I use the Kennicott-Morphy game in II, but Morphy's game against Judge Meek could have been used as well for our purposes.
Your points are well taken: ...d6 is good, ...d5 is good, and so is ...Qh4, once Black gets to take a breath.
My point was to get to the Vitzthum Attack (Part III), in case Black plays best moves. I believe a lot of players could learn something about strategy and tactics from study of this variation, even if they never play it from the white side. Otherwise I would not have spent a few hours putting it together.
In my heart I want Black to blast White off the board after 5.Ng5, but it is not possible, if White doesn't go off the rails. I have lots of games in my databases of ELO <2200 winning with regularity with White in serious games of recent vintage.
I would rather have the black side in a blitz game, but I might get disappointed with a draw in the end.
I simply can't prove the Sarratt Variation loses for White by force, despite my researches. I liken it to Fischer's article "A Bust of the King's Gambit" published decades ago, after he lost to Spassky from the black side. All the 10000 people around the world, who won today with the white side of the KG, are sure to be amused that they were playing a "busted" opening.
Bullet and Blitz players are always looking for some off-the-wall opening variation to confound their opponents. If the variation is not quite sound and their opponents prove up to the task of finding their way through the maze, well, they might win on time anyway, because of their opponents' difficulties in finding the way.
Such an opening is the Scotch Gambit Sarrat Variation, which has more soundness in the end than a lot of other craziness.
However, to show all the attacking and defensive ideas in this somewhat complicated variation would require a really long post. So, I tackle that problem by breaking it up into the historical development of the variation, which makes it easier to remember for most people, and divide it among 3 posts. [I use the Roman numerals I, II, and III to differentiate among the posts, just like "Star Wars"!]
The Scotch Gambit (1.e4 e5 2.Nf3 Nc6 3.d4 ed 4.Bc4) is usually answered by 4...Bc5, although other moves are tried, such as 4...Qf6. With 4...Bc5, though, Black supports the gambit pawn by developing a piece to a good square in openings of this sort.
In the position under consideration, it appears White will have to expend some effort to regain a pawn or not bother at all and try to get compensation in some other way. One time-tested method for White is 5.c3. Then, after 5...dc 6.Nxc3, White has developed with tempo and has half-open c and d files for his heavy pieces, plus his pawn on e5 is uncontested. All-in-all White has adequate compensation for a pawn at the moment in this case.
However, sometime around 1818 Jacob Henry Sarrat turned his attention to the Scotch Gambit and wondered if there wasn't something better than 5.c3. After all Black is not forced to play 5...dc and might play 5...d3. Whether B or Q takes on d3, neither piece is particularly well-placed there and will usually need to redeploy, wasting time. Plus, it's not like the Pc3 is helping out, and would probably be better back on c2 to give the Nb1 a good square.
Well, Sarrat came up with a totally radical idea: 5.Ng5, which doubly attacks the weak spot f7 and opens the diagonal d1-h5 for the Q. The latter is not an incidental factor in forming an evaluation of the position. By geometrical motif, Qd1-h5 can attack the unsupported Bc5, after all the trades are made at f7 and the king captures last. Since Qd1-h5 would be with check, the Bc5 must fall. Therefore, White has really serious threats.
After 5.Ng5 many players of the black pieces have lost their heads and thrashed out with moves like 5...Bb4+ or 5...d4. These hot heads didn't always lose after these sorts of moves -- sometimes they even won -- but it wasn't because they had a better or even equal position after their rash plays.
Cooler heads playing the black pieces would take a moment to evaluate. First, Black has made no bad moves -- or even a pair of 2nd-best moves -- in the opening; so, why should Black panic? Second, opening principles tell us that moving pieces twice in the opening to start an attack is premature and should result in failure; so, that should fortify Black's spirits as well.
The reality on the board, though, is that White threatens to crash through on f7, winning a R, Q, or even K. As if that weren't enough... if Black survives the first wave, there is always Qd1-h5 and the Bc5 could be lost.
With these thoughts in mind, what defense would you consider best? [See Part II for some historical developments in this line and a couple of interesting complete games for some answers.]