OpenDyslexic is a typeface designed from a specific theory of dyslexia. Its letters have weighted bottoms — heavier at the base — on the premise that dyslexic readers confuse letters by mentally rotating them. A 'b' flips into a 'd'; a 'p' becomes a 'q'. Gravity anchors prevent this: the heavier base creates visual asymmetry that resists rotation. The logic is clean. The implementation is careful.
The font doesn't work.
Controlled experiments found no improvement in reading rate or accuracy for students with dyslexia. In alternating treatment designs comparing OpenDyslexic against Arial and Times New Roman, OpenDyslexic produced equal or worse performance across measures. None of the participants preferred it. The font designed to help was rejected by the people it was designed for.
The failure isn't in the execution but in the premise. The visual confusion hypothesis — that dyslexia is primarily about seeing letters incorrectly — has been progressively displaced by phonological models. The dominant understanding now locates the difficulty in mapping written symbols to sound units, not in distinguishing the symbols visually. Most dyslexic readers can tell a 'b' from a 'd' when shown them in isolation. The problem surfaces when they need to associate each letter with a phoneme rapidly enough to sustain fluent reading.
OpenDyslexic solved the wrong bottleneck. It engineered against visual rotation when the constraint was phonological decoding. The weighted bottom that was supposed to anchor meaning instead introduced an unfamiliar shape that slowed recognition without compensating elsewhere.
The design lesson is uncomfortable: an intervention built on an intuitive model of the problem can be worse than no intervention at all. The font worked perfectly — for a version of dyslexia that doesn't dominate.