Scale is a fantastic book on the hidden order of complex systems | Hacker News Reader

Scale is a fantastic book on the hidden order of complex systems

thinkthinkthink.substack.com

100 pointsdbaboci17 comments

I read this book awhile back, and I wasn’t really impressed. There was a pervasive emphasis on the superiority of cities, but it was a highly superficial analysis of largely cherry-picked statistics; for example, using patents as a proxy for economic productivity. I also read a quote (on HN, in fact) that resonated with my skepticism in the context of Scale, (paraphrasing) “like a vacuum, Nature abhors a bare exponential; it’s usually a sigmoid in disguise.” Maybe there is a universal power law relationship in complex systems, but that seems like only part of the story.

A really good book around the same subject is Human Scale by Kirkpatrick Sale. The book was written 40 years ago and talked about how society was getting too far out of scale for humans to control.

I think the research in the field is fairly limited - which might make the book sound limited, but I didn't really see any picking of cherries, rather than very few cherries to pick from :).

Also West makes it pretty clear what he thinks of a bare exponential & corollary phase transition. The scaling power-law relationship however does not imply any sort of bare exponential growth.

"Scaling laws"[1] is the google phase (missing from OP).

Take "power law" with a grain of salt.[2] They often turn out better fit with log-normals, log-logistics, etc.

[1] https://www.google.com/search?q=scaling+laws https://scholar.google.com/scholar?as_ylo=2016&q=scaling+law... [2] A classification of natural and social distributions Part one: the descriptions https://arxiv.org/abs/1507.03408

I know that the point of this book is stated to be "that there's a clearly visible coarse-grained order that drives biological and social life."

I'm sorry, I'm probably dumb, but I'm not getting the "point" of the book based on this summary.

All the examples shown seem to say: See there are macro trends happening, some of them are exponential. And here are some correlations that prove they have an affect on the real world, seee?

Can someone explain what I'm missing?

Edit: I guess maybe the point is we need to pick up the pace of innovation as a society?

From my perspective the point of the book is to shed light on some of these macro trends that are not typically intuitive. So we expect walking speed to be random, but it is driven by the population of a city. So a lot of decisions in business or urban planning for example could learn a lot by knowing these macro trends (West has been one of the first to explore how these trends relate to scaling)

I only skimmed the book but I got the impression that it was a commentary on how scaling "laws" pervade all classes of complex systems, systems as different as a city and the human vascular system, and how understanding these led to insights across domains. I believe there is a good sam harris podcast with the author.

Read on and you find that the end result of picking up the pace is infinite growth in finite time, which is impossible, which means we will instead have a phase transition. Which is an obscure way of saying "a cataclysmic change". If we keep up the pace of innovation, something is going to change very severely, and we have no idea what or how.

I think about it in terms of weapons. Over time we increasingly advanced the innovation of weapons. In an extremely short amount of time we peaked at nuclear weapons, which, if they were actually used regularly, would reach a phase transition: virtually all life on the planet destroyed, with a small collection of humanity transitioning to underground bunkers and underground warfare. And then some new life would emerge on the planet and take over, like a radioactive slime mold or something. So far we have simply decided not to trigger this change, but all someone has to do right now is push a big red button.

So the infinitely increasing innovation leads to a cataclysmic phase transition. I think the point is that there's lots of these things that are moving towards phase transition.

> Very interesting to see how on that scale there is a large gap between the blue whale and the size of the earth.

They could have put in the moon at two orders of magnitude less than Earth, and the Great Pyramid at 4 to 5 orders of magnitude greater than a blue whale.

Maybe should have clarified in text: the "very interesting..." bit doesn't come from the author. It's just something that sounded off to me.

Even adding the Moon or the Great Pyramid would still leave a large gap in between. Which is totally normal, does not suggest anything weird. It just feels like a vacuum (that might be filled with huge spaceships in sci-fi futures:)

Students gaining an operational and transferable understanding of the physical world, is regrettably almost a non-objective for current K-13 science education. To oversimplify, it's poorly incentivized, and might not fit anyway. XR and associated edtech, seemingly has the potential to alter these pedagogical constraints and incentives. So looking ahead...

