And one detail will surprise anyone who has used a Koch trainer: the canonical character order starting K, M, R… is, as far as the record shows, probably not Koch's at all. It appears to be a later addition from the trainer software era. Treating it as sacred has no evidential basis.
So is the method good or not?
Both. The Koch method is best understood as a sound didactic bet, not a proven system. Its core — full-speed characters, mastery-gated growth — aligns well with what modern learning science has since established about skill acquisition. Its specifics — the exact 90 percent gate, the fixed character order, the pair-wise introduction — are conventions, and some of them look actively suboptimal today: a learner held at 95+ percent accuracy is being parked, not challenged, and research on optimal difficulty suggests the sweet spot for progress lies a little lower than the tradition assumes.
That is exactly where the story continues: what modern learning research adds to Koch — and how we built it into a trainer. If you just want to start, the practical version of all of this is in our learning guide.