Yeah, Bill's tangential alright, but usually in a good way. Explaining to him how he's wrong makes me think better, so it's a bonus
Forgetting surfactants for the moment...I've seen some amazing floatation chemicals that are so dense normal rocks float in them, and this is a fast process. If it floats, you wash it away (catch in a filter and recycle the fancy heavy liquid/slurry). Now you are a lot more concentrated....think U enrichment which takes many steps with a little improvement at each one. The farther apart the densities, the faster you can run a process and still get decent separation. Not to mention, the evenness of the "fineness". For example, at very fine grits, it can take quite a long time (hours) for a heavy solid to settle even in still water -- this is how you grade some abrasives. So very fine stuff will tend to float, and if it's gold, it will be lost (think how it acts in GoldSchlager) if you go too fast. Once things get to "sand size" it's not so much a problem unless you're going really fast and with a high Reynold's number (turbulence).
A lot of the surfactant stuff is mainly used to separate fine from coarse -- the fines get caught up in the bubbles regardless of density. You might not want that?
So you might want to start by separating by size, or crush to uniform size.
At any rate, to combine fast with effective, think about a multi-stage setup where it just gets better each stage -- by the last one, there's not much left (unless you are insanely lucky and it's all gold), so it will naturally be able to go slower if it's the same size.
At some point....the old mercury method comes to mind for these small amounts (again assuming one isn't insanely lucky). A solvent beats a surfactant any day.
B&W has designed and has data on some related stuff that separates coal and rocks -- important to coal-crusher life etc, but the same problem in another guise. Miners in general do some version of this, only super rare stuff gets hand-cobbed (or in places where labor is real cheap). I'd be looking for tricks for ore enrichment from anywhere in the mining business -- the same tricks will more or less apply here, modified to suit the densities involved.
I easily pulled all the black sand out of a gold dust sample for another guy the other day with a magnet. The trick there is to go slow too, and shake the magnet to get all the gold dust loose off the black (iron oxide, magnetite) sand before pulling out of the top of the vessel. Near perfect separation in a few seconds, after he'd worked it with a pan till he was blue in the face and couldn't get that last bit out of the gold (about half the volume). So you might think about a combination of techniques here....
After all, if you can get to a place where you remove 5% of the junk, but zero gold per pass -- you can get there fairly quick.
Now if you want truly tangential -- how about eddy-current braking that would only work on electrically conductive stuff. Get out that superconducting magnet! Of course, you'd have to separate the iron compounds in another stage.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.