"The more we reduce life's decisions to quantifiable formulas, the more automated we can make the simple tasks — freeing the human mind for true, ambitious deep thinking."

I do not know where the letter e is.

I can type it four hundred times an hour without an error, and if you asked me to point at it on a diagram of a keyboard I would have to mime the motion first and read the answer off my own hand. The knowledge is in the hand. It was put there in a keyboarding class at Sts. Joseph and John, a long time ago, by somebody drilling a room full of children on a skill none of us wanted, and it has been running unsupervised ever since.

That is the creed's tenth claim in its purest form, and the best evidence for it. A decision that used to be a decision is not one anymore. The capacity that used to go into finding the key is available for the sentence. Everything I have ever written was written on top of that transfer, and I have never once thought about it while writing.

Multiply that by the morning. Lights, music, radio, the dresser, the route. Every one of those was a decision and is not one now, and each one hands back a small amount of whatever it is that decisions spend.

Here is the question I did not ask myself for about twenty years. Where does it go?

Not rhetorically. Literally. The morning frees up some measurable quantity of attention and deliberation, and that quantity does not evaporate; it gets spent on something. On a good day it gets spent thinking about a problem at work on the drive. On an ordinary day it gets spent on sports radio and whatever is on the phone at a red light. On a bad day it gets spent worrying.

My own ism says automating the simple tasks frees the human mind for true, ambitious deep thinking. It says that as though the second clause followed from the first. It does not follow. It is a hope attached to a mechanism, and the mechanism works reliably and the hope does not.

Stated as a rule, and the statement has to include an admission the original hides:

The variables are the decisions in a person's day, sorted into those that have been converted into formulas and those that have not, and the capacity that conversion releases.

The quantity maximized is that released capacity. The original is explicit about it and correct about it.

The failure signal is a person who has automated a great deal and cannot say what they got. That is a real diagnostic and it is available to anyone: name what the freed time went to over the last month, specifically. Most people, including me on most months, cannot.

The domain of validity is where the released capacity is both protectable and wanted by its owner. Two conditions, neither of which the original mentions.

And the admission: the first half of the sentence is an empirical claim and the second half is a value commitment. That automating simple tasks releases capacity is a fact about how minds work, well supported, and the next section makes the case. That the released capacity should go to true, ambitious deep thinking is a statement about what a life is for. I believe it. It is not a finding, nothing in the literature establishes it, and the rest of this chapter would be dishonest if it pretended otherwise.

The empirical half has the best pedigree of anything in the creed, and the two people who stated it most clearly stated it more than a century ago.

Alfred North Whitehead put it in a book about mathematics in 1911, and he framed it as a direct attack on received wisdom, which is why it still reads as fresh. It is a profoundly erroneous truism, he wrote, repeated by all copy-books and eminent people when they are making speeches, that we should cultivate the habit of thinking of what we are doing. The precise opposite is the case. Civilization advances by extending the number of important operations which we can perform without thinking about them.

And then the image that makes the argument rather than merely asserting it: operations of thought are like cavalry charges in a battle. They are strictly limited in number, they require fresh horses, and must only be made at decisive moments.

That is a claim about a budget, and it is the same budget Simon was describing sixty years later when he said that a wealth of information creates a poverty of attention. Deliberate thought is expensive, scarce, and non-renewable within a day. Anything that spends it on a decision already made is spending a cavalry charge on an empty field.

William James had said the same thing in 1890 about habit, and his version supplies the mechanism. In habitual action, he wrote, mere sensation is a sufficient guide, and the upper regions of brain and mind are set comparatively free. The famous sentence follows: the more of the details of our daily life we can hand over to the effortless custody of automatism, the more our higher powers of mind will be set free for their own proper work.

Read those two side by side and something beyond the mechanism is shared. Both are arguments about release, and both simply assume the released capacity goes somewhere worth going. James says the higher powers will be set free for their own proper work, and the phrase their own proper work is doing enormous unexamined labor. Whitehead says civilization advances, which is a claim about aggregate history rather than about any particular person's Tuesday.

