Bureaulogy

2024-12-21<br>Bureaulogy

I'm fascinated by the formation, evolution, and perpetuation of bureaucracies.

If I were to ever lose my mind and decide to go do a sociology PhD, I'd absolutely do it about bureaucracy. I'd want to establish a new subdiscipline of bureaulogy — the study of bureaucracies.

I think this subdiscipline is just as real and interesting as something like anthropology or psychology.

Some principal questions of the field would be:

Why do bureaucracies inevitably form in every organization of humans on a long enough time scale?

What are the common elements shared by all bureaucracies, and what leads to these commonalities?

Why do bureaucracies lead to inevitable organizational decline and eventual collapse?

What methods or techniques can be used to slow bureaucratic decline?

Is bureaucratic decline deliberately reversible?

This post attempts to start answering a few of these questions.

Why do bureaucracies form?

The following reasons are in no particular order.

Dunbar's number

Dunbar's number is a hypothetical number of relationships the typical person can keep track of. It's around 150. It's a little pop-psychology, but there's certainly something there.

In my experience, once the size of an organization exceeds a critical threshold in this vicinity, bureaucratic tendencies accelerate rapidly.

Below this number, it's possible for everyone to have some context on what everyone else is doing, and if they don't, the context is easily attainable by talking to 1-2 people.

Here's an example scenario where bureaucracy forms:

Alice is a longtime employee in an organization that has grown to 300 people. Bob is a new employee on the other side of the org from Alice.

One day, Bob does something stupid and causes an outage.

Alice could have told Bob the thing was stupid and would cause an outage if only she had known, but the org is so big, she can't keep track of everyone anymore, especially all these new-hires.

Alice proposes that all changes must first create a change proposal document that must be circulated and approved by all the senior engineers of each sub-organization so outages like this won't happen again.

Voila, a pretty extreme bureaucratic process has been formed (with good intentions!).

Upside-Downside asymmetry

It's generally true that actual losses are perceived as worse than missing out on possible gains.

Concretely, imagine an opportunity with a 60% chance of gaining $100, and a 40% chance of losing $50. A rational actor should probably take this chance! The expected value is $40.

But that expected value treats the dollars that might be gained as equal to those that might be lost. But the employee is not necessarily optimizing for dollars to the company — they are optimizing for their own individual status within the company.

For psychological reasons, it could be that losing $1 hurts one's status as much as gaining $4 helps it. Under such a ratio, the expected value of this opportunity to the individual decision-maker is -20 status-adjusted dollars. The decision-maker decides to do nothing instead of take the opportunity.

I've seen this play out, and the ratios are actually far worse in real life. One project blows up and loses $1M. Another is wildly successful and gains $100MM. The magnitude of the status gain and status loss are similar, despite 100x different monetary magnitude.

This is somewhat related to the idea of loss aversion but it's not quite the same thing, because it's not the individual perceiving a loss to themselves as being worse than a gain. It's their peers actually acting as if the loss is much worse than the gain.

Inclusivity impulses

It's often the case that a decision only needs buy-in from 1-3 key people to be finalized and executed.

However, in a team situation, this dynamic creates an implicit hierarchy of who is high-judgment, valuable, etc — the implication being the excluded members are not.

So most people, not wanting to be seen as rude, will simply invite the whole team to the meeting.

Blame-assignment hesitance

It's natural to avoid blame for oneself. But the flip side of this is most people also want to avoid blaming others for social cohesion reasons. It's awkward to place the blame on a colleague.

This tendency will lead to the creation of well-defined bureaucratic processes that allows one to blame the process instead of the person, because blaming the process is much less painful.

Team-oriented ownership

This one usually comes hand-in-hand with blame-avoidance tendency. If you have a service or process owned by a single individual, and that thing has an issue, it's clear whose fault it is. However, if it's owned by a team, the blame is more diffuse and less painful.

Team-oriented ownership creates communication overhead and necessitates processes to coordinate, thus creating bureaucracy.

Promotion legibility

In a very small company, small enough that there are no "teams" — less than 20 people,...