The side cores did a standard one engine landing burn. The center core, because it was coming in so hot or SpaceX was pushing the envelope, was supposed to do a three engine hoverslam. SpaceX would know from data what went wrong and they say there was a shortage of TEA-TEB. They have no reason to lie about that. So the question is really what caused the shortage? I have full confidence they will know and correct whatever it was. I doubt that there just wasn't enough TEA-TEB loaded. Lots of interesting takes on it here.
What was different?
Probably the hottest reentry yet. You could see the camera get clogged with crud a lot earlier than ever before. That tells me it was very hot.
The center core had more bits on it than a standard F9 does. Maybe dealing with different aerodynamics had some factor in the shortage.
Three cores instead of one and doing different things. Maybe a software error prevented or purged excess TEA-TEB before the final landing burn.
Would a center core use a 3-1-3 burn on final? This shot shows all 3 were running (starting at 59 seconds before impact), but then it was down to one engine a few seconds later.
I went back and watched it. There was a single engine start, that expanded to three engines for maybe a second, and then it went back to a single engine for the landing.
I really like the corrected final FH launch video. You can see what they really wanted and where they went wrong originally. You can catch some of the original video in other people's livestream, now just a video.
Note that this is the first landing failure since June 15, 2016. The Eutelsat GTO hot landing where they just ran out of fuel just above the ASDS.