My own sentence inherits that assumption directly and states it more baldly than either of them, because I named the destination. True, ambitious deep thinking. Neither James nor Whitehead promised that, and the difference between their claim and mine is the difference between saying a resource is released and saying what it gets spent on.

The evidence on release is good. The evidence on destination is bad, and the best test anyone ever ran on it failed.

John Maynard Keynes ran that test by accident in 1930, in an essay written during a depression and pointed a hundred years forward. His economic prediction was that the standard of life in progressive countries one hundred years hence would be between four and eight times as high as it was then. That has held up remarkably well. He coined the term technological unemployment in the same essay, defining it this way:

And from the economic prediction he drew a conclusion about hours: three-hour shifts or a fifteen-hour week might be enough. The productivity would arrive, and the freed capacity would be taken as freedom.

The productivity arrived. The fifteen-hour week did not. That is close to a controlled experiment on freeing the human mind for true, ambitious deep thinking, run across a century, on an entire civilization, and the result is that an enormous release of capacity did not convert into the thing everyone assumed it would convert into.

What makes Keynes worth more than a punchline is that he saw the difficulty himself and named it before anyone had the data. Once economic necessity ends, he wrote, humanity faces its real, its permanent problem: how to use freedom from pressing economic cares, how to occupy the leisure that science and compound interest will have won. And he worried specifically that this would be a fearful problem for the ordinary person, with no special talents, to occupy himself, especially if he no longer has roots in the soil or in custom or in the beloved conventions of a traditional society.

The man who predicted the release also predicted that the release would be the hard part. My sentence asserts the release and treats the destination as automatic, which puts me a century behind an essay written in 1930.

One counter-claim I went looking for and could not sustain. David Graeber argued that a large share of modern work is meaningless, which would suggest the freed capacity has gone into invented labor, and his figures are widely repeated: something like twenty to sixty percent of jobs. Magdalena Soffia, Alex Wood and Brendan Burchell tested this against European Working Conditions Survey data across 2005 to 2015, with samples growing from around twenty-one thousand to nearly thirty thousand respondents, asking whether people report doing useful work. The proportion saying they rarely or never do useful work was four point eight percent, not twenty to sixty. And it declined over the decade rather than rising, from seven point eight percent. Their finding on what actually predicts feeling useless is about management and autonomy rather than about the job's function: respect, being able to use one's own ideas, and having enough time to do the work well.

So the most quotable version of the pessimistic case does not survive contact with the data, and I am reporting that even though the pessimistic case is the one arguing against me. The freed capacity did not obviously go into useless jobs. Where it went is a question the evidence does not answer, which is different from and weaker than the claim I wanted to make.

The design case where I watched the mechanism work and the destination fail is the one from three chapters ago.

Automating a classification removed a recurring decision from several people's days. The decision had been taking real time and real deliberation, it was being made inconsistently, and the automation was correct on the merits. Capacity was released, exactly as the claim predicts.

Nothing was done with it. There was no plan for it, because the business case had been written in the language every business case is written in, which is hours saved. Hours saved is a number that goes in a spreadsheet and it is not a destination. The hours went where unallocated hours go: into the pile of everything else that was already waiting, which absorbed them without trace, and six months later nobody could have told you the automation happened except by looking at the system.

That is not a failure of the automation. The classification is still better. It is a failure of the sentence as a complete statement, because it promises a second thing that requires a second act nobody performed.

And there is a sharper version of it in the same story. The judgment that got automated had been produced by an activity, and the activity had been producing other things nobody counted. That is the finding from the standardization chapter. Put the two together and the accounting gets uncomfortable: the automation released capacity and destroyed an unmeasured capability, and the released capacity was not directed anywhere, so the trade was a known gain in consistency, an unknown loss in situational awareness, and an unbanked credit of attention. I would still do it. I would do it differently, and the difference is entirely in what happens after the automation ships.

The boundary is protection, and it is the condition the original sentence leaves out entirely.

Released capacity is not held anywhere. It does not accumulate in an account with the person's name on it. It is simply time and attention that has become available, in a world with many claimants, and it goes to whoever claims it. That is not a moral failing on anybody's part; it is the default behavior of an unclaimed resource.

