August 29, 2023

Fast vs Frenzied pt 2

Longtime friend-of-the-list, Tom Ryan wrote in with a fascinating follow-up to my recent message on doing less to move faster (shared with permission):

This stuff is modeled really, really well in the math of performance, utilisation & congestion: queuing theory. Recommend. Was mind-blowing for me in the same way as How to Measure Anything: persuasive, pervasive, impossible to unsee.

At its core (the core of my superficial understanding, tbh) I found this dichotomy:

Utilisation is bounded. Congestion ain’t.

The problem is almost always congestion, and congestion isn’t best solved by increasing capacity or, god forbid, by increasing utilisation.

The best way to improve the overall throughput is to do less work for each request/customer/project/chapter/commit/whatever.

The second-best way is to batch similar stuff.

Here’s a nice intro vid in case you’re curious, otherwise thanks for reading this far 😂

(Actually 26mins, not 48 as claimed)


I love the distinction between utilization and congestion!

Thanks so much to Tom for sharing this :-)