I agree; I don't see a clear resolution either.
The obvious argument which I see favoring 3NT as the weaker
("frivolous") slam probe (and this may be Andrew's point; I am not
sure) is that a strong hand probably has several controls include one
cheap one to cue bid, whereas a weak hand is more likely to lack
control in the cheapest suit. The frivolous approach therefore
conserves a lot of bidding space with a hand like AKQxx, AKx, x, xxxx:
1S 2D
2S 3S
3NT frivolous leaves 4C for partner to bid, whereas if you had to bid
4H the bidding would be cramped.
As against that, slam is always more likely when you have max values
(serious slam try values), then when you are marginal. So even though
the serious 3NT method saves on average less space, it may save space
at more important times.
An alternative approach, which I invented many years before the advent
of serious/non-serious 3NT, is in effect to use 3NT just to save
space. Playing an Italian cuebidding method -- bid 1st and 2nd round
controls indifferently below game level, but above game level bid a
2nd round control only if the 1st round control is known to be held --
we bid 3NT when interested in a 2nd round club control from partner.
The logic is that 4C is the only chance the partnership will have to
show a 2nd round control, so you don't want to preempt the partnership
from having the opportunity to bid it. Works quite well.
1S ... 3S then
cue bid 4C = probably 2nd round control exactly; could also have 1st.
cue bid 4D/H = 2 club losers
3NT - probably 1st round club control but not 2nd.
We also played serious 4NT. Of course I realize that no player today
will give up his beloved Blackwood.
Charles