FD is a good idea, rather poorly implemented. IMO what's really needed is a sort of programming language for bridge auctions, that would generate actual auctions. Syntax could be similar to the shorthand often used to describe general agreements, something like:
1m-1M-1N-2om = 5M, 10+ HCP
1m-1X(!1N)-2N = 18-19 HCP
(1w)-Dbl-(1y)-Dbl = 4+ x, 4+ z, 8+ HCP
1M[1,2]-3M = 4M, 10-11 HCP
1M[3,4]-3M = 4+M, <6 HCP
1M[3,4]-2c = 3M, 10+ HCP
1M[3,4]-2d = 4+M, 10+ HCP
Those 7 rules define basic versions of NMF, opener's jump to 2N, responsive doubles, and a basic major suit raise complex. To implement those in FD would require literally hundreds of sequences. At one point when I was fooling around with a relay precision I actually wrote a program to bulk generate an FD card for some of our relay auctions. Even with just a few steps in the relay, the FD card was several MB in size, with thousands of auctions, despite being describle in simple english in a few paragraphs.