Thread by @johncutlefish

How a single shared infra/ops team bring an org to its knees, while no one does anything about the problem? The problem is so obvious and solvable. Well...

1. Rampant Tetris playing puts these teams in a no-win situation. Trust drops.

2. When trust drops, it is hard to hire up. So the shared team stays perpetually under-supported. It is commonly seen as a "management issue", so more human load balancers are brought in to "fix it".

"Oh well, you know Ops...dropping balls again"

3. Then...more elaborate Tetris playing. Big quarterly planning sessions, dependency bingo, "rough estimates", 20%-20%-20%-20%-20% splits, too busy to actually slow down and assess needs/actual problems

4. Then...other teams work *around* the shared team, even if that work is less valuable, or if the work actually subverts the shared team's effort. At this point someone will say...

"Do we really need [Shared Team]? Seems like teams are doing just fine by ignoring them!"

Anyway. All sorts of things conspire to make a relatively straightforward issue way more opaque and slippery. So you find big teams floundering *more* because all of their sub-optimizations and circumventing the problem, than the actual problem itself.