Every successful large system is a redesign of a somewhat smaller working system. I know of no exceptions to this rule
Page 709 on the use of models. In the broader context of 'The development process'
Maybe the most important criterion for a prototype is that it has to be so incomplete that it is obviously an experimental vehicle
and cannot be turned into a product without a major redesign and reimplementation.
711 on experimentation and prototyping
Most organizations reward individuals and groups that choose to re-invent the wheel
Page 714 on reusability
If productivity is measured in terms of number of lines of code produced, a sucessful application of reuse will appear to cause negative performance of programmers
716-717, in a discussion on individuals and reusability
Some problems are best solved writing a set of procedures
Page 727, in the context of ignoring classes and 'design and programming language'
Forcing everything into a single hierarchy can introduce artificial similarities and obscure real ones
page 732, using class hierarchies exclusively. Class Object...
Naturally, a derived class depends on its base classes
It is less often appreciated that [...] if a class has a virtual function, the class depends on derived classes [as well].
Page 737 in the context of dependencies within a class hierarchy.
A one-to-one mapping between the design concepts and the programming language concepts should be used whereever possible
Page 747. As an example: delegatation.
A major concern of a class design is to get an object [of that class] into a well-defined state
The simple act of defining invariants and using them during debugging is an invaluable help in getting the code right