Picture the version of competence that depends on everyone around you doing their job correctly. Your subordinates report accurately. Your allies follow through. Your own judgment gets confirmed, again and again, by people you can rely on.
Now picture what happens when that stops being true. Ulysses S. Grant, early in the Civil War, found out. His authority didn't collapse because he lost a battle. It collapsed slowly, from the inside, because he could no longer trust the information reaching him.
This is Robert Greene's fourth illustration of Strategy 3, the Counterbalance Strategy — presence of mind as something actively rebuilt, not just naturally possessed.1
Early in the war, subordinates passed Grant inaccurate terrain reports. Captains failed to follow through on his orders. Other generals openly criticized his plans.1
Grant was stoical by temperament — not a man prone to visible panic. But his diminished control over his own troops produced a diminished control over himself. He drank to cope with the slippage.1
It's tempting to read this as a personal failing, and Greene doesn't excuse it as anything else. But the mechanism matters more than the moral verdict. A commander who can't trust the information reaching him is, functionally, operating half-blind — every decision now carries the added uncertainty of not knowing whether the inputs themselves are reliable. That's a genuinely different, harder problem than making a hard call with good information. Grant's drinking wasn't a separate failure alongside the command problem. It was one visible symptom of trying to operate inside an epistemic fog nobody around him was helping to clear.
By the 1862–63 Vicksburg campaign, Grant had changed his method entirely.
He rode the terrain himself instead of trusting secondhand reports. He reviewed intelligence personally rather than relying on filtered summaries. He honed the precision of his own orders, making them harder for subordinates to misread or ignore.1
And once he'd made a decision, he stopped waiting for his fellow generals' approval. He trusted his own judgment and moved.1
The change wasn't that Grant suddenly trusted people less. It's that he stopped needing to trust them as much, because he'd rebuilt his own direct access to the information that mattered. Self-reliance, in Greene's telling, isn't about isolation or refusing help — it's about not letting your composure depend entirely on other people's competence, especially in a role where their competence is genuinely outside your control.
Vicksburg fell to Grant's forces after a long, difficult campaign — not because the terrain had gotten easier or his subordinates had suddenly become more reliable, but because Grant no longer needed them to be reliable in the same way. His feelings of helplessness, and the emotional volatility that came with them, dissolved along with the dependency that had produced them.1
There's a version of this that could easily curdle into paranoia — a commander who trusts no one, verifies everything personally, and burns out from the sheer workload of never delegating. That's not what Greene's account describes, and it's worth distinguishing carefully. Grant didn't stop delegating command decisions. He stopped delegating the specific inputs his own judgment depended on — terrain, intelligence, the raw shape of the battlefield. Everything downstream of that verified picture, he still handed to his officers to execute. Self-reliance here means owning the foundation your own composure rests on, not owning every task underneath it.
You've been running a team project where your read of progress depends entirely on status updates from two team leads. Lately those updates have started contradicting what you're seeing in the actual deliverables — optimistic reports, slipping deadlines, a gap between what you're told and what you can verify yourself.
You notice you've started dreading the weekly check-in. That dread is Grant's early-war symptom, not a separate problem from the reporting gap itself.
Run the Vicksburg fix instead of the coping-mechanism fix. Go look at the actual work yourself, the way Grant rode the terrain. Review the raw data instead of the summary, the way Grant reviewed intelligence personally. Tighten your own instructions so there's less room for a status update to paper over a real problem, the way Grant sharpened his orders.
None of this requires firing anyone or assuming bad faith. It requires closing the gap between what you're told and what you actually know — because the dread wasn't really about the project. It was about not trusting your own picture of it.
Not every gap between report and reality justifies going around your team entirely, and doing so as a reflex would recreate the paranoia trap named above. The Grant move is proportionate: verify the specific inputs that have actually started diverging from what you're told, not every input across the whole project. If the discrepancy is isolated to two status updates, check those two things directly. Rebuilding your own composure doesn't require rebuilding your entire information architecture from scratch.
Grant's early Civil War struggles with alcohol and his subsequent transformation into one of the war's most effective commanders are well-documented across independent biographical sources, not solely Greene's retelling. [POPULAR SOURCE] for Greene's specific framing and emphasis.
Open tension: Greene's account risks implying self-reliance is a simple switch Grant flipped once he decided to. The historical record is messier — Grant's drinking recurred at points later in his life and career, suggesting the underlying vulnerability wasn't permanently resolved by the Vicksburg rebuild so much as it was managed more successfully under that specific set of conditions. The chapter's clean before/after structure understates how much ongoing maintenance a genuinely rebuilt self-reliance actually requires.
That recurrence isn't evidence the Vicksburg fix was fake or temporary. It's evidence that self-reliance is a practice tied to specific conditions, not a permanent character trait acquired once. Grant's method worked reliably when he had direct battlefield access to verify against. It's a fair question — one the chapter doesn't ask — whether the same vulnerability resurfaced later precisely in contexts (the presidency, for instance) where that kind of direct, personal verification was structurally much harder to arrange.
Set this beside Nelson and the Blind Eye, another case in this same chapter. Nelson's presence of mind was built through repeated exposure to external danger. Grant's was rebuilt after an internal collapse — not a failure of nerve under fire, but a failure of trust in his own information pipeline.
Read together, they suggest presence of mind has at least two distinct failure modes with two distinct fixes: courage that erodes under repeated danger needs exposure to rebuild, but judgment that erodes under unreliable inputs needs something closer to what Grant did — cutting out the unreliable middle layer and re-establishing direct contact with reality.
Business → Founder Dependency on Filtered Reporting. A founder who scales past the point of personally touching the product often experiences a milder version of Grant's exact problem: decisions increasingly made on the basis of summarized reports rather than direct observation, with no easy way to tell whether the summary is accurate.
The insight the pairing produces: "founders should stay close to the customer" is usually framed as a quality or empathy argument. Grant's case reframes it as a composure argument — the founder who loses direct contact with ground truth doesn't just make worse decisions, they become measurably more anxious and reactive, because part of what they've lost is the ability to verify whether their own fear is proportionate to what's actually happening.
Psychology → Locus of Control. Psychological research on locus of control distinguishes people who attribute outcomes to their own actions from those who attribute them to external forces — an internal locus is reliably associated with lower anxiety and better stress tolerance, even when the actual level of control hasn't changed. The insight: Grant's transformation wasn't just tactical, it was a locus-of-control shift. By taking direct action to verify information rather than depending on others to supply it accurately, he moved from an externally-dependent position (my composure depends on other people doing their jobs) to an internally-anchored one (my composure depends on what I've personally verified) — and the drop in his visible anxiety tracked that shift almost exactly.
Grant didn't become calmer because the war got easier. He became calmer because he stopped needing other people's reliability to feel in control of his own judgment. The lesson generalizes past command: dependency doesn't just create risk of being let down. It creates a specific, corrosive kind of anxiety that no amount of the other person improving will ever fully resolve, because the real fix was never actually about them.