Are Our Offices Over Optimised?

I just finished reading an essay called How to Beat an AI System , Lessons From the Boston Celtics. It’s about over-optimisation: how chasing the statistically “best” move can also make you fragile. The piece used basketball as a metaphor, showing how the Celtics leaned too hard into three-point shooting. On average it made sense. But on nights when the shots didn’t fall, their whole system collapsed.

It made me wonder: are offices in danger of falling into the same trap?

A lovely bright NO HQ workspace in the heart of the City

Optimisation Has a Ceiling

Most workplaces of the past two decades have been optimised within an inch of their lives. Facilities teams, consultants, and landlords love numbers: desk utilisation rates, meeting-room occupancy, cost per square foot. From these metrics came open-plan floors, hot-desking, and endless meeting pods. The logic was simple: reduce friction, increase efficiency, maximise usage.

But like the Celtics’ three-point strategy, what looks efficient on paper doesn’t always hold up in practice. The average day is not every day. Variance matters. A system designed to hum along under typical conditions can feel brittle and joyless when conditions shift when the office is half-empty, when a team needs quiet, or when people crave a bit of personality instead of uniform desks.

A NO HQ workspace in the City of London.

The Value of Friction

In our push to optimise, have we removed too much friction from the workplace?

Older offices were often inefficient: too many corridors, odd-sized rooms, a strange sofa nobody knew what to do with. But those quirks created moments. You bumped into someone in the stairwell and sparked an idea. You wandered into a room too big for its purpose and had a conversation you wouldn’t have had otherwise.

Friction slowed things down, but it also produced culture. Just as a bad shot in basketball can keep a defence honest, or an odd play can disrupt an opponent’s rhythm, the small inefficiencies of an office gave work texture.

When Community Gets Over-Optimised

Coworking providers fell into the same trap. They talk endlessly about “community,” but often try to manufacture it in ways that feel forced. The classic case is WeWork: events, workshops, after-hours drinks, all engineered to create the appearance of belonging. At scale, it became formulaic, optimised for marketing decks and investor pitches more than for the messy, organic ties that make a real community.

True community has a certain looseness to it. You can’t spreadsheet it into existence. Yet many coworking brands still treat it like desk utilisation: a metric to be maximised.

Plenty of collaboration points in this workspace.

The Real Bottleneck: Medieval Leases

Meanwhile, landlords cling to archaic leases, rigid, multi-year contracts designed to protect their income streams rather than meet the realities of modern work. These structures are essentially medieval, locking companies into spaces they don’t always need, with terms that bear little resemblance to how teams actually operate.

And here’s the problem: both models are optimised in favour of themselves, not the people who use them. Landlords optimise leases for stability. Coworking operators optimise space and community for marketing. Both strategies work until they don’t. When the three-pointers stop falling — when leases become financial shackles in a downturn, or when the sheen of manufactured community wears off, it’s the tenants and members who are left carrying the risk.

Designing for Variance

If AI systems are vulnerable because they optimise for averages, then perhaps offices and coworking spaces are too. Optimisation assumes predictability. But work isn’t predictable especially in an Ai & hybrid era. Some days a team needs total focus. Other days they need to spread out on whiteboards, or gather in larger groups, or simply spend unstructured time together. Some days you want to connect with your coworking neighbours; other days you just want to be left alone.

Spaces that design for variance and lease models that allow it, will always outperform those stuck optimising the wrong thing.

The Next Chapter

We’ve seen what happens when landlords and coworking giants optimise for their own metrics: rigidity, fragility, collapse. The next chapter belongs to spaces designed around variance contracts that flex with reality, communities that grow organically, and offices that feel alive rather than optimised. That’s where the real work and the real value will be found. The next generation of offices won’t be measured by how efficiently they’re filled, but by how resilient they are to variance: empty days, bad days, and yes the coming AI days.

It’s the question we’re quietly exploring at NOHQ: what happens when you design space around people, not the other way round?