Domdaniel: Unlike so many other things - which we can call opinions - novelties are a matter of fact. A move in the opening has either been played before, or it hasn't.
Of course, mistakes can be made. A move may have been played somewhere by an unknown player and not found its way into the databases. Back in the 1960s, before personal computers, GMs used to guard their access to tournament bulletins and other sources of data. But this game dates from 1990, well into the computer age, and it should be possible to verify any novelties.
I see that my Chessbase book adds a '?' to Black 6th move (6...Nc6?) and White's 10th (10.Bb3?), though both have been played hundreds of times. Such annotations, however, are not facts: just the opinion of the book's compiler.
Note that the Dragon is probably the most TN-sensitive of all openings: whole lines have gone in and out of fashion when new moves were found for one side or another. Anyone using 'old' theory against a well-informed opponent is liable to find themself in a lost position.
Just using CG's opening explorer as a rough guide, one can see that up to 11...Ne5 is a very well-trodden path, with over 500 games (eg Lombardy-Byrne, 1958). But with 12.Bh6 (?) the numbers drop sharply - the main line is (and was) 12.h4, with 12.Kb1 in a distant 2nd place. Chessbase book gives 12.Bh6 a '?'. White, with this over-eager move, drifts away from any kind of Dragon mainline.
In a sense, this makes the follow-up less significant: 12.Bh6 is busted, and had been busted before 1990. But perhaps Black found a new or more forcible way to demonstrate the bust? Let's see.
12.Bh6 Bxh6 13.Qxh6 Rxc3
The thematic Exchange sac, played by Velimirovic and Korchnoi in the 1960-70s, is standard.
The crux. Chessbase suggests only 14...a5 and 14...Qa5 for Black. In the CG database, this is one of two games with 14...b5. The other one, Slavotinek-Travers, was played in the Australian Women's Championship in 1981 (a game won by White, incidentally).
As this was nine years earlier, 14...b5 was not a novelty here. The TN laurels go to Ms Catherine Travers. The Black player here, however, may have been at least the first *male* player to uncork the move. Or first to win with it? Let's dig deeper.
Chesslab.com has only one example of 14...b5, in its 'historic' (pre-1991) section: it's Davidson-Goldsby.
365chess has 48 games reaching the key position, but none at all with 14...b5 --- only 14...a5, 14...Qa5, and 14...Qc7. Chessbase also turns up a blank.
And that's as far as I go. There are other databases I could check, but I think we've learned enough (and the relevant info was here at CG all along). I'm no Dragon expert - I've tried studying it, but was defeated by the amount of work involved - but it's safe to state some conclusions:
14...b5 was not a theoretical novelty, having been played earlier in Australia. But it seems unlikely that the players here were aware of that precedent.
More significantly, 14...b5 didn't subsequently catch on. On the rare occasions when this line (with 12.Bh6?) was later played, the Black players used other moves to follow up the Exchange sac (14...a5, 14...Qa5, etc).
It's hard for a non-expert to state definitively which 14th move is best for Black. And, in any case, that would just be an opinion: we're trying to restrict ourselves to facts.
The first novelty in this game was whenever it stopped following the Aussie women: 15...Qa5 rather than 15...Nc4. But of course 15...Qa5 transposes into games going 14...Qa5 15.h4 b5!? ... which would require a whole nother search. I'm passing on the torch.