Ashby's Law of Requisite Variety: Why Your Simple HR System Cannot Manage a Complex Workforce

60-Second Summary

Ashby (1956): a controlling system must have at least as much variety as the system it seeks to control. 'Only variety absorbs variety.'
In HR: a single competency framework, one review template, or one 'high-performer' profile has less variety than the workforce it manages. It will systematically fail on the edges.
Variety has a cost — infinite bespoke systems are unmanageable. Design question: where do we need requisite variety, and where can we compress?
Practical response: variable pay by role family, differentiated reviews by function, tiered onboarding — not one process for everyone.
Under-varied systems fail through silent attrition of people who don't fit the mold — often your most valuable outliers.

A People team rolls out one competency framework across the company. Sales loves it. Marketing tolerates it. Engineering routes around it within six months. HR calls this 'engineer exceptionalism'. Ashby would call it inevitable — because the framework has less variety than the population it's trying to govern.

What Ashby actually proved

“Only variety can destroy variety.”
— W. Ross Ashby, An Introduction to Cybernetics (1956)

Ashby's Law of Requisite Variety is a mathematical result from cybernetics: a regulator can only control a system to the extent that the regulator's variety (number of distinguishable states) matches the system's variety. A thermostat with two settings cannot regulate a room with 20 possible states. A single HR policy with one shape cannot regulate a workforce with 500 role shapes.

Where HR under-varies

HR system	Variety it has	Variety in the workforce	Symptom
Single career ladder	6–8 levels, one shape	IC, manager, TPM, staff+, deep specialist, hybrid	Great engineers becoming bad managers to progress
One review template	Same rubric all roles	Sales, research, support are different beasts	Ratings drift to middle for 'not-fit' roles
One engagement survey	40 standard questions	Frontline, remote, exec, R&D care about different things	Aggregate score means nothing
One 'high performer' profile	Confident, articulate, self-promoting	Introverts, deep specialists, cross-cultural styles	Systematic under-rating of quiet excellence
One PIP model	30-day formal process	Some issues are skill, some fit, some context	PIP used as pre-firing paperwork, not development

Requisite variety without chaos

Under-varied vs deliberately varied

One-size (under-varied)

Single career ladder company-wide
One review rubric
One onboarding path
One leadership program
One engagement survey for all

Right-sized variety

IC and manager tracks; deep-specialist paths in R&D/eng
Role-family-specific rubrics with shared spine
Tiered onboarding: role complexity, remote/on-site, seniority
Frontline / new-manager / manager-of-manager / exec tracks
Common core + segment-specific modules

Four rules for adding variety without exploding complexity

1
Shared spine, varied branches
Common principles and vocabulary at the top; differentiated implementations by role family or context.
2
Vary where the workforce varies most
The dimensions with the most workforce variety deserve the most system variety. Don't add variety on dimensions that don't matter.
3
Compress low-variety areas
Payroll, statutory compliance, standard benefits — aggressive standardisation here funds variety elsewhere.
4
Measure the miss rate on the edges
Track how the system serves the tails — deep specialists, non-standard role shapes, non-dominant cultures. Tail attrition is your variety-mismatch metric.

The quiet failure mode

Under-varied HR systems don't visibly break. They quietly select for the people who fit the mold and lose the people who don't — usually your most differentiated talent.

FAQ

Frequently asked questions

Doesn't variety add cost?

Yes. That's why you don't add it everywhere. Add it where the workforce variance actually is, compress the areas where it isn't.

How does this differ from personalisation?

Personalisation is downstream expression. Ashby's law is upstream — it says the design must have enough distinguishable states.

Where do most HR teams over-vary?

In administrative processes (invoicing, forms, tools) where uniformity would help, and under-vary in performance/reward/career where diversity would help.

Takeaways

Only variety absorbs variety — one HR system cannot govern a diverse workforce well.
Under-varied systems fail through silent attrition of edge cases.
Design shared spine + varied branches.
Add variety where workforce variance is highest; compress everywhere else.