I have just had time for a quick look at the first game in TN´s pgn file while grabbing a coffee in my lunch hour:
I´ll paste the game below, heres my initial thoughts: Watson makes a big thing of ...Nh6 not being given more treatment, he´s right of course but wrong to lambast the book. As for this game:
9.c3 as played by Perelshteyn is better in this move-order with ...Nh6
13... Nh4 you say is given equal by Watson - an example of Watsons rose-tinted glasses. 14.0-0-0! and if 14...Nxf3 15.gxf3 the absence of many pieces around Blacks king will make the g-file extremely dangerous once occupied by White rooks. Maybe Black should try 15...Qh4 intending ...Qf4 to swap Queens, but even in endgames White has a dangerous initiative, he has play on both sides of the boards (Q-side via Kb1, Rc1 etc) and Black has the usual problem with the bishop - check this with the engines and they also love White.
15.g4? is the mistake in my view as you rightly point out.
15. Bxf5! is strong, lets look at the improvement you give:
15... Qxb5 16. Bc2 Nb4
Here White has 17.Ng5! with a massive advantage (close to winning). He has no objection to his bishop being exchanged for the Knight since it leaves him the classic good Knight v Bishop. Also now ...g6 looks practically forced when off-hand something like Rf6! followed by invasion on the dark squares. Its hopeless for Black.
So lets see if we can now reverse Watsons words - Whites Bd2,f4 strategy is already beginning to look quite good .....
Quote:[White "Perelshteyn, Eugene"]
[Black "Shaked, Tal"]
[Result "0-1"]
[ECO "C17"]
[WhiteElo "2360"]
[BlackElo "2445"]
[Annotator "TN, Watson"]
[PlyCount "50"]
[EventDate "1997.??.??"]
[EventRounds "9"]
[EventCountry "USA"]
[EventCategory "3"]
[Source "ChessBase"]
[SourceDate "2000.11.22"]
1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. Bd2 Nh6 6. Nb5 Bxd2+ 7. Qxd2 O-O 8. f4
Nc6 (8... a6 {- next game}) 9. c3 {This will transpose to 9.Nf3, see the game.}
(9. Nf3 {Watson: 'and at this point the authors say merely 'Compare 5...Ne7'
and stop! What a strange way to avoid the issue: it takes almost no thought to
see how different the two lines are. A knight on h6 defends against the
tactics involving Bxh7+ as well as the move Nf7 that DPA feature in the lines
after 5...Ne7. Furthermore, Black's knight can help attack the centre from f7
after he plays ...f6, and ...Nh6 also leaves Black's queen in contact with the
kingside. Finally, the knight can (and does) go to g4 with great effect when a
pawn on d4 anchors e3 for occupation via ...Ne3 and/or when Black's queen on
b6 supports that move. A knight on e7, on the other hand, has its own
advantages in that it can go to c6 (often with tempo), or to g6, or c8 (from
which square it can challenge a knight on d6). Did the authors simply not
understand this?'} f6 $5 (9... a6 $1 {is also good for Black - see Watson's
analysis. This had not been played at the time of the publication of the book,
but should still have been covered as the position after 10.Nd6 occurs by
transposition in another game.}) 10. c3) 9... f6 10. Nf3 fxe5 11. fxe5 cxd4 12.
cxd4 Nf5 13. Bd3 Qb6 $5 (13... Nh4 {is equal according to Watson - I agree
with this assessment.}) 14. O-O Bd7 15. g4 $2 (15. Bxf5 Rxf5 ({I suggest the
improvement} 15... Qxb5 $5 16. Bc2 Nb4 17. Bd1 Rac8 $132 {with good play for
Black.}) 16. Nd6 Rff8 17. Kh1 h6 18. a4 a5 {and White has only a tiny edge at
best.}) 15... Nfxd4 16. Nfxd4 Nxe5 17. Kg2 Nxd3 18. Qxd3 e5 $19 19. Qb3 exd4
20. Qxd5+ Be6 21. Qe5 Bc4 22. Rxf8+ Rxf8 23. a4 d3 24. Nd6 Qf2+ 25. Kh1 Qf3+ {
Watson: 'Okay, that's embarrassing, and surely given this personal experience
Perelshteyn shouldn't have hidden behind the 'see 5...Ne7' comment. Just as
importantly, if one is to advocate this line for White and even superficially
analyse it, one would have the obligation to justify playing into the most
difficult positions which directly arise from it, especially right away on the
ninth move!'} 0-1