I thought it would be interesting to let Houdini 1.5 x64, Fire 2.2 xTreme x64, Naum 4.2 and Critter 1.6a 64 do a comparative analysis on the game just to see how their evaluations stacked up against each other and against the analysis of the day.
It seems to me that Gligoric’s evaluations were more accurate than O’Kelly’s. It appears that the analyst of the day missed a promising line for Black at move 19 when he could have played 19…Bh4+. My database shows the only move for White as 19.c4 which was played 52 times. Black played 19…Bh4+ in 29 of those games with a performance rating of 2722 which is around 200 points higher than was achieved with 19…Ra7 (18 games), 19…c5 (2 games) and 19…Be7 (2 games). However, looking at the actual results of games where both players were rated over 2400 the results for 19…Bh4+ were +2 -1 =3.
Once again, engine evaluations are not always 100 percent reliable, especially at fast time controls. The position is very complicated and trying to make an accurate determination is way beyond my ability.