Comment by dr_dshiv | Hacker News Reader

Comment by dr_dshiv | Hacker News Reader

He had really interesting computational ideas about harmony. From The Nature of Order (2002) “As architects, builders, and artists, we are called upon constantly—every moment of the working day—to make judgments about relative harmony. We are constantly trying to make decisions about what is better and what is worse…”

Following this, he published a paper on “harmony-seeking computation” as an approach to optimize “wholeness.” He identified 15 elements of wholeness in designs that might be measured, like the presence of coherent centers, strong boundaries, local symmetries or roughness/imperfection. He proposes that “the harmony that is sought in these computations is indeed what we otherwise call ‘beauty’. But the result of harmony-seeking computations are not merely pretty or artistic. In most cases, they are also better functionally and technically.”

Alexander, Christopher "Harmony-Seeking Computations." International Journal of Unconventional Computing 4 (2008).

I wasn't aware of this paper before now and am excited to read it.

The 15 properties of life were, I believe, his attempt to find the universal rules underlying the design patterns he described in A Pattern Language. I've long thought that most of those properties could be described algorithmically and was curious what sort of system they would converge to. I suppose nature is one easy answer, but it's not a very useful one.

For those who are interested, there's a nice summary of the properties and analysis of how they relate to each other at http://web.sfc.keio.ac.jp/~iba/papers/PURPLSOC14_Properties.... and a more architecturally-focused take at https://www.archdaily.com/626429/unified-architectural-theor....

Awesome, thank you for sharing these great resources. How did you find them?

The main thrust of Nature Of Order is that these 15 elements of wholeness are itself a quantifiable measure of life, of literally how alive a system is. Christopher believed you didn't design and build a structure and then fill it with life/humans, rather he believed the process of architecture was itself about literally increasing the life of a space. An architect's job was to recognize life-creating patterns and imbue structures with them.

He believed that a meadow with a river bend, an old willow tree and a rock to sit on below it was not a scene composed of living and non-living entities, but rather the arrangement of these entities (both internally alive and not) gave rise to a wholistic quantity of life in that space. Depending on where the rock was placed the space itself was more or less alive.

It sounds like a far out, "new agey" idea but I think he argued quite rationally that this framing of life makes far more sense than a binary determination based upon near arbitrary delineations such as sex, reproduction and consumption of resources.

I think he really was onto something personally.

Fascinating. Saw a paper pop up on one of my Semantic Scholar feeds just the other day with the title "Symmetry and simplicity spontaneously emerge from the algorithmic nature of evolution". Here's the link https://www.biorxiv.org/content/10.1101/2021.07.28.454038v2

Is it this ?

https://patterns.architexturez.net/doc/az-cf-173014

Yes

“Harmony” was also the metric used for the first multi-layer neural network by Smolensky (1983). Then Hinton called it (free) “energy”—same metric with a sign reversal. Then, with Rummelhart, they compromised and called it “goodness of fit.” Same as Harmony. So, neural nets still optimize for Harmony, by another name. I find that wonderful, somehow.

Meh. The whole "wholeness" and "structure-preserving transformations" thing just sounded to me like he was describing a literal cell division algorithm. He even gave cell division as an example of structured behavior rules producing wholeness out of chaos.

And is cell division any less magical for that comparison? We still don’t understand it well enough to build with anything close to the robustness & scalability of biological systems.

Reply on news.ycombinator.com