At various times, PPC was indeed faster than Intel's finest, either outright or in certain niches (particularly SIMD).
> Also Apple knew the truth since they were always running an x86 OS X for 5 years and knew the numbers and just were fine spreading "Apple Truths."
Consider what would have happened if the Intel switch had happened two years earlier. Take the iBook (then, AFAIK, Apple's best-selling product). They could have gone to Pentium M (the P3-based one), which was at best in the same speed range as the G4s used in Apple laptops, and used somewhat more power, resulting in diminished battery life. Or they could have gone to Pentium 4-M (the P4-based one), which would have been somewhat faster, and used vastly more power, resulting in massively diminished battery life.
Apple also at the time had the not unreasonable expectation that IBM would produce faster PPC970s, and mobile-suitable PPC970s, as was in their roadmap.
The Intel switch happened because it made sense at the time; the Core Duo laptops resulted in better performance vs the G4 and only a modest battery life it, while the high-end desktop waited for the Core 2 Duo Xeons, which resulted in better performance vs the G5. It would not, however, have made sense much earlier.
Obviously, they had MacOS 10 running on x86; NeXTStep did, after all, and it'd never be wise to close off options. I mean, _Windows_ supported PPC, Alpha and MIPS for a long time.