He was pretty specifically talking about the Newton, which produced either decently captured raster handwriting or questionably recognized plain text. You don't want to be the one telling people to use a device in a way which is best described as questionable. The iPhone produces reasonably keyed plain text, or questionably smudged handwriting. Since adding cellular modems to handheld devices was straightforward engineering at the time, that inversion of input is probably what made the iPhone take another five years of work when they returned to the idea after stabilizing OS X.