Gossamer networks

I’m reading paper trails, which is a fascinating history of the post office, as studied geographically from maps of post office locations, as both a dispersed organisation and a key component of colonisation of the western US, and the associated genocide of native American people.

There’s a lot of other important stuff in there, but the bit that’s knocking around in my head is how the office functioned as a network with minimal central control, and with quick addition and removal of nodes- a Gossamer network in the terms of the book.

There was no 100% true contemporary map of the postal network which could be viewed in Washington. New post offices were technically commissioned centrally- by congressional order- and Technically appointed centrally(all postmasters were appointed by the postmaster general, giving him the power to personally appoint friends and allies to tens of thousands of positions across the country, making him one of the most powerful political figures around.

But in reality, these were rubber stamps and technicalities, and most of the work of adjusting the network was done outside of washington. A petition comes in to their congressman from a new town for a post office, which he rubber stamps, and the central office passes on an edict to add a node to the network to the local large city. A postmaster is appointed by someone closer to the edge, and the requests are basically never rejected.

Thousands of post offices opened and closed a year, and it took at least a week to get a letter from a rural office to the central one, which is effectively an impossible latency for something which is so nominally centrally controlled. But still it functioned almost perfectly- it coordinated troop movements and provided the essential backbone for a social sphere among the white settler population which was frequently moving hundreds of miles every couple of years, a level of mobility not really seen even among renters today.

Nodes on the network had incredibly cheap startup costs which made this possible- if they were already on a contracted mail route, it was simply picking a person, swearing them an oath and giving them a book of stamps to sell. So they opened a lot and they closed down a lot. The larger mail towns knew which offices were closed or opened and they passed it along, but again the latency meant this was not known or ordered from a central point.

The colonisation of the western US would not have happened at the rate it did without a system that reacted locally- managing it centrally would have been impossible with that latency, and subject to the normal problems of central planning.

I think there’s a couple of key things here which made the network work:

It had a simple, binary, success or failure condition. You can have a lot of corruption and inefficiencies in the system as long as the people who’re doing the corruption are also passing along packages- You either pass it along or you don’t. And while this gives a lot of power to a small town postmaster, but the postmaster is also highly interdependent on the other settlers around him. They can’t easily decide to go down the road to the next store- but if they do it ruins him.
Plentiful resources. The network had immense capital(from the economic vortex that was the NYC banking sector, running on colonialist extraction), and “wasted” a lot of it in comparison to a contemporary mail network. But more important were the capital surrounding the network- Most postmasters were only deriving a small fraction of their income from the post(10-20% afaict) and most from other endeavours(running a store or hotel seems common). In a space with plentiful capital there’s a lot of people who can take on a job like that.

And the advantages of the network were massive(from the perspective of colonisation- which needs a reliable and convenient communications system to encourage and understand colonisation of the west):

The network essentially reconfigured itself at the edges. Technically congress asked the postmaster general to open a new office at a location and the postmaster general chose a postmaster and the postmaster opened the office, but in reality most of that work was incidental to the actual opening of the office- the work of deciding where to open the office was delegated to anyone who could get a congressmans ear to have their staff write a short letter- basically anyone with a population the network needed to serve. New town? It had a post office in weeks or months.
It was cheap to open and close new nodes for the central bureaucracy- You can make lots of bets on new towns, and if they get dysentery, well, you lost some stamps.
Highly fault resistant. Humans in the loop gave it the vaunted property of the internet, and it routed around damage. If a post office closed and a letter had been posted it it, the letter can be passed on to the next nearest post office.
It was a poor tool of central control due to it’s unknowability and distributed decision making. When settlers moved on to reservations despite supposed federal bans(even sometimes enforced federal bans, where it was opportune for the state) the post office quickly opened new offices to serve them- They barley had most of the offices on a map, and new offices essentially opened with local permission.

Tl;Dr: The post office as was a reactive system with broadly distributed decision making which enabled the colonisation of the American west and the genocide of native American people, in contravention to narratives of a totalising state.

For all James Scott would have us ascribe the ills of colonisation and genocide to central sight, the post office did this without being able to map their territory or maintain an up to date realistic list of central post offices. Colonisation wasn’t a regimented state project, but a case of setting up incentives for the systems on the ground to enact themselves.

There’s lots of other examples of systems with distributed decision making doing evil things, and I think we need to be suspicious of the “because it’s decentralised it’s good” example.

Al-Qaeda is essentially a gossamer network- Cell structures lend themselves to the model, and there is a degree of central control, as with the US post office, but it is mostly locals with grievances against US imperialism self organising, and maybe receiving a bit of(cheap) training in a mountainous region. When a group fades or is disrupted it doesn’t matter- the network routes around it.

Not all cell structures are gossamer networks- a more traditional state actor has much more investment in many of their nodes(a several clean faces for an overseas operation, along with equipment, etc is far more expensive than a training camp in the highlands) but operates them in isolation- so they don’t behave like a gossamer network. But combine those low startup costs and easy affiliation with a cel structure you get a network that behaves like the postal office(to do insurgency and/or psychological warfare we call terrorism not deliver mail).

So are many fash networks too(join this telegram, buy this sticker printer, print these stickers, onboarding into the white rose completed), and I think there’s an argument that the mutual aid organisations can be considered as such(make this fb group, post these leaflets, do these simple tasks- you’re onboarded. Pass along referrals to the nearest group. )

I think gossamer networks, as defined by low startup costs for nodes, high interoperability, simple operations producing complex effects, and low levels of central knowledge are an interesting opportunity for building effective systems on a large scale which don’t have central decision making. Which is essentially the anarchist project. And also a strong warning- The colonisation of the US and genocide of Native American people would have been impossible without an organisation operating with far distributed decision making, much in our own models of organising. Distribution alone is not enough.

I’m interested in other examples of gossamer networks people can think of. Email me