How might scaling laws be used in education? Now or eventually? Perhaps one possibility might be as part of skill cluster emphasizing rough quantitative reasoning and Fermi questions, as rules of thumb, in conjunction with an order of magnitude feel for reasonable numbers for physical properties. But that's an "eventually". For "now", area-vs-volume length scaling is a familiar part of intro biology. For "near term", maybe as part of intros emphasizing quantitative and physical biology? So what else...?

Any brainstormy thoughts on how scaling laws might be used in education, now or soon or eventually?

I advice everyone interested in ecology to learn about the Club of Rome and the Limits of Growth. This is the true inconvenient truth specially for people in businesses that treat growth as the ultimate objective.

I am quite intrigued by the graph about walking speed in cities. Apparently there are a few cities where people walk... Backwards?

Hehehe true. The graph is from the book, it might be mislabeled. I think it represents the change from average walking speed (which is 1.3 - 1.8 meters / second)

"If the truth can be told so as to be understood, it will be believed..."

I had high hopes for this book but got a niggling sense that the author started with a thesis and then found data to fit it. The author also comes dangerously close to Lysenkoism where he says that "we evolved by interacting and adapting to our environment". An unforgivable sin in my book and sloppy beyond belief.

Then there is the missing data. Aaron Brown's review on amazon [1] puts it best:

"Another chart shows number of heartbeats per lifetime of "animals" versus weight. It looks constant, because the range seems to be about 30 million to 150 million, but the vertical scale runs from 100 to one trillion. "Animals" turns out to be a few selected mammals (whales are listed twice with different values). If you go to the paper, the author emphasizes that the interest is in the deviations from the typical relation shown by the animals on the chart; the chart only shows the typical animals. So far from a universal constant in nature, we find that a subset of mammals happen to have values within a factor of five, with other mammals and non-mammals outside that range, but missing from the chart even though the vertical axis is scaled to accommodate them.

A better chart shows metabolic rate versus weight (here labeled "mass" despite being the same scale as the previous graph). This includes selected mammals and birds, and does illustrate rough linearity in log-log space. But here the main interest is in the slope of the line rather than the linearity. Metabolic rate increases not linearly with mass but at about the 3/4 or perhaps 2/3 power. This is key, because a lot of things also go up with powers of mass and people disagree on which ones of them are important for setting the metabolic rate.

Finally, there is a chart purporting to show that net income and assets of companies are linear in log-log space with number of employees. This is clearly nonsense. Technology companies often have hundreds of thousands of dollars of net income per employee, and few assets, while retailers have an order of magnitude lower profits per employee but much higher assets. There are companies that own and lease things with huge assets and few employees, and service companies that own nothing but a few desks and computers with many employees. It turns out if you read the notes at the back of the book that the 22 points are actually averages of over 30,000 companies (by the way, page, chapter and figure numbers are wrong in the notes, but I assume this will be corrected before publication). So all the chart tells us is when you average over large numbers of companies of different types but similar size, you get similar relations of employees to income and assets as the average for large numbers of companies of a different size."

These would be de-rigueur for a journalist without a science/math background but Geoffrey West is a physicist, albeit a theoretical one.

[1] https://www.amazon.com/gp/customer-reviews/RQT3GP7W8NUFL/ref...

Could you elaborate why the quote you mention is a faux pas. I think from the evolution / morphogenesis perspective he describes how constraints (i.e. gravitational, energetic) shape and direct the space state. Lewontin (also affiliated with Santa Fe) goes a bit further here on that I think: http://www.accuracyingenesis.com/evolution/complications.htm

I read the review and agree with some of what Aaron is saying. But the book is meant to appeal to a general audience. West's research is well represented in peer-reviewed journals, where most of his concerns are properly addressed (example: https://www.pnas.org/content/104/17/7301).

Reply on news.ycombinator.com