Saturday, February 20, 2010

Seeing like a slime mold

Land reforms instituted in Vietnam under French rule, in India under the British, and in rural czarist Russia introduced simplified rights of ownership and standardized measurements of size and shape that were primarily for the benefit of the state, e.g., for tax purposes. James C. Scott’s Seeing as a State: How Certain Schemes to Improve the Human Condition Have Failed gives these and numerous other examples of ways in which standardization and simplification have been used by the state to make legible and control resources (and people) within its borders. He recounts cases of how the imposition of such standardization often fails or at least has unintended negative consequences, such as his example of German scientific forestry’s introduction of a monoculture of Norway spruce or Scotch pine designed to maximize lumber production, but which led to die-offs a century later. (The monoculture problem of reduced resilience/increased vulnerability is one which has been recognized in an information security context, as well, e.g., in Dan Geer et al.'s paper on Microsoft monoculture that got him fired from @stake and his more recent work.)

Scott’s examples of state-imposed uniformity should not, however, be misconstrued to infer that any case of uniformity is state-imposed, or that such regularities, even if state-imposed, don't have underlying natural constraints. Formalized institutions of property registration and title have appeared in the crevices between states, for example in the squatter community of Kowloon Walled City that existed from 1947-1993 on a piece of the Kowloon peninsula that was claimed by both China and Britain, yet governed by neither. While the institutions of Kowloon Walled City may have been patterned after those familiar to its residents from the outside world, they were internally imposed rather than by a state.

Patterns of highway network design present another apparent counterexample. Scott discusses the design of highways around Paris as being designed by the state to intentionally route traffic through Paris, as well as to allow for military and law enforcement activity within the city in order to put down insurrections. But motorway patterns in the UK appear to have a more organic structure, as a recent experiment with slime molds oddly confirmed. Two researchers at the University of West of England constructed a map of the UK out of agar, putting clumps of oat flakes at the locations of the nine most populous cities. They then introduced a slime mold colony to the mix, and in many cases it extruded tendrils to feed on the oat flakes creating patterns which aligned with the existing motorway design, with some variations. A similar experiment with a map of cities around Tokyo duplicated the Tokyo railway network, slime-mold style. The similarity between transportation networks and evolved biological systems for transporting blood and sap may simply be because they are efficient and resilient solutions.

These examples, while not refuting Scott’s point about frequent failures in top-down imposition of order, suggest that it may be possible for states to achieve success in certain projects by facilitating bottom-up development of ordered structures. The state often imposes an order that has already been developed via some other means--e.g., electrical standards were set up by industry bodies before being codified, IETF standards for IP which don't have the force of law yet are globally implemented. In other cases, states may ratify an emerging order by, e.g., preempting a diversity of state rules with a set that have been demonstrated to be successful, though that runs the risk of turning into a case like Scott describes, if there are local reasons for the diversity.

[A slightly different version of the above was written as a comment on the first two chapters of Scott's book for my Human and Social Dimensions of Science and Technology core seminar. I've ordered a copy of the book since I found the first two chapters to be both lucidly written and extremely interesting. Thanks to Gretchen G. for her comments that I've used to improve (I hope) the above.]

UPDATE (April 25, 2010): Nature 407:470 features "Intelligence: Maze-solving by an amoeboid organism."


Alex said...

The slime moulds didn't really "duplicate" the real Tokyo maps, and reading this paper, neither did the UK maps.

In order to compare them, they first looked at the maps, and ignore where the real railway lines/motorways were, instead they just drew straight lines between each node - this makes what mathematicians call a graph, not to be confused with a chart, which is what everyone else thinks of.

They after each slime mould run, they would do the same. Then then would compare the two graphs.

The details of each map could vary considerably, with the real world having obstacles to avoid like rivers and mountains, and the slime moulds having their own variations.

However, even though they threw out all the details, and concentrated on the links, there wasn't really a good match. In the UK example, the best they could do was the real life graph was a sub-graph of the slime mould.

Now, back to the 'how unlikely is this'.

If you have a limited number of nodes, and a small number of connections between them, then you are going to get a graph which looks kinda like a road or rail network. The only other possibilities are either really unlikely (like a circular route where every node has links to exactly 2 other nodes), or eliminated by the rules, such as linking every node to every other node.

It's really just another example of pareidolia.

Jim Lippard said...

Alex: Did you look at the published articles? I only read the second-hand reports that I linked to. From your description, it sounds like not very good science--are they really just making subjective visual comparisons?

One of the articles suggested they planned to add topographic features, which would add some more real-life constraint to the model.

Given your descriptions, the slime mold cases gave me a good title but not good evidence. The sorts of claims Jared Diamond makes about geographical features in _Guns, Germs, and Steel_ would be a similar kind of evidence--some historians find his work to inappropriately attribute "agency" to non-humans, but that seems to me to be a mistaken notion of the term "agency."

KingPong said...
This comment has been removed by a blog administrator.
Jim Lippard said...

The previous comment has been reposted on the relevant blog post (from 2007).

Alex said...

The UK paper is availble at, for free.

The Japanese paper is at the web page for Science magazine, but it's per pay. However, the key sentence from the abstract says it all

We show that the slime mold Physarum polycephalum forms networks with comparable efficiency, fault tolerance, and cost to those of real-world infrastructure networks—in this case, the Tokyo rail system

Comparable efficiency is not the same as identical.