[tpm] New Parrot release!! .... one year since last update :-)

[arocker at Vex.Net]

Trying to deliver a first iteration of a product with all the features
suggested for the final version is too ambitious. No matter what the
field, ATTAM ("All Things To All Men") products tend to be mediocre at
best. Compromises for unimportant aspects degrade essential performance
areas.

Producing an MVP which can evolve seems to work more often, though in that
case it's ssential to keep the feedback loops short, so that "need queues"
don't become an obstacle to progress. I suspect that developing several
levels of a stack from scratch simultaneously is an unusually difficult
task. Once it's developed, provided the APIs are well designed, evolving
from a working state to a better one is probably much easier.

Herbert Simon's "Sciences of the Artificial" is an interesting read about
these problems.