Which means automating the simple tasks only produces deep thinking if something protects the released capacity for that purpose, deliberately, against competition. A person who automates their morning and does not defend the resulting half hour has not bought thinking time. They have bought half an hour of whatever arrives, and what arrives is engineered by people who are extremely good at arriving.

The second boundary is whether the person wants it. I have built automation for people who did not experience the released capacity as a gift, because the task I removed was the part of the job they liked, or because what filled the gap was more of the work they found draining. The sentence speaks in the first person plural and assumes a shared preference for ambitious thinking, and that assumption is mine rather than universal, and imposing it through system design is a thing I have done without noticing I was doing it.

Three things this rule cannot see.

It cannot see that deep thinking is undefined, and I have been running the same trick I criticize everywhere else in this book. The chapter on defining variables says a term is defined when there is a test. There is no test here. I cannot tell you what would count as having spent an hour on true, ambitious deep thinking, which means I cannot tell whether the claim has ever succeeded.

It cannot see that some capacity should not be reallocated at all. Rest is not failed deep work. A person whose every released hour gets converted into more demanding cognitive labor has not been freed; they have been re-tasked, and the sentence as stated provides the rationale for doing that to somebody.

That one deserves more than a line, because it is the reading most likely to do harm and it is available on the surface of the words. Freeing the human mind for true, ambitious deep thinking names one acceptable destination, which by implication makes every other destination a waste. Applied to yourself that produces a low-grade guilt about any hour that was not productive. Applied to somebody who works for you it produces a justification for filling every gap you create.

I do not believe that and the sentence says it. What I actually think is that the released capacity belongs to whoever it was released from, that ambitious thought is the best thing I know to spend it on, and that those are two different claims and only the first is binding. An automation that gives somebody back an hour and no say over the hour has not freed anything. It has moved the hour from one column of somebody else's ledger to another.

And it cannot see that the ability to do the deep work may itself depend on the practice being removed. That is the ironies-of-automation argument arriving in the last place I want it: if the simple tasks were where judgment was exercised and the material was encountered, then automating all of them frees a mind that has less to think with.

The keyboarding class is the case that gives that objection its limits rather than the case that answers it. Learning where the letter e is cost me something real at nine years old and bought me every sentence since, and nothing about writing requires conscious access to the key positions. The knowledge that got automated was purely mechanical, and the capacity it fed was genuinely separate from it.

That is the easy end. Reading a drawing set, classifying a request, judging whether a customer is about to become a problem: none of those are key positions. The doing is where the understanding comes from, and the understanding is what the freed mind was supposed to think with. The distinction I would draw, tentatively, is between automating an execution and automating a discrimination. An execution can go with no loss. A discrimination cannot go without taking some of the person's model of the world with it, and the sentence as written does not know the difference.

Here is the case against, and it does not come from Keynes's grandchildren or from Graeber's meaningless jobs. It comes from what happens to any unprotected resource.

Parkinson's Law says work expands so as to fill the time available for its completion. Stated as a joke about bureaucracy in 1955, it is a claim about equilibrium: a release of capacity in a system that measures itself by occupancy will be absorbed rather than banked. The automation saves four hours a week and the four hours are gone by the following month, into meetings, into more of the same work, into the general expansion of everything that was already there.

The harder version is not about work at all. The freed capacity is a resource, and there is an entire industry whose economics depend on capturing exactly that resource, staffed by people whose full-time job is to be better at claiming it than you are at defending it. An unprotected half hour is not a neutral half hour. It is a half hour with a market value to somebody else, and they have instrumented their side of the transaction far more carefully than you have instrumented yours.

Which produces the objection in its strongest form, and it is uncomfortable because the mechanism is identical to the one I advocate. Everything in this book is a method for converting deliberate decisions into automatic responses. That method works. It is the same method used to build systems that convert a person's deliberate attention into an automatic response to a notification. I have spent a career refining a technique whose most successful large-scale application is aimed at the people I am telling to use it.

And there is no version of the creed that distinguishes the two. Reduce life's decisions to quantifiable formulas, automate the simple tasks. A recommendation engine is that sentence executed at industrial scale, on somebody else's behalf, and it is more effective at it than I will ever be at building anything for myself. The creed has no term that rules it out, because the difference is not in the mechanism. It is in who chose it and who benefits, and my sentence mentions neither.

