Seneca, Darwin and Jefferson kept one: a book of passages worth keeping from whatever they read, so that reading added up instead of evaporating. This is mine. Every passage is short, verbatim and attributed, carries a note on why it matters, and links to the mental models it bears on.
29 passages · search, filter, or let one resurface
Resurfaced today
“Frequently the messages have meaning; that is they refer to or are correlated according to some system with certain physical or conceptual entities. These semantic aspects of communication are irrelevant to the engineering problem.”
Kept because it shows what to leave out. Shannon got a theory by refusing to care what the messages meant, which is the hardest kind of simplification: throwing away the part everyone else thinks is the point.
Kept because the whole supply-chain security field is a footnote to this one sentence. He proved it with a compiler that planted a back door in itself and left no trace in the source. The lesson is not paranoia; it is knowing where your trust actually bottoms out.
“We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3 %.”
Kept with its second sentence, which nearly everyone drops. The quote is used to excuse slow code and to forbid thinking about speed. Knuth asked for neither: measure, find the three percent, and work hard there.
Kept because it is obvious and almost nobody acts on it. Hamming's test was to ask colleagues at lunch what the important problems in their field were, then why they were not working on one. Most could not answer the second question.
Kept as the antidote to waiting for the right conditions. Hamming's examples were people who did their best work with worse equipment than they wanted, because the constraint forced the idea. Treat the missing tool as the problem statement.
Kept because every bias on this site is a way of fooling yourself, and Feynman names the fix before the problem. The rest of the talk is about what he called bending over backwards: reporting what might make you wrong, not just what makes you right.
“I'm talking about a specific, extra type of integrity that is not lying, but bending over backwards to show how you're maybe wrong, that you ought to do when acting as a scientist.”
Kept as the standard for any public claim, including the ones on this site. Not lying is the floor. The extra integrity is volunteering the evidence against yourself before anyone asks for it.
“The whole acts as one market, not because any of its members survey the whole field, but because their limited individual fields of vision sufficiently overlap so that through many intermediaries the relevant information is communicated to all.”
Kept because it describes how any large organisation actually works, not just markets. Nobody sees the whole. Coordination comes from overlapping local views and a signal that carries between them, which is why the signal matters more than the plan.
“The most significant fact about this system is the economy of knowledge with which it operates, or how little the individual participants need to know in order to be able to take the right action.”
Kept because it inverts the usual praise of markets. The point is not that prices are clever but that they let people be ignorant and still act well. A good interface does the same thing for a programmer.
Kept for its compression. Each herdsman gains the whole benefit of one more animal and bears a fraction of the cost, so each adds one, and the pasture dies. The sentence applies to shared test suites and on-call rotas as well as pastures.
Kept because the whole paper is built on a borrowed sentence about something else. Some problems are shaped so that a better machine cannot fix them, only a change in the rules. Knowing which kind you face saves a great deal of engineering.
Kept because planning what to do is easier than doing anything, and feels like progress. Graham's claim is that the right problem only becomes visible from inside some problem. Start, then steer.
Kept as a decision rule for the many choices that have no data. When two options look equal on paper, the one you keep reading about at midnight is not equal.
“If you're excited about some possibility that everyone else ignores, and you have enough expertise to say precisely what they're all overlooking, that's as good a bet as you'll find.”
Kept because it has two conditions, and people remember only the first. Excitement about a neglected idea is cheap. Being able to say precisely what the crowd is missing is the part that makes it a bet rather than a hope.
Kept because the cost of a meeting is not its length. The afternoon is a unit, and a thirty-minute call in the middle destroys the unit. Whoever owns the calendar should read this once a quarter.
Kept because it names the thing ambitious people refuse to do. Recruiting users one at a time feels beneath a company that intends to be large. Graham's examples show it is how the large ones started.
Kept because it sets the right expectation. An abstraction saves you from understanding what is underneath, until the day it does not, and that day arrives for every abstraction worth having. Budget for learning the layer below.
“With a sufficient number of users of an API, it does not matter what you promise in the contract: all observable behaviors of your system will be depended on by somebody.”
Kept as the reason that changing anything public is harder than it looks. The contract is what you meant. The interface is everything anyone can see, including the bugs and the timing.
“Organizations which design systems (in the broad sense used here) are constrained to produce designs which are copies of the communication structures of these organizations.”
Kept in its original wording, which is stronger than the slogan. Not tend to, not often: constrained. If the architecture you want needs two teams to talk and they do not, draw the org chart first.
Kept in the 1980 wording. The principle built the internet and is now argued over, because being liberal in what you accept lets errors live forever. Both the rule and the argument against it are worth holding at once.
Kept as the first question to ask when behaviour makes no sense. Munger's examples were a courier firm paying by the hour and a salesman who believed his own pitch. Before explaining people as foolish, ask what they are paid to do.
“We might as reasonably dispute whether it is the upper or the under blade of a pair of scissors that cuts a piece of paper, as whether value is governed by utility or cost of production.”
Kept because it ended a century of argument with a household object. It also settles many arguments that are not about prices. When two factors are each called the real cause, check whether they are two blades.
“They took away the old timbers from time to time, and put new and sound ones in their places, so that the vessel became a standing illustration for the philosophers in the mooted question of growth, some declaring that it remained the same, others that it was not the same vessel.”
Kept because the oldest puzzle about identity is a maintenance log. Every long-lived codebase is this ship. The question of whether it is the same system matters less than who still knows why each timber was replaced.
Kept as the best example of a field being founded by a definition. Everything that followed, from compression to error correction, sits inside this sentence. Note the word selected, which is where the whole theory hides.
“Frequently the messages have meaning; that is they refer to or are correlated according to some system with certain physical or conceptual entities. These semantic aspects of communication are irrelevant to the engineering problem.”
Kept because it shows what to leave out. Shannon got a theory by refusing to care what the messages meant, which is the hardest kind of simplification: throwing away the part everyone else thinks is the point.
“There is no single development, in either technology or management technique, which by itself promises even one order of magnitude improvement in productivity, in reliability, in simplicity.”
Kept because it has been tested against forty years of silver bullets and holds. Each new tool, including the current one, is sold as the exception. The base rate says to expect a good gain and not a tenfold one.
“The hardest single part of building a software system is deciding precisely what to build. No other part of the conceptual work is so difficult as establishing the detailed technical requirements, including all the interfaces to people, to machines, and to other software systems.”
Kept because it moves the difficulty to where it actually is. Code that does the wrong thing well is the commonest failure, and no amount of typing speed helps with it.
Kept because it quietly gives up something everyone assumes: that events have an order. In a system with more than one clock, before is only partly defined, and most distributed bugs are someone forgetting that.
“It operates by association. With one item in its grasp, it snaps instantly to the next that is suggested by the association of thoughts, in accordance with some intricate web of trails carried by the cells of the brain.”
Kept because it is the design brief for the web, the wiki and this book, written in 1945. Filing systems sort; minds link. A commonplace book is an attempt to keep notes the way the mind keeps them.
“The summation of human experience is being expanded at a prodigious rate, and the means we use for threading through the consequent maze to the momentarily important item is the same as was used in the days of square-rigged ships.”
Kept because it is still true, with better ships. The volume of what is known grows faster than the tools for finding the right piece of it, and the gap is where most of the useful work in software lives.
