The 18MP thing could be a huge misdirection. Being very specific, great way to isolate a leak in your organization.
But it is plausible.
Canon IS a practical kind of company in what it offers. It wants to offer what is best for REAL photography, not necessarily moronic tests on DXO, DP, or to please the tech junkies who want specs. This is why Canon likes the AA filter. It improves IQ. To the layman or amateur, current day use of an AA filter in a DSLR seems like an archaic practice.
The same is true of megapixels. The 1DX absolutely and undeniably proves that its IQ is superior to the higher MP 5D3 as the ISO goes up. Megapixels certainly isn't everything. Less noise more than makes up for less resolution. High resolution is of no help to anyone and cannot be realized or used when it is utterly destroyed by noise. This is a fact. There is no debate. The thousands of images made in comparison is the evidence to all who are not blind.
Also, real pros doing real work know that 18MP is perfectly fine. While the 5D3 is the wedding photography king, a lot of event shooters use the 1DX also for its low light capability. They aren't bothered in the slightest about the MP count because they know the images these cameras produce is more than enough for any wedding album or print.
Really, if you need more MP for a larger print - they now offer the 5DS for that. And once you are in the realm of creating large prints with such high detail - you're either doing studio or landscape work, and high ISO isn't in the game any more. This was a good segregation of specialties by Canon.
All that said, for the NEEDS of real pros (not the wants of internet reviewers), 18MP would not be viewed at all as a negative selling point, not if the ISO was improved at least 1 stop.
Thus, this is a plausible rumor because it appears that Canon is basically giving the 1DX sensor to the 5D4. Not a bad move at all. Nothing wrong with it. Now, if it is an improved 1DX sensor at 18MP (using newer process), even better. But we'll see.
Now, from a marketing point of view - this would be disastrous. All the tech junkies, nerds and people who argue and debate incessantly about the most minute aspects of technical specs and THEORETICAL IQ will be in uproar over this. To them, going down in MP signifies a step backwards. Even though direct evidence to the contrary is in their face as to why this is not.
But this all depends on who Canon is marketing the camera to. If the 5D line is going back to its roots as being for pros - no problem. If there is still a large enough market of prosumers and techies with lots of disposable cash, it could be a problem. They'll miss out sales to these people.
There's still rumor of a two-flavor 5D4, one with the 18MP and the other with 28MP to satisfy everyone. Personally, I don't see splitting that model out happening like it did with the 5DS.
If the 6D2 is decent enough, an 18MP 5D4 is possible.
The 6D2 though will be sufficiently crippled to protect any 5D4. Namely, weaker AF system and single-card slot. In order to have those extra 10 megapixels, you'll give up probably one full stop of ISO, a high end AF system and the data safety of a 2nd card slot. (and likely 4K video). All of which, in my opinion, is a major deal breaker for anyone thinking about saving money by doing professional work with the 6D2. Precisely what Canon wants.
So a 6D2 has got to be the prosumer all purpose camera to appease the masses, while the 5D4 becomes a serious tool for the event photographer.
Another thing to note is, 18MP would be a bit of a gamble for Canon. By going "backwards" in MP, and taking a major hit in marketing - Canon would really, truly need to come through on ISO performance. It would need to be a TRUE 1 stop improvement across the board at minimum. But I would say it would have to be more. 1.5 stops would be sufficient to balance out the lower MP count issue.
If it is anything short of 1 stop, this will be a flop. It will need to be easily noticeable in IQ side by side comparisons with the 5D3. None of this drawn out side by side comparing to try and find the differences. On a basic zoom in of the image, it will need to be immediately apparent. A short glance to say "wow, big difference" ...not tedious pixel peeping of very similarly noisy images.
1.5 stops isn't unreasonable. the 1DX is basically a stop better than the 5D3, so right there that sensor has it. Add in newer sensor technology and it shouldn't be out of reach I would think.
Overall though, I'm not too certain about this rumor. If I had to bet, I would say the 5D4 is going UP in megapixels. There's just something wrong about going down, and Canon knows it. Marketing will prevail. After all, Canon did not have to put out 50MP in the 5DS, they could have done 36 - 46 MP instead. 50 was a marketing driven goal.
I expect the 5D4 to be anywhere from 24 - 28MP, and improve ISO by 3/4 stop, but no more than 1. It will get close to the 1DX in ISO, but with more MP. I expect a 20 or 22MP sensor with one more stop of ISO on the 1DX2. Which will be amazing.
If there's going to be any real big leaps (like full stop improvements), it's going to be on the flagship. Not the 5D line.
Figure it this way. With 18MP in a 5D4, Canon pleases pros and those doing real practical photography. But kills themselves in the media and marketing world. On the other hand, if they go 24-28MP, now they don't have to come up nearly as big in ISO because they don't have to offset a "downgrade" in MP. They'll still improve ISO a bit, just not as much. This overall is still an improvement. They will still please the pros and event shooters, all without killing themselves in media and marketing.
Which is most likely? The latter. The latter is because it is positive in all areas without causing damage. The former is problematic and thus unrealistic.
18MP in a 5D4 is ONLY happening as a two-model offering, if that even happens.