Perhaps, but in order to learn from failure, you also have to recognize failure, and avoid it.
Most of the DISCO and NUCLEO boards are very good, the adoption of the Arduino shield has helped focus design goals. Both the DISCO and EVAL boards tend to be very pin constrained, and this is compounded by the inflexibility of the pin muxing options. Now I understand why you don't want to have to route high speed signals all over the die and pad-ring, but when you're left with NO OPTIONS, it does suggest that those in the design meeting(s) didn't advocate sufficiently for usability. And this gets to be particularly telling in low-pin count offerings. And for things like the F429-DISCO where there's very limited option for any expansion or usage beyond "here's a cheap board with a good screen and lots of memory", but has poor connectivity options.
The STM32F429I-DISCO is however the most broken design ST has delivered, at least in its ability to expose usable pins. I think introspection on this, and warning people away and to better choices, is the most prudent path. I also find that straight-forward communication saves everybody time. And while at times this can be indelicate, it does need to be done in order for things to change.
I will also note that many of the EVAL/DISCO boards use parts which are denser than most people are comfortable with manufacturing with, on boards of complexity/cost that most cannot accommodate.
