Context
In computer science, thrashing is a specific kind of failure.
The system is not idle. It is not broken. It is, by every visible metric, working hard. CPU usage is high. Processes are running. Activity is constant.
But throughput has collapsed.
The machine is spending more time managing context switches than completing work. Each new task added to the queue makes the problem worse. Recovery becomes harder the longer it runs in this state.
The reason it’s dangerous: from the outside, it looks like diligence.
Insight
Knowledge work has the same failure mode. And most people don’t recognise it until they’re already deep inside it.
The cause is almost never volume alone. It’s fragmentation.
Too many partially-open loops — unresolved decisions, ambient messages, shifting priorities, strategic and operational thinking interspersed every 15 minutes — begin degrading cognitive coherence. The brain starts paging. You stop thinking in complete arcs. You skim, react, and context-switch. Simple tasks feel inexplicably heavy. Not because they are difficult, but because the underlying state is corrupt.
Thrashing has a recognisable psychological signature:
- You feel simultaneously urgent and unable to start
- You seek motion as relief — checking Slack, reorganising notes, opening tabs
- Everything feels important and nothing resolves
- Rest feels unavailable even when load briefly drops
The trap is that this state impersonates productivity. The thrashing person often appears highly committed. They are responsive. They are in every thread. They are visibly moving.
But motion is not throughput.
The dangerous part is that thrashing impersonates diligence. The overwhelmed person often appears highly committed.
Implication
The problem compounds at the organisational level.
When individuals thrash, teams inherit the pattern. Coordination overhead rises. Meetings multiply to manage the fragmentation caused by previous meetings. Ownership blurs. Escalation becomes the default resolution mechanism.
The result is institutional thrashing:
| Symptom | What it looks like | What it signals |
|---|---|---|
| Endless coordination | Recurring syncs that don’t close | Fragmented ownership |
| Duplicated reporting | Same data reformatted everywhere | Low trust + low clarity |
| Escalation dependency | Everything needs a senior decision | Unclear decision rights |
| Perpetual reprioritisation | Quarterly plans revised monthly | No forcing function |
Everyone moves. Little resolves.
The deeper issue is temporal geometry. Modern knowledge work delivers every input at the same psychological priority — a CEO escalation and a Slack notification hit the nervous system identically. Without deliberate hierarchy, the system flattens all signals into urgency. This is catastrophic for judgment.
A calm operator is not necessarily less loaded. They are better at preserving dimensionality: distinguishing what matters from what is noise, what is reversible from what is waiting, what deserves depth from what deserves abandonment.
Action
Computer systems recover from thrashing by reducing active load. Humans are similar.
The recovery is not intensity. It is reduction.
Reduce active cognitive inventory. Fewer concurrent priorities. Fewer open decisions. Fewer status surfaces. The goal is not laziness — it is preserving the conditions for uninterrupted cognition.
Finish before starting. Partially-completed work is dead weight in working memory. It costs attention even when untouched.
Externalise memory. Systems, not recall. Trusted external structure frees working memory for actual thinking.
Protect depth blocks. Uninterrupted time is not a luxury. It is the only state in which complex work actually completes.
Distinguish signal from emotional urgency. Not everything that feels urgent is important. Not everything important feels urgent. The nervous system is a bad prioritisation engine without deliberate override.
One diagnostic question worth carrying:
“Am I doing work, or managing the consequences of too much simultaneous work?”
The distinction matters. Many organisations accidentally optimise for visible activity rather than completed thought.
A machine at 100% utilisation may already be collapsing.
So may a human.