Running multiple organizational experiments flies in the face of common wisdom

Many startup founders want to fix all the broken things they’ve seen in companies, redesigning all the mechanisms from scratch. This often admirable instinct is attributable in large part to the fact in order to start a startup in the first place you need to naturally think about how the world could be improved and different. Almost by definition startup founders are more comfortable changing the world by building something new than working within existing systems, and this attitude often applies to how they organize and run their companies in addition to the products they build. Examples include Zappos’ Holacracy, Google’s attempt to not have managers, not having an HR department or a sales team, attempts to put literally every decision in writing, etc.

While these experiments are essential to introducing new ideas to how organizations can run, each new change you introduce creates multiplicative risk for the organization as a whole.^1 As a result, common wisdom has become that you can get away with changing one thing at a time. Exactly one is of course a heuristic rather than a hard limit. The more changes you make, the higher the chances that either the organization will fail to accomplish its main task (create and sell a great product and make lots of money) or revert to more ‘traditional’ structures - bringing in management and an HR team, introducing pay tiers, etc.

Startups are not the only examples of this phenomena. People have set out to completely reinvent how an organizational structure worked and came to the conclusion that it didn’t work in all sorts of situations. Kibbutzes started off as radically communal structures (even the kids were communal) that eschewed the idea of money and now resemble relatively quaint self-sustaining towns. Various country-level communist experiments are another example - most of them started off as radical departures from more ‘traditional’ government structures and over time have experienced some combination of failure or reversion to the mean.

At the end of the day, this is basically applied Chesterton’s Fence

^1: It feels analogous to mutations in DNA. Occasional mutations are good for the species but too many can quickly make an individual inviable.

Web URL for this note

Comment on this note