"Do I really need 1/8000 second shutter speeds, or is a camera with only 1/4000 enough?”
Excellent question. The answer, as to a lot of questions, is “maybe.”
If you’re shooting at base ISO all the time, it’s somewhat unlikely that you’d be bumping up against a 1/4000 shutter limit. For ISO 100 Sunny 16 is 1/100 and f/16, so you’ve got at least five stops of leeway (e.g. you're at 1/3200 at f/2.8). That’s a good enough range for most types of photography. If your camera is base ISO 200, then Sunny 16 is 1/200 and f/16 and you’re going to lose one of those stops of flexibility. You could be over the top at f/2.8.
Where you certainly might hit the shutter speed bar is when you really want to use a fast (f/1.4) lens in bright light to produce shallow DOF (perhaps some portraiture and street scenes). You usually can’t lower the ISO and thus you hit maximum shutter speed before you get to that f/1.4 aperture. (Many cameras allow a false ISO of half the base, but they’re cheating on highlights, and you need to be careful using that.)
Those of us who shoot sports and wildlife action tend to hit the shutter speed bar a different way. I’m at the darker shadow end of the stadium/arena maxed out on ISO and trying to maintain a 1/1250 or higher shutter speed (depends upon sport). I pan with the subject to a brighter area still lit by sun and I may find myself suddenly hitting the top shutter speed.
Auto ISO certainly can help in the second situation, particularly when you want maximum aperture, as I tend to want in the sports situation.
While I’m discussing this, the 1/2000 top shutter speed imposed by some cameras in some electronic shutter modes on modern cameras can be very frustrating to deal with. So I’d tend to say: Avoid 1/2000 top ends; double check whether a 1/4000 top end will work for everything you do; or just get a 1/8000 top end.
“I’m having trouble seeing the effect of a circular polarizer in my EVF. What am I doing wrong?”
You’re letting Nikon pick what to do about contrast, saturation, and tonal curves. In other words, you’ve selected the Automatic Picture Control, which is the default these days. Coupled with the EVF showing you what the image will look like, the camera is fighting your polarizer, trying to produce the same look. Try Neutral as a Picture Control and you won’t be having that fight.
"Are you sure XQD cards never fail?"
No I'm not, and I never wrote that. What I wrote originally several years ago was that I'd not then heard of an XQD failure. All of us have had a lot more experience with XQD since then. What I can say is that an outright physical failure—which is relatively common with SD cards used for a long time—is extremely rare. I know of two such circumstances, and it appears that in one the culprit wasn't the card itself, but the card holder in the camera. In the other case, I personally have a card where an edge of the plastic case has finally succumbed and started to break off. Super Glue to the rescue.
Logical failures also seem to be rare in the XQD world. I know of four that have been reported to me, most from high volume shooters. Like all NAND-based storage, XQD cards have both a fixed life and like any semiconductor, can fail. Buying from an established, vetted manufacturer is better than trying to find a bargain somewhere. That's been true of SD cards, and as XQD/CFexpress gets more popular, it will be true of them, as well.
What people don't talk about, and what's difficult to get a clear handle on, is that when you produce cards, you source both memory controllers and NAND parts. You can buy cheap, lightly tested units, or you can buy more expensive, highly tested units to put in your product. When you buy a US$20 SD card, which do you think the manufacturer has used? When you buy from an XQD card from a primary vendor who values reputation, such as Sony, which do you think the manufacturer has used? Some brands, such as Delkin and ProGrade tend to make claims backed up with some form of warranty, and they probably do so because they're buying and integrating the "better" parts rather than the "cheapest" parts.
Unfortunately, there's no easy way to verify any of this. I've had long discussions with card vendors, and they generally won't tell you exactly what parts they're using and how they were tested. No one wants to get into specifics, as if they're hiding some trade secret from each other.
I can speak to my own use, for sure. I shoot tens of thousands of images a year. Virtually all of that now is SD or XQD. I encounter fairly regular SD card failures, even from top brands, and physical failure is definitely an issue I have to watch for. I've literally had an SD card fall out of the camera in parts. XQD? I'm not had a physical failure, and I've not had a logical failure to date, either. That's better than my experience with CompactFlash with about the same volume of use (and of course, CompactFlash card/slots are highly subject to pin bending, which is a form of physical failure and fairly common).
All else equal, I'd take an XQD/CFexpress over an SD card today. But all else is not equal; CFexpress cards, slots, and readers are clearly faster than the top SD ones we're currently using.
So I'll continue to support XQD and its successor CFexpress. To me, these cards seem like the best of both worlds: robust physical attributes and high performance. Will I ever experience an XQD/CFexpress failure? I'm sure I will. Use something enough and it will eventually fail. What I don't like is using something a little and having it fail.
"Where are the CFexpress card firmware updates?"
This is a bit of a chicken and egg scenario, and I think some eggs got broken in the process. For CFexpress to really hit the market, it has to satisfy a lot of needs now, not just those of Nikon cameras. Canon and Panasonic will be supporting CFexpress. Plus we've got video devices and industrial devices using these cards, too.
Nikon has finally issued a firmware update for the Z6 and Z7 to use CFexpress cards, about a month later than anyone thought. But note that Nikon says that this firmware has only been verified to work with Sony CFexpress cards. Reports of problems with SanDisk CFexpress cards abound.
What I heard is that in the final testing of early production cards someone (everyone?) found small issues that needed addressing. In other words, the first chickens didn't produce the expected eggs. Or was it the eggs didn't produce the expected chickens? I'm told that these are normal birthing issues and that everyone went into deeper testing and resolving cycles than they expected to have to.
You also need finished, or at least near finished, devices upon which to test. The upcoming D6 and 1DX m3 are two of the likely best test beds, as they both have dual CFexpress slots and can stress the write/read performance. The Nikon D5, D500, and D850 all have early XQD slots that need firmware upgrades, and I have to wonder if their slots are slightly different than the Z6/Z7 ones. Whatever the reason, it'll be a little longer before we get firmware updates for those cameras, I think.
CFexpress cards now have to pass Canon, Nikon, Panasonic, and Sony device verification at this point. The rumble out of Tokyo a month or so ago was that there was an issue in the software stack that kept a complete sign off of all players on all devices from happening, and that issue was being resolved by the standards group (which extends to the card makers, card reader makers, etc.).
Some things take time. Patience, grasshopper. We'd rather have this done right than quickly.
