The 1000 Second Method

I spent a year writing about habits for dads in tech. About 100 people subscribed. The newsletter died a quiet death and I am genuinely fine about it.

But one thing kept working when everything else didn't.

16 minutes and 40 seconds.

That's 1000 seconds. It's the smallest chunk of focused work I've ever found that actually stuck. Stuck across a director-of-product job, two small kids, a personal AI agent I'm building on the side, a marriage I care about, and a body I keep trying to keep alive past 40. Not 25 minutes. Not 45. Not a "deep work block." A thousand fuckin seconds.

I didn't pick the number. The number fell out of the constraints.

The unit

Here's the move. You pick one outcome. You give it 1000 seconds. You install it daily, with a fixed trigger and a written intention. You don't celebrate completing it. You log it and move on.

That's it. That's the whole thing.

The reason it works is not "small habits compound" or any of the James Clear stuff (which is great, by the way, just not the lens here). It works because 1000 seconds is the smallest duration that satisfies four behavioral-science constraints at once. If you go shorter you miss one. If you go longer you violate a different one. 1000 is the floor.

Let me show you the four.

The four constraints

1. Implementation intention (Gollwitzer & Sheeran, 2006)

The single most replicated finding in behavior-change research is that intentions don't change behavior. Implementation intentions do. Peter Gollwitzer and Paschal Sheeran's 2006 meta-analysis of 94 independent studies found that specifying when, where, and how you'll do a thing produces a medium-to-large effect on follow-through (d=0.65). Just intending to do it produces basically nothing.

1000 seconds is short enough that you can pre-commit a specific 1000-second window: 6:20 AM after coffee, 1:00 PM after standup, whatever. The window is concrete. That makes the implementation intention possible.

Try writing a 90-minute implementation intention for a Wednesday. You can't. Your day won't hold it. The window is too big to anchor.

2. Chunking under the attention floor (Huang, Jin & Zhang, 2017)

Stephanie Huang and her collaborators ran a sequence of experiments on goal pursuit and found that sub-goals of an intermediate, identifiable size produce more follow-through than either tiny micro-actions or large multi-hour pushes. Tiny actions feel insulting. Large actions feel like a commitment your future self has to make on behalf of your present self, and your future self always votes no.

(Side note: the chunking literature is messier than I'm making it sound. There's no magic 1000-second number in the papers. What there is, is a robust finding that the "right size" sits in the 15-25 minute range for cognitively demanding tasks. 1000 happens to be a clean number inside that range and it makes the math easy. Don't @ me.)

3. Commitment without consequence inflation (Ariely, Karlan & Giné)

Dean Karlan and Xavier Giné's commitment savings work in the Philippines, and Dan Ariely's broader writing on pre-commitment, both point at the same trap: if your commitment device has too big a downside, you avoid the commitment entirely. If it has no downside, you don't follow through.

1000 seconds is small enough to commit publicly without bracing for catastrophe. "I'm gonna run my 1000 seconds at 6:20 AM" is a sentence you can say out loud without your nervous system filing a protest. "I'm going to do 90 minutes of deep work every morning before the kids wake up" is, for most parents, a lie you are telling yourself in front of witnesses.

4. Identity-confirming threshold (Fishbach, plus my own dumb experiments)

Ayelet Fishbach's research on goal pursuit shows that the moments most likely to derail follow-through are the moments right after you celebrate a sub-goal. You complete something small, you give yourself a hit of "I did the thing," and then you license yourself to quit. Sub-goal celebration is the productivity equivalent of eating one Pringle.

1000 seconds is long enough that completing one feels like you actually did the thing for the day, but short enough that you don't need to celebrate it. There's no "I crushed it today" feeling. There's a quiet "did the work" feeling. You log it. You move on.

This is the constraint most people violate. They try to make the 1000 seconds feel important. The whole point is that it doesn't feel important. It just happens.

The trap (the one I'm still working on)

The trap is Hershfield and Benartzi's future-self problem. Hal Hershfield's research is the one where they age-progress photos of people and show that subjects who see their older selves save more money for retirement. The general principle: humans treat their future self as a stranger. Most of our broken habits are us defecting on the stranger who lives in our body next year.

1000 seconds works against this because the future-self horizon for "did I do my 1000 seconds today?" is 24 hours, not 24 years. You're not bargaining with the stranger. You're bargaining with tomorrow morning's version of yourself who already agreed to this. That's a fight you can win.

But the trap is real. The 1000 seconds has to be in service of something that matters to the long-horizon you. Otherwise you're just doing a daily ritual that compounds into nothing. NGL, I've watched myself do 1000 seconds of "writing" that was actually 1000 seconds of editing one paragraph 13 times. Doesn't count. Has to be in service of an outcome the year-from-now you would recognize as worth it.

5 things I actually do for 1000 seconds a day

In rough order of how non-negotiable they are:

1. Movement. 1000 seconds. Could be a kettlebell circuit, could be a walk with a podcast, could be a 4-minute warmup then 12 minutes of Zone 2 on the rower. Not a workout. A floor.
2. Writing. 1000 seconds of putting words on a page. No editing, no research, no Twitter. Words on a page. This is the one I'm worst at and the one with the highest return.
3. Deliberate practice on one product skill. Right now: writing PRDs that one-shot in Claude Code. Two months ago: building MCPs. The skill rotates. The 1000 seconds doesn't.
4. Connection. 1000 seconds of fully present time with my wife or one of the kids. Phone in the other room. No agenda. This one is harder than the workout and I rank as a 6.5/10 on it. Honest grading.
5. Agent brief. 1000 seconds writing or refining one brief for my personal AI agent. This is the one that's compounded fastest in 2026. The agent gets sharper, my asks get sharper, the loop tightens.

Five outcomes. 5000 seconds. 83 minutes. Adds up to less than a single "deep work block" that I've never actually completed.

Why a cohort changes everything

Here's where I'll come clean. The 1000 Second Method works alone. It works better in a small group. Not because of accountability theater (which is mostly bullshit) but because the implementation intentions get stronger when other people know the window. You don't want to be the operator who skipped Tuesday's 1000 in front of seven peers who didn't.

I'm running this as a 6-week cohort starting in a couple months. 8 to 12 seats. It's called the Grit Collective: Operator Cohort. I'm not pitching it today. I'm flagging it because if you try the method solo and hit the installation wall, the cohort exists.

For now, the move is simpler.

Try it

Pick one of the five. Pick a window. Pick a 1000-second timer. Run it tomorrow.

If you want a faster way in, I built a 15-minute audit that tells you which of your last 1000-second blocks were leverage and which were noise. Free. PDF plus a Google Sheet you can run on a single day of your last week.

Want to try it? The 1000 Second Audit takes 15 minutes. Free.

Try it. Let me know what breaks.