If you do this, why wouldn't you, in the case of Brunswick Anniversary or Gold Crown rails, just tap the wood for the 3/8-16 machine thread of the rail retaining bolt? Why didn't Brunswick just tap the wood for this thread instead of going through the trouble of the round or figure-eight nut plates?
I take it you are being rhetorical in order to set up the explanation.
However, to address it with some other points:
1.) with threads essentially parallel with the wood grain (as opposed to going into end grain hence cutting across all the fibers) increasing diameter makes a rapid very large improvement in pull-out force required. So the 3/8" machine screw fastener into the .720 (or so) dia insert acts like a .720 dia bolt as far as when/whether the wood fails.
2.) inserts are usually used for KD assembly. The threads wear out faster in wood than in a compatible metal, also over the service life of the article (say a pool table) the metal part is sacrificial/can be replaced.
3.) (minor point) it is probably less likely for most people to torque out the threads in a metallic nut than in some wood sections or types. The insert gives (as per 1.) above) can give over twice the withdrawal force of the actual bolt size in wood for the same penetration.
No-tap fasteners are an installation improvement, not necessarily a significant strength enhancement over threaded coarse machine thread inserts of the same diameter. IOW they obviate the tapping step. I sometimes make and use tapped systems because the fastener can more reliably be placed in locations that might split with the no-tap type. Including plywood where the no-tap insert can sort of spread the grain/"swell" the panel near an edge including leading to visible splits from the edge view. OTOH where knife edge version of no-tap fasteners can be used (as probably in pool table rails) they will most likely attain the ultimate possible for a given diameter.
AFA Brunswick and the figure 8 plates, i have always wondered that. Why? inserts could be so much stronger and at the factory level have to be faster to install with machines.
smt
PS, will be curious to learn if your barge cement experiment gives any improvement. Properly used, filled epoxy could. But fundamentally the figure 8 design is just a poor one for a connector that sees significant static and dynamic loads as a matter of normal function. Or another way to put it, in that installation they have proven to be too slight in terms of attachment details.