Yes I agree, your point is well taken, the recovery needs to be decreased- as the OP said - in future sessions. The difficulty lies in the fact that this is NOT a predictive session. Time-trials, simulations, etc. are predictive sessions, whereas this is a "training to train" LA/O2 energy system session. Martin/Coe (20 to 30 reps) indicate that this may be defined as a 1500m pace session.
The difference between "training to race" and "training to train" sessions are about proximity to racing preparation. For training, the Bowerman/Dellinger (amongst others) idea of Date pace to Goal Pace to determine intensity of training is the intelligent approach, where training intensity follows racing performance (or at least a history of it) NOT the other way round. For developed runners, doing specific sessions at Goal pace (as B.kennedy often stressed) is the way to go - but one assumes sufficient knowledge from coach/athlete to be able to not let sessions get out of control. In other words I could go out and "race" a session of 200's at 32sec (which is what you are saying), but that won't necessarily mean that I can run a 4 minute 1500m unless there is a relationship to something I have done before or a trial at a similar pace.
If you want to "test" for a race - do a time-trial (e.g. 1000-1200m, for 1500m/Mile) or a simulation (500+300 with short 30sec recovery for an 800m). Save the training sessions (20x200m) for what they are supposed to be about - advancing your training (for speed endurance/anaerobic capacity or whatever).