IMO, it is a big help once you learn to work the tree. The potential (no guarantees) improvement comes from two areas:
1) Reacting to the top bulb (first flash), can be more consistant
2) Adjusting the total RT with a computer is easier and more consistant than adjusting the driver or the car with mechanical changes.

So when you add those together, plus a good driver, the average RT of a delay box class is much better than without it. Roughly 0.010 vs 0.030 comparison, plus or minus. Your mileage may vary...

Ask the track operator to average all the RTs from last years delay box class and the same for the non-delay box class. Then post the results. Even an average from one big race (7-9 rounds) should be enough to make the point.