Hacker News new | ask | show | jobs
by cmenge 80 days ago
Read a super interesting paper by McEntire lately, “The Organizational Physics of Multi-Agent AI”.

In short, he proved that even AI agents exhibit all the dysfunctions one would normally attribute to human shortcomings / politics / laziness etc.

Either way, I think the point is strong: if the organization is bad, you end up doing mostly work about work which is exhausting.

Small, effective teams with super high accountability are more fun, but don't look "reproducible" or "repeatable".

Shameless plug on my take: https://www.menge.io/blog/where-to-cut/

4 comments

I feel a bit wacky even saying this, but I just started re-reading Team Topologies last week because it's starting to feel like the whole orchestration pattern only works reliably when roles and structure are clearly defined.
I love this insight, and it generalizes. Just swapping out humans with AIs won't just fix everything, because many of the biggest problems are structural or emergent.

I'm hopeful that we can use AI models to pressure test better options of social organization etc.

> but don't look "reproducible" or "repeatable".

IMO, "ish". You can reliably and repeatedly produce good teams _if_ you reliably and repeatedly invest in your people.

IMO, what's really happening is that small, effective teams aren't _fungible_ - you can't just swap people around without breaking the magic in a team, and you can't just move a team around an organization without similarly breaking the magic (although the latter _is_ way more possible).

IMO, it's sort of an organizational version of "context switching". It takes time for a team to get up to gel and get up to speed. If you're swapping out team members, you break that cohesion. If you move around teams, you (somewhat) reset that "getting ramped up" process.

"So God created mankind in his own image"

I wonder if that made it into the training set intentionally, or just as an unexpected side effect of stealing every character of text available on the internet with absolutely no curation?

Your take is far better than the OP. I didn't figure out what point the OP was trying to make.