# Blog Archives

## Fixing broken government — Philip K. Howard at the Long Now

Philip K. Howard has been active in public affairs his entire adult life and has advised national political leaders on legal and regulatory reform for fifteen years. Philip writes periodically for the Wall Street Journal, the Washington Post, and the New York Times

Philip K. Howard gives his diagnosis and prescription for fixing government in the US at the Long Now.

People feel powerless in the face of government. The reason is simple: they are.

It isn’t just citizens. Govnt employees are equally shackled. We as a nation have become strangled in a dense and growing jungle of regulatory constraints.

The root is that we have changed the nature of law. Instead of laying down timeless general principles it now specifies exactly what actions are or are not allowed. Individual discretion has been removed from the system. Worse, the laws don’t sunset, they just continue to pile up on top of each other. Simple example: OSHA regs specify that a hammer used in a business must meet certain standards which may make sense for a construction site, but not for a guy in an office who puts up a picture once a year (and got fined for not having the right kind of hammer).

The solution is not the end of regulation. Regulation is needed to manage our increasingly complex society. Government is necessary to safeguard the common good, especially in an increasingly interdependent yet anonymous society.

The needed change is to once again allow human judgment to determine appropriate action.

The proposed change is threefold, which each change requiring the sacrifice of a sacred cow.

1. Spring cleaning of the law, or at a minimum any law with budgetary implications. Force legislators to review all existing law, removing that which no longer makes sense, and revising that which does according to principles 2 and 3, below.
2. Simplification. Law must be comprehensible to laypersons, not just specialists. The purpose of law is to set goals and objectives, not to specify exact procedures. For example, a workplace safety law could require just that equipment be appropriate to the job and in concordance with industry norms.
3. Accountability. The counterpart to relying on human judgment is to make the person who makes the call accountable. If they consistently make bad judgment, they loose their job. If they make really bad judgments, they go to jail.

This is difficult. Special interests will fight to keep regulations complex and written as they are. Accountability requires trust, and neither political party is willing to extend this to the other.

But such changes have happened before. The progressive era saw the end of laizer faire capitalism as people realized that business left to itself quickly becomes exploitative (child labor, etc). The New Deal saw the creation of a safety net because our farmers were starving due to forces well outside of their control. The rights movement resulted in sweeping changes to society.

Normally these movements run for a long time, building up momentum, but with little “outward” signs of progress. Then, when they are ripe, they spring up “overnight” with incredible force. The change happens quickly.

He also notes that major legal reforms, such as he proposes, are universally followed by massive increases in economic growth.

One place to start would be reforming public schools. Give power to the principles and teachers to do what they think is best in a given situation.

## Why the West rules for now — Ian Morris at the Long Now

Geography shapes history, history shapes geography.

Geography determines which societies will do well, yet culture determines what aspects of geography are relevant. Shifts in what is relevant make the fortunes of cultures.

Agriculture started where it was easy. But the development of irrigation allowed river societies to take off. Rivers then become trade routes.

Then the Med, then ocean going ships conquered the Atlantic, etc.

Talk is here.

Supported by Niall Ferguson (The Ascent of Money) and Jared Diamond (Guns, Germs, and Steel).

## Matt Ridley — deep optimism at the Long Now

Matt Ridley explains why he is such an optimist at the Long Now.

Humanity’s lot has been improving for the last 10,000 years and he sees no reason this will stop, despite everyone’s claim that the world/civilization is going to heck in a handbasket.

He was a pessimist when he graduated college. He marks his conversion to the day he realized acid rain was not a problem.

Trade as the key thing which separates man from beast, and which powers all of our cultural growth.

His greatest concern is that religious fundamentalism will shut down trade and innovation. He is an athiest, though raised as an anglican. He considers this the mildest form of the virus, practically a vaccine.

## Structural preferential attachment: network organization beyond the link (balancing order and chaos)

Preferential attachment is ubiquitous. It predicts power-law distributions where the probability of an event of size k decreases as an inverse power of k: $P_k \propto k^{-\gamma}$.

