Facial Recognition - Most effective processing

Sorry, I missed your reply in your other thread.  I will answer your current questions here, and respond to any remaining questions in the other thread.

I'm averaging about 1.5% per day...it's been running 12 days and I'm at 23%. It started rather fast, and as the percentage increased it appears the speed is slowing down.

That seems a little on the slow side.  What OS are you using and what are your computer specs?  Are you using a USB 2.0 or 3.0 port for your external drives? 

As I said in the other thread, performance has improved over the years.  Given past history, a new version may be coming out in two or three months.  Because of your enormous catalog of photos, you may want to upgrade when it does.

Is it more efficient to let Organizer complete the scan, (which at this rate will be somewhere around the beginning of October if it keeps slowing down), or is it OK to add names as it goes.

There should be no problem with you starting to name your stacks and to use the Don't Show Again command to remove stacks of unwanted faces, before the complete analysis is done.  For one thing, it will improve the creation of stacks.  Once you start naming people, the program's AI uses the named faces to create stacks of the identified person.  In fact, as I understand it, the analysis slows down over time because each new photo is compared to all faces that have already been analyzed to see whether they are a match. 

Sure would be nice to see a count of the number of pics that have already been scanned to make sure it's still running.

As josec34178218 states, there is a dynamic status bar in the lower right.  However, this sometimes disappears and does not accurately display the true status.  The only real way to confirm whether activity is continuing is to look at Task Manager (in Windows) and/or look in your catalog's Face Analysis/json folder and see whether the number of .json files is increasing.  You can also look at the dates of the .json files within the folder.  A .json file is created for every photo that has been analyzed.

I was advised that the 2020 version of PSE runs a bit faster. However, what would be the impact of updating before the initial facial scan is complete? Will it start all over again, or pick up where it left off?

If you upgrade versions, your existing catalog will be converted to the new version and all face analysis that has been completed will convert to the new catalog.

Also, is it possible to adjust the sensitivity of the facial recognition while it was still running without screwing up what's already been done? I can't find where to do this but someone posted a screen shot (might have been n earlier version). I just can't find it.

The only way to adjust the sensitivity of the facial recognition is to start naming faces. 

It's identified roughly 1800 faces, although a great number of them are duplicates. Even to go though them once it's done is going to take ages. I'd love to restrict the scanner to prominent faces only.

Unfortunately, that can't be done.  However, you can generally multi-select stacks of faces using shift-click, ctrl-click and even shift-ctrl-click to select contiguous and non-contiguous batches of faces and then use the Don't Show Again command to quickly remove stacks of unwanted faces.  I have deleted several hundred stacks at a time.

I'll respond to your other thread tomorrow.

Look at the bottom right of your screen when in the 'People' menu. You will see how many photos of faces are being analyzes out of the total number of photos. I started the program analyzing people faces earlier today. So far 12% of my 48000 files with faces are being analyzed.