[I buried this question in another post in another forum where it probably got lost so I'm asking again here where it probably belongs anyway.]

In an SOS article on level setting pointed to by dmbaer, Hugh Robjohns writes :

"Start with the individual monitor or power‑amplifier volume controls at their minimum position, and if you have a variable output on the computer interface or D‑A, set that to maximum."

Then in another article on building a volume control to go between audio interface and monitors, Rob Squire writes :

"Technically, this is a variable pad and with the pot rotated fully clockwise for maximum volume it will still attenuate the signal by around 6dB. This is not a problem, however, and in fact generally speaking we want to throw away quite a lot of level between the output of a typical recording system and the speakers. If you’re operating your system at healthy professional levels as has been discussed a few times now in AT, your average signal level will be around the same mark as the maximum level that amplifiers or self-powered monitors will accept before clipping. By that I mean; the average level will be running the amp/speaker combination flat out and as loud as they’ll ever get. You’ll be wanting to turn it down, and turn it down by quite a bit!"

Not quite the same thing, but along the same lines.  So is it standard practice to have the monitor output from the audio interface always set to maximum and then adjust the actual monitor SPL with its input level control?  Should I do the same ?  I've always avoided maximum settings on anything.
 

I'm probably missing something, but it sounds like he's just trying to minimize noise to signal ratios where possible. You don't want to trickle signal to your speakers and then turn your speaks way up (along with the related noise). You also don't want to send so much you distort anything in the chain.
 
For me, I don't have a pre amp. I just have an amp and speakers... so, I use the fader on the RME Output to control the volume in the room. In effect, the RME is my pre-amp.

The answer may depend on whether the final level control, in the digital to analog interface, is in the digital or analog domain.
 
If the level control is controlling the digital output of the D to A converter than the general recommendation is that you run the digital stream at max to output the full dynamic range of your digital recording.
 
If the level control is controlling the analog output, even if it is a digital control of an analog stage, after the D to A conversion than traditional gain staging protocol should be observed with the goal of maintaining a maximum signal to noise ratio along your signal chain.
 
How do you know what the level control on your D to A device controls? That's not always an easy question to find the answer to.

The Quad Capture master out controls both the headphone and amp send, so the headphones tend to dictate the output level from my soundcard

Another thing to consider is the idea that in the past when analog transfers to other devices were commonplace the output of a device like a digital to analog converter would be calibrated so that the meters actually meant something with reference to analog signal strength. For example; -18dBFS, (or -12dBFS, or -20dBFS) could equate to +4dBu or 1.23vAC RMS. That's when a level control can come in handy.
 
Etc. etc.

Scoot
The Quad Capture master out controls both the headphone and amp send, so the headphones tend to dictate the output level from my soundcard

My interface, which is close at hand, has separate controls for monitor and headphone outs.  I've just been setting the monitor out to somewhere around the middle, for no particular reason, then adjusting my amp/speakers accordingly which I can adjust through a remote.
 
I imagine that for those with active monitors that are a bit far away, it would be awkward to adjust the listening level with the controls on the monitors.  I can understand wanting to minimize noise, but I would have thought that turning the output all the way up might get into some sort of distortion, but maybe not on the input end of things.
 
 

Paul P
 
 I've just been setting the monitor out to somewhere around the middle, for no particular reason, then adjusting my amp/speakers accordingly which I can adjust through a remote.

If you haven't had a particular reason, try thinking in terms of what I mentioned earlier:
 
"traditional gain staging protocol should be observed with the goal of maintaining a maximum signal to noise ratio along your signal chain."
 
The better the SNR of your "send" the better the SNR of your final output will be.
 
The fact that you have a remote control should make enjoying best case SNR, and SPL that you find comfortable, a breeze.

My system is old school with a old Yamaha P 2100 passive cooled power amp and the NSM10's. 
At max unattenuated level there will be a little bit of hiss from the power amp so I have always set it at 75%. The output control of all sorts of interfaces I've owned seem to work from 10 o clock to around 3 o clock before things are pretty darn loud. So 12 o clock is normal for most mixing. 
I would probably turn any powered speakers to the same 75% setting out of habit more than logic. It seems the goal is to have that working volume at 12 o clock. This gives me a quick set up point as I'm in the habit of turning the output volume off before shut down. 
Having the front panel level controls for headphones and main monitors is a must have. I'm missing the balance control my Tascam has ( input/computer) as the Focusrite uses mix control software and it's a bit inconvenient to not have a hardware option. Mix Control is powerful but you need 2 monitors to make use of it while working with clients who are fussy about the headphone mix.