Preferential attachment can create either order, i.e. distinct growing structures (here, Simon’s balls-and-bins, more generally, a Dirichlet process model) or chaos, i.e. random networks (here BA).

Preferential attachment can form order (balls in bins) or chaos (random network). The real world falls in between (community structure).

Girvan and Newman showed that most real world networks have community structure. (PNAS, 2002, non-paywall here). They mix order and chaos.

This work presents the Structural Preferential Attachment (SPA) algorithm for creating networks. The algorithm is as follows:

At every time step, a node joins a community. The node is either new with probability q or existing. Similarly, the community is either new with probability p or existing. Choice between existing nodes is with probability proportional to its membership number. Between existing communities is proportional to their size.

The model allows closed-form solutions for the membership, community size, and community degree distributions. Further, fitting the model to real-world data shows that it accurately re-creates their structure.

Builds network connections by growing communities shows that scale-free properties observed at the node level are inherited from preferential attachment at the level of the community.

Abstract:

We introduce a mechanism which models the emergence of the universal properties of complex networks, such as scale independence, modularity and self-similarity, and uniﬁes them under a scale-free organization beyond the link. This brings a new perspective on network organization where communities, instead of links, are the fundamental building blocks of complex systems. We show how our simple model can reproduce social and information networks by predicting their community structure and more importantly, how their nodes or communities are interconnected, often in a self-similar manner.

## All scale free networks are sparse

Scale free networks have a degree distribution which follows a power law, $P(k) \sim k^{-\gamma}$ This letter explains why $\leq < \gamma \leq 2$ cannot occur (for large networks).

Note that $\gamma <0$ implies a very dense network (and hence not really “scale free” in the traditional sense), and $\gamma >2$ implies a sparse network.

The core of their argument comes from looking at the scaling of the largest and the lowest degree nodes. For $\gamma \leq 2$, the number of nodes with the largest and second largest degree is $\O(N)$, while for $\gamma >2$, the number of such nodes grows sublinearly with N. Also, the number of nodes with degree 1 (or order 1) grows linearly with N. Thus, if $\gamma \leq 2$, one needs $\O(N)$ degree one nodes to associate with the highest degree node, leaving no way to place edges for the second highest degree node.

Note these results are asymptotic. One could have a finite network with $0 \leq \gamma \leq 2$. To grow such networks, however, requires either a cutoff on the allowed node degree, or that $\gamma$ must increase over 2.

http://prl.aps.org/abstract/PRL/v107/i17/e178701

Charo I. Del Genio, Thilo Gross, and Kevin E. Bassler
Phys. Rev. Lett. 107, 178701 (2011)
All Scale-Free Networks Are Sparse

## Measuring the dimension of a spatially embedded network

From Nat Phys, feb 2011

The authors propose measuring the dimension of a spatially embedded network by:

1. measuring
• the number of nodes and
• the mean euclidean distance

of all nodes at (graph) distance l, for all l, from a given start node

2. averaging these quantities over a large number of starting nodes
3. fitting the average to the equation M ~ r^d

where M is the mass (number of nodes), r is the mean distance, and d is the dimension

Background & fun facts: Previous work has only considered lattices or random networks (all nodes have equal prob of being connected). In a lattice,

• dimension == embedding space.
• mean distance between two nodes scales as N^(1/d)

In a random network,

• dimension is infinite
• mean distance scales at log N or log log N

Observed real-world networks:

• mobile phone: prob of a friend at distance r scales as P(r) ~ r^-2
• global airline: prob of a direct flight to an airport of distance r scales as P(r) ~ r^-3

The authors claim that dimension is related to diffusion (the probability that a particle returns to its start point after t steps), and percolation (the critical fraction of nodes which can be removed before the network dissolves)

The paper has many formatting errors, both in the text and the main figure.

Dimension of spatially embedded networks
Daqing, Li; Kosmidis, Kosmas; Bunde, Armin; Havlin, Shlomo
Nature Physics, Volume 7, Issue 6, pp. 481-484 (2011).