Since the distinction is now needed in three chapters, here is a first cut, offered the way the complicated-versus-complex heuristics were offered a few chapters back: provisional, cheap, and not the instrument.

Who wrote the rule, and can the person it runs on read it? My morning sequence is a rule I wrote, can read, and can change in ninety seconds. The sequence deciding what appears on a phone is a rule somebody else wrote, cannot be read, and changes without notice. That is not a detail of implementation. It determines whether the automation is a tool or an environment.

Where does the released capacity go, and who gets it? An automation that hands the capacity back to the person it acted on is serving them. An automation that converts the released capacity into engagement with itself has released nothing; it has changed the form of the occupancy. The test is whether you end the interaction with more of your day available than when you started, which anyone can answer honestly about any system they use.

And can it be turned off by the person it runs on, at no cost, today? A thing you cannot stop is not a formula you installed. It is a condition you live in.

None of those three is decisive. All three are answerable in under a minute, and all three come out the same way on every case I have tested them against, which is weak evidence that they point at something real. Chapter 18 is where they get replaced, because a creed cannot recommend a mechanism while having nothing to say about that mechanism's largest application.

So the sentence gets an entire clause it never had, and the clause is the part that does the work.

Release without allocation is not freedom. It is capacity with no owner, and the default owner is whoever is best at collecting it. The mechanism in the first half of my sentence is real and the outcome in the second half requires a separate, deliberate, defended act that the sentence does not mention and that I did not perform for about twenty years.

What that act looks like, concretely, is smaller than it sounds, and each part is a defense against a specific way the capacity disappears.

Decide what the released capacity is for before you release it, as part of the same work. Not afterward, when the hours have already been absorbed and there is nothing left to allocate. A business case saying this saves four hours a week is incomplete in a way nobody treats as incomplete; the finished version says what the four hours are for, and if there is no answer, that is worth knowing before the work starts rather than after.

Put it somewhere it cannot be silently reallocated, which in practice means a time on a calendar rather than an intention. An intention loses every contest it enters, because everything competing with it is a specific claim on a specific hour and the intention is neither.

And check, after some months, whether the thing you named is what happened. The honest answer is often no, and the answer is only available if you named something, which is the main reason to name it. A stated destination is not a guarantee. It is an instrument for finding out that the capacity went somewhere else.

That is three lines and it is the difference between the claim being true and being a pleasant thing to say.

On the mechanism I concede nothing. Whitehead's cavalry charges and James's effortless custody of automatism are right, and the evidence is strong. Deliberate thought is a cavalry charge, the number available per day is small, and spending one on where the letter e is would be a waste of a horse. Every formula that takes a decision off the list is a real gain and it happens whether or not anybody plans for it.

On the destination I concede almost everything. A century of enormous productivity growth did not produce Keynes's fifteen-hour week. The automation in my own work released capacity that went nowhere in particular. And the most effective users of this method in the world today are using it to capture attention rather than to free it, with the identical mechanism and better instrumentation. Anyone who reads this chapter as a promise that automating the small things will make them a deeper thinker has read it wrong, and the fault would be in my original sentence rather than in their reading.

What I will not give up is the value commitment, stated as one. I think a life with fewer small decisions in it is better, that the capacity released is worth protecting, and that protecting it for ambitious thought is a good use of a human being. That is a position rather than a finding. It is the most important thing in the creed and the least defensible by evidence, and the honest form of the sentence has to carry both of those facts at once.

What survives:

The original ran an empirical claim and a value claim together with a dash between them, in a way that made the second look like it followed from the first. The revision separates them, keeps both, and states which is which.

Three things go forward. Deep thinking is undefined by the standard this book applies to everything else, which means the claim cannot be tested. The creed has no term distinguishing automation that serves the person running it from automation that serves somebody else, and that distinction is now needed in three separate chapters. And the ironies-of-automation objection reaches this claim too: if the simple tasks were where the material was encountered, a fully freed mind may be a mind with less to be ambitious about. Chapter 16 has all three.