The only way I can think of to do it is if all phones and computers had really good voice control systems that nonetheless only understood the standard pronunciation (like early versions of Dragon dictation). The voice system makers have a vested interest in not doing that, but in understanding everyone as they talk. Arabic has tried for centuries, with the full force of a major religion, and yet still the language is splintered into mutually unintelligible dialects.
People talk how they talk, and it's constantly changing. I don't think anyone can control that.
Also, English has many pronunciation variations, and links between many words may be obvious in writing but not as much in sound (breath / breathe). English needs something more of a morphology-based writing system, not a pure phonetic representation (like Spanish).
* No baseline or other orienting structure, so vertical shifts easily change meaning ("ab" turns to "bc"), same for rotations.
* No redundancy to allow for recognizing incomplete / garbled / damaged letters.
Both features are prominent in any machine-readable encoding (like bar codes or QR codes), and are relatively easy to find in any human script.
The idea behind making more compact visual representation of letters has merit. This particular implementation of it is impractical, though. It only looks acceptable on a demo page.
(While at it, I know two well-designed phonetic writing systems: Tengwar and Hangul. Neither is super compact, but both are easy to learn and read, because of the internal logic. Dotsies are way more arbitrary; they don't exploit any connections between letter or sounds they represent.)