The First Law of Complexodynamics (2011) | Hacker News Reader

The First Law of Complexodynamics (2011)

scottaaronson.blog

40 pointsibobev10 comments

There’s been a bunch of progress in the field since then. Oliver de Weck from MIT and his group have devised a way to calculate system complexity, correlates that with both cost and cognitive effort and shown, that it indeed has a hump-like shape on simulated systems. Furthermore, one can demonstrate, that in some sense optimal number of connections between system elements exists and that this optimum is linear in relation to the number of nodes for k>10. Thus, you can calculate the number of connections between your microservices beyond which you are better off with a monolith.

Can you say more about what sense a linear number of connections is optimal? Minimising complexity doesn't always strike me as the right goal. I believe many problems require complex solutions, and view microservices more as a way of helping humans manage that complexity (at some cost, of course).

I have previously encountered a strong working definition of complexity in ergodic theory (divide a wandering set into subsets based on their behavior, graph their respective measures over time and do some analysis) and I'm surprised to see no use of its tools referred to here. If anybody has followed along with this project, do you know if they've since investigated the applicability of the notions of topological entropy and Hopf decomposition to their question?

Discussed at the time:

The First Law of Complexodynamics (2011) - https://news.ycombinator.com/item?id=14966873 - Aug 2017 (6 comments)

Shouldn’t it be that we don’t talk about complexodynamics?

Burn, karma, burn.

This was not discussed in the older thread, but the complexodynamic graph can probably be used to model technological society.

My fave example is Mongolia.

The area started off developing horseback archery at the end of the neolithic, then in the 12th C they had the world's most advanced metallurgy, gunpowder, bioweapons, postal system, siege engines, and nowadays we are back at admiring their horseback archery.

How does China fit in that scheme [1]?

[1] https://en.m.wikipedia.org/wiki/History_of_China

For the benefit of the model you can consider china to be part of Mongolia from 12th C to 14th C

The background of my question is, that China is around for a long time, sometimes technically advanced, sometimes not. It is still around today. It does not seem to fit into your scheme.

That's why I focussed on Mongolia, and should have emphasized that china should not be thought of as part of mongolia after the fall of the Yuan. It was indeed most of the driving force for the temporary increase in Mongolian sophistication. Maybe we don't want to think of the "china phase" as geographic china but Chinese knowhow?

Another way in which geographic china doesn't fit into the scheme but historical Mongolia does is that china has never really been a closed system. Geographic Mongolia however was landlocked like the closed system in the featured article. (And this is why they fought hard but failed to expand to the seas)

Reply on news.ycombinator.com