So the lack of comparative data about different DD-23 developers is real.
So far the best source has really been http://unblinkingeye.com/Articles/DD-23/dd-23.html . There's also reference list for literature.
There's also Jim Veenstra's developing times as an example. These are way different than usual DD-23 instructions: 'develop 3 or 4 mins in both solutions', taking care of developing time in both baths just like Gerald mentioned above.
This also suggests that DD-23 should be used as one shot developer. As it is really D-23 + alkaline after-bath. Not a 'silver-bullet' developer that can give similar results even when re-using up to 20 times.
Veenstra suggest that only N- developments needs bath B. N/N+ developments are done with only bath A.
Thornton suggested that contrast should be controlled by varying bath B, not caring about developing times. For normal contrast, use 12g Sodium Metaborate solution. More contrasty or dull light use 7g and 20g Sodium Metaborate solution (He suggested similar method for His DiXactol developer when used as divided developer).
As the most instructions are on the 're-use both solutions and develop 4+4 minutes' (or 3+3, ...), I would like to raise a question.. As the solution A is really developer, how come that no one instructs for longer developing times to compensate re-used and more exhausted solution? Like D-76 suggestion of adding 15% to developing time after each 4 rolls / 8x10 sheets of film if re-using solution.
Is the DD-23 so forgiving developer that no one cares about under development or am I missing something?