Most of this is copied directly from the bouquet of post-its spewing out of my copy of Notes on the Synthesis of Form, save for the obvious excerpts.

Page 15

Independent discussion of conceptual integrity, a decade before MMM

Page 17
Reread regarding the hazards of the form-context boundary.
Page 22
We recognize good fit only as an absence of misfits.
Page 24
Alexander remarks on the absurdity of conventional requirements gathering.
Page 37
Alexander makes an argument against architectural darwinism
Page 50
Alexander remarks that in "unselfconscious" cultures, it is natural to correct a failure upon recognizing it, such that there is "no deliberation between failure and correction".
Page 53
It is much easier to be a critic than an artist.
Page 65
There is no reason to believe that the cognitive categorization of fitness variables will correspond to the form's subsystems
Page 66
Introduction to extension and intension. Confer Meaning and Necessity by Rudolf Carnap.

Every concept can be defined and understood in two complementary ways. We may think of it as the name of a class of objects or subsidiary concepts; or we may think of what it means. We define a concept in extension when we specify all the elements of the class it refers to. And we define a concept in intension when we try to explain its meaning analytically in terms of other concepts at the same level.

Page 68
Here we have Alexander talking about the overwhelming likelihood of a mismatch between the optimal decomposition of a complex problem with the mereotopology of well-known concepts with tidy labels such as acoustics or neighbourhood. The effect of trying to shoehorn the decomposition pattern of the problem into the structure of the language yields subsets like the striped, amoeba-shaped one in the third figure below.

From the neighbouring text:

Take the simple problem of the kettle. I have listed 21 requirements which must take values within specified limits in an acceptably designed kettle. Given a set of n things, there are 2n different subsets of these things. This means that there are 221 distinct subsets of variables any one of which may possibly be an important component subsystem of the kettle problem. To name each of these components alone we should already need more than a million different words—more than there are in the English language.

His point is that committing to a prescriptive taxonomy/typology/ontology designed up front has deleterious effects on the ability to solve design problems, because the preordained concept scheme cuts across strong connections between structural features of the actual problem space, and bundle up weak ones.

Page 69
Once you name it you can't change it.
Page 75
What is Alexander's definition of invention?
Page 76
I'm beginning to think there is some value in just plodding along, just not as much room for grandiosity.
Page 83
Just like how organizational structures are decomposed.
Page 90
Functional specifications do little to define forms
Page 92
This is what I was talking about WRT throwing the project out (bottom of page)
(That is, intension vs. extension)
Page 95
Strip the problem of semantic undertones implied by nominalization
Page 102
Since the set of misfits can never be called complete, the client must understand that his product will never be perfect

For a problem like an urban dwelling, if we ask different designers to state the problem, we may find it hard even to get agreement about what the relevant issues are. Probably each designer has his private set of hunches about where the issue really lies. The designer is free to look at a problem in any way he chooses; all we can hope to do is put a fruitful structure on his view of it. It is for this reason that M cannot be thought of as objectively complete, and has been presented, instead, in Chapter 6, as a picture of a designer's view of a problem. However, it should be pointed out that in spite of the natural bias which any one designer's statement of a problem is sure to carry, at the same time the use of the set M as a means of representation does have in it one great claim to neutrality. What designers disagree about is the relative importance of different requirements. In the present theory this would have to be expressed, if it were expressed, at all, by assigning some sorts of weights or values to different variables. However, few designers will actually disagree about the variables themselves. While the relative importance of different requirements usually is a matter of personal opinion, the decision that a requirement either is a requirement or isn't, is less personal.

One more time, with feeling:

What designers disagree about is the relative importance of different requirements…However, few designers will actually disagree about the variables themselves.

Page 103

Prioritize by opportunity (also this is where architectural authority comes in handy)

By leaving the designer to work out the relative importance of different requirements at his own discretion during the diagram phase of the design process, it is therefore possible for designers to agree about the contents of the set M, whether or not they agree about their relative importance, because mere inclusion of a requirement in M, as such, attaches no weight to it.

Page 105
Do some math-larnin'
Page 107
Step 1: get the goddamn data!
Page 109
Argument in favour of a narrative and an audit trail of decisions
Page 112
Groom set M to separate higher-order interactions
Page 113
Groom set M so that all members have the same rank and scope
Page 114
Page 115
Making the case for ethnographic research, personae etc; do not attempt top-down design at this phase
Page 125
Mix it up a little!
(I have no idea what I meant by this)

Aaand the rest of the book is appendices and notes, and I refuse to take notes on notes on Notes on the Synthesis of Form.