Liberum Veto

A Case Study in the Consensus Trap

While visiting a friend in Warsaw recently, I visited several of the city’s museums. Much of Polish history is fascinating, but what struck me was the demise of the Polish-Lithuanian Commonwealth¹ in the 17th and 18th centuries, leading to the region being partitioned and absorbed by the larger surrounding powers: Russia, Prussia, and the Habsburgs. The Commonwealth had several interesting features, not least of which was the liberum veto — a parliamentary mechanism where any member of the parliament could not only block legislation but could dissolve an entire session of the assembly.

The liberum veto might seem like a reasonable device, acting as a check against both monarchical and majoritarian power while ensuring that everyone’s interests are taken into consideration. It played out terribly, however: for the better part of a century the government was hobbled, if not paralyzed, by the ability of any member to end proceedings. Earnest dissent was not the only issue: the major powers around Poland soon learned they could block any reforms not in their interest simply by bribing a single deputy. If you take the narratives in Warsaw’s museums at face value, the liberum veto was central to the collapse of the Polish-Lithuanian Commonwealth in 1795².

This echoes the corporate culture so many firms today, striving for consensus by handing a veto to all involved in a decision, no matter how remotely. In some ways, it’s very much in the spirit of the times: by construction, everyone’s voice is not only taken into account; no decision can be made that leaves someone feeling put out or left behind; and ultimately, all points of view are leveled to a rough equality, irrespective of role, knowledge, or experience. Perhaps most importantly for friction-averse management teams³, it’s a convenient recipe for avoiding conflict.

Driving this is the laudable idea that it is very human to want to be heard. Where consensus breaks down, however, is in failing to acknowledge the equally human tendency to be motivated by one’s own interests, which aren’t always aligned with the organization’s. This manifests in a number of ways⁴, but in one area where it can be crippling is in trying to change a firm’s technology or product direction.

Software products are like living organisms; as markets and technologies evolve, they need to adapt or will eventually die. Any product with even minimal success⁵ will have its internal supporters, however. The developers who built it, or maintain it currently, may defend it out of pride or fear for their jobs. Sales reps might worry about their existing clients or about having to develop a new market segment. Senior managers who claimed it as an win earlier in their career will hesitate to damage that narrative. In a culture that mandates consensus, stakeholders like these can block desperately needed transitions indefinitely.

If you find yourself at a firm with an outsized amount of legacy platforms and technical debt, it’s worth looking at how decisions are made. Consensus-driven environments are very good at providing a veneer of collegiality. For firms that need positive change to meet their objectives⁶, however, the liberum veto can be a disaster.

¹ An interesting government in its own right, one I somehow knew nothing about beforehand. Unfortunately, it’s well past the statute of limitations to blame my high school history teachers.

² Consensus came at a huge cost — it would be 128 years before Poland would again have independence.

³ Different topic, but these are increasingly the norm, particularly in firms that have a large, effectively exogenous revenue stream from a first-mover advantage or monopoly position.

⁴ Hiring is an interesting case, as a number of firms give a de facto veto to all interviewers. I’ve seen some howlingly bad non-hiring decisions made on the basis of one outlying view.

⁵ There are certainly cases where abject failures have their supporters as well, which is a bit sad

⁶ This includes virtually every firm with a software stack more than five years old