The problem with the 603 (not the 603e) was three-fold. It has small split I/D caches, which severely hurt the performance on the 68K emulator, unlike the unified cache of the 601. The 603 also has a rather weak FPU. But what Byrd is talking about is that the 603 was more deeply pipelined to reduce the complexity of the processor, make it cheaper to manufature and ramp up speeds, at the cost of requiring a higher clock speed to do the same work. We all remember the megahertz myth from the G4 days, but it really started with the 603. Although it's not quite THAT bad, assuming half the clock speed 601 equivalent is not a bad rule of thumb.
Do note that the 603 is not really a bad design, it was just marketed wrong. The 601 was not scalable because of its legacy POWER instructions and its complex internals, and it did not implement the entire PowerPC instruction set -- all of which the 603 fixed -- and the 603e finally added enough cache to make it competitive. But at the time it came out, consumers were used to looking at clock speeds, and when they found out the 603 was not actually grinding through work faster than their old 601s it rapidly gave the chip a black eye. Early Performas that completely lacked an L2 cache of any kind were the worst.
When the 603e came out, the marketing glitch had dissipated, the caches were larger and the chips were faster. And as proof that the 603 was a design meant for the future, the G3 is essentially just a refined 603 with backside cache and instruction tuning (though it still has that weak FPU to this day, which was corrected in the 604 and the G4s).
In this specific case, the 2300 (and all 603 PowerBooks) is actually a 603e, so it's not nearly as dire. However, it will probably not run the software significantly faster than the 601, and its video is no faster either.