Canikon MUST switch to EVFs

No, people buy whatever fits their needs. How do you explain Olympus/Panasonic/Sony's mirrorless camera sales? Clearly, people were attracted by the relatively large sensor being offered in a smaller package, and were willing to suck up the disadvantages (no OVF, no PDAF) to get that small size.

Look - adopting a technology that's inferior in many respects based on the promise that future iterations of that technology will bring improvement isn't a sound plan at all.

How about...story time!

Remember the Intel Pentium 4 and its NetBurst architecture? Intel thought that was the future. Using a longer pipeline would mean that each pipeline stage has to do less during a clock cycle, making it easy to increase clock speeds. However, that longer pipeline increased the penalty from missed branch predictions. Initial Pentium 4s were disappointing, failing to match lower clocked AMD processors using a shorter, more traditional pipeline length. In fact, those early Pentium 4s even fell behind Intel's Pentium III processors, which also used shorter pipeline.

But no problem, right? A longer pipeline is absolutely the future, because of the incredible trajectory of advances in branch predictor units (reducing miss penalties) and clock speed. And besides, how will competitors catch up with their old processor architectures when NetBurst gives so much potential for improvement? AMD's architecture, or Intel's old P6 architecture could do more per clock cycle, but who cares when you can just make the clock speed skyrocket?

Intel stuck with the Pentium 4's NetBurst architecture for a while. Pentium 4 processors reached incredible clock speeds (3.8 GHz, pretty amazing even now, achieved with a ridiculous 31 stage pipeline). Advanced branch prediction units meant that Pentium 4 processors were predicting branches more accurately than AMD processors.

In the end, though, even advanced branch prediction units can't predict things that are unpredictable, and penalties from missed branch predictions hurt the Pentium 4's performance. Higher clock speeds allowed Pentium 4s to match AMD's more traditional architecture in many areas, but also brought problems like high power consumption and cooling problems. Intel thought that technological advances would let Pentium 4s beat out competitors using more traditional architectures. That didn't happen.

Today, the NetBurst architecture, once viewed by Intel (and many ardent supporters) as the way of the future, is abandoned. When Intel launched the Core brand, they went back to using an improved variant of the old P6 architecture that dated back to 1995, and achieved superior performance. They never looked back.

Long story short: "it doesn't compete well now, but it's cool shiny technology and it'll do better in the future" isn't such a sound plan.

My prediction is that Nikon and Canon will stick with proven technology that does a better job, even if another option seems to have potential for more improvement. And that decision will serve them very well, because customers don't buy expensive things based on the promise that newer versions will be better.