Working Hard Means Working Deep and Resting on Purpose
I keep coming back to one idea about work ethic: hours matter less than how they’re spent. A hard worker spends them on one hard thing at a time and rests as deliberately as they work. Cal Newport’s Deep Work is the clearest statement of it I’ve read, and I’m writing down the parts I plan to use this term.
One thing at a time
Newport defines deep work as professional activity done in a state of distraction-free concentration that pushes your abilities to their limit. The argument for it rests on attention residue, a finding from Sophie Leroy’s research: when you switch tasks, part of your mind stays on the last one, and your performance on the next one suffers.
Switching costs more than the seconds it takes to open another tab. A morning split between a paper, email, and a half-written proof gives you three tasks that each got a fraction of your attention. The alternative is to finish one, or to reach a clear stopping point, before touching the next.
Close the drawer
Single-tasking only works if you can put a task down cleanly. I picture it as a cupboard with a drawer for each subject. To move to something else you shut one drawer, and to stop working you shut them all.
Most people can’t do that on command, so it helps to make it physical. Before I stop, I write two lines: where I am, and what the next step is. Newport suggests a shutdown ritual at the end of the workday, where you check that everything is captured and then say the day is over. The point is to give your mind permission to stop, because it keeps circling open loops until it knows they’re written down.
Build systems so fewer decisions need you
Good workers keep their attention for the decisions that need it. For a student that means a reading queue instead of deciding daily which paper is next, a weekly plan instead of a daily scramble, and a fixed place for notes. Each of these removes a small decision from the day.
Plan for the setback
Deadlines rarely go as planned. It helps to write down, before the work starts, what could go wrong and what you will do if it does. If the proof doesn’t work by Thursday, I switch to the simpler case. If the experiment fails, I check the data pipeline first. A short list like this costs ten minutes and keeps a bad afternoon from turning into a lost week.
Rest as its own block
Newport argues that deep concentration is limited. His reading of the research on deliberate practice is that even experts sustain around four hours a day, and beginners less. Past that, extra hours mostly buy tiredness.
That changes what rest is for. It is part of the work. Downtime lets your mind keep processing problems you’ve stopped thinking about, and a firm stop in the evening means you return with energy. What doesn’t count is half-rest, where you are technically off but still checking messages. I want rest to be complete, the same way I want work to be. Sleep belongs in the plan too.
What I’m doing this term
I’m taking Human-AI Interaction and Formal Verification, which means a lot of paper reading and proof work. I’m putting that into long blocks with the phone in another room, stopping at a set time each day, and writing two lines to close each session. Rest goes on the calendar as a real block, with a proper break after the hard days.
I’ll share the paper readings and projects from these courses here on the blog.
Enjoy Reading This Article?
Here are some more articles you might like to read next:
- Building a fast 1BRC solver in C# on Apple Silicon
- Memory Management in Production: Avoiding the Silent Killers
- Growing as a Software Engineer in the Age of LLMs and Vibe Coding
- Taming the Imagination: A Comprehensive Guide to Handling Hallucinations and Implementing Guardrails in Agentic AI
- Building an Agentic AI Analytics Dashboard: A Deep Dive with LangChain, LangGraph, and Fine-tuned LLAMA
- Neural Algorithmic Reasoning: Teaching Neural Networks to Think Like Algorithms
- OpenTelemetry in Microservices: How We Transformed Our Observability Strategy
- Using AI Tools in Software Engineering and Research Workflows
- The Performance Crisis: How We Rescued a .NET 8 Microservice from 10 Critical Bottlenecks
- Zero Downtime Deployments in Kubernetes: How We Keep Our Services Running While We Ship