What makes them instructive is that they are generally full of tactics. Peter Clarke wrote miniatures are entertainment; they are full of thrills, comedy and farce. Technically the quality of play isn't a criteria, but for the game to have any instructional value it should be free of gross blunders. Clarke noted that when the loser disregards basic principles or makes gross blunders, miniatures have no appeal. Miniatures are the result of mistakes, of course, but they do not arise out of a lack of ability. They are usually the result, he said, of over ambition where the loser tries for too much in the position.
The following game does not fit Clarke's definition precisely as the two players, Reshevsky vs. an ordinary master, can't be considered a game in which the players were evenly matched. USCF Master Gallagher didn't lose because of over ambition or trying for too much. He made an elementary mistake, but that's what makes this game instructive...the master made a typical mistake that we all make.
C.J.S. Purdy constantly drove home the point in his writings that at EVERY move you have to look around for a SOUND tactic and he gave some helpful advice on how to spot them. See my post Tactics, the Pornography of Chess.
Always try to sniff out a tactic if you see: pins or forks, pieces with limited mobility, a piece that could be attacked if a piece that stands between it and a piece of lesser value could be removed (a masked piece), or a piece that is undefended. Always take a gander at Pawn breaks, sacrifices and bizarre moves; something may turn up there. You should also look at what Purdy called "jump checks" which are checks that could be played if a piece could jump over any intervening pieces.
In this game Gallagher's mistake was pretty obvious; his 14th move placed a N on a square where it was undefended and Reshevsky immediately took advantage of it. Prior to Gallagher's tactical error though Reshevsky himself made a mistake when he took with the wrong piece at move 14. Reshevsky, who was a fine tactician and excellent at calculating variations, may have seen that 14.Nxd6 was better and believed that he could win the game against a weaker player even if he did find the best defense with 14...Bxb5. But, that's conjecture. After all, Reshevsky was was 79 years old and his rating had slipped to 2459 in 1990 and this was the last US Open (won by Yasser Seirawan) he played in, so he may have just miscalculated. Heck, the year before even I managed to draw a game against him, so he wasn't the Reshevsky of old!