A game from the first round - Smeets-Reinderman - caught my eye, from a variation of the Open Ruy invented by Igor Zaitsev and famously played in two world championship matches - its debut in Karpov-Korchnoi, 1978 and its absolute high point in Kasparov-Anand, 1995. I knew it was theory pretty far in, so I wondered if this 28-move draw (by perpetual check) simply repeated pre-existing theory. The novelty came on move 24, but what was surprising is that Black improved, or "improved", on an earlier game Black won in 32 moves! Here's the position before Reinderman's novelty:

It's Black to move, and he played 24...c2, threatening both 25...Bxa1 and 25...d1Q. White's situation would be dire, if he didn't have the neat tactical resource 25.Rxa6!, with the point that 25...d1Q, as played by Reinderman, allows perpetual check starting with 26.Ba7+ Kc8 27.Bd4!. White threatens 28.Ra8#, and moves like 27...Qxf1+ change nothing after 28.Kxf1. Black will have to play ...Kb8 sooner or later, and that's what happened: 27...Kb8 28.Ba7+ Draw.
This looks neat enough, but what about the stem game, won by Black? That was Burnett-Becerra, from a 2006 USCL clash, and there (going back to the diagram position) Black played 24...Rhe8. Now there are two questions: (1) Why can't White go for the same perpetual? and (2) How did the game finish? The answer to the first question makes for a nice mini-quiz, and I recommend you try to solve it before proceeding.