Hierarchy, Networks, Generals and Clouds of Purpose
At the Futures Forum, the event we ran in May, Anton Andrews talked a bit about how it was the new cool thing to go from hierarchies to networks. I was pretty taken by this idea1 so I did a bit of digging. It gets credited to General Stan McChrystal, but it’s got a long history before that!
Tl;dr. There are two main ways of running organisations: hierarchies and networks. Hierarchies are brittle and aren’t conducive to good communication. Networks are more robust and are very good at communication! If you want it even shorter, hierarchies are uncool, networks are cool.
If you click the button this will show a bit of information trying to get from bottom right to bottom left. It’s slow and there are any number of ways that it could go wrong!
Hierarchies feel like an obvious way to control a big group. Kings and Queens, controlling their Lords and Dukes, who then instruct their knights… and so on until the peasants and the slaves. The problem is that hierarchies are really slow to move information around. If we imagine it with the king at the bottom, and the peasants at the top, then it’s just like a tree. That’ll give us lots of easy metaphorical hooks. Moving information from one leaf to another can take a really long time. Sometimes the information needs to go all the way down/up2 to the king before it can go back out to another leaf! This might be OK, but what if one of the lords is on holiday, or the messenger gets waylaid in a brothel? The message passing falls down because there is no redundancy or robustness in the system. What if it’s bad news and someone is too scared to tell their superior? This piece is going to try to explain that networks are better for a whole bunch of reasons.
The story goes that the US army needed to find a way to beat al-Qaeda. The mantra It takes a network to defeat a network.
became something of a driving force in Afghanistan. To give this a bit of colour, here’s Tim Ferriss talking to Stan McChrystal about how it ends up working:
If you think about it, if you and I were playing a game of chess right now and we each had our 16 chess pieces. I moved, you moved, I’m against you and we’re both micromanaging our teams. What if for the dynamics of shared consciousness all of my chess pieces could decide on their own, and they could communicate amongst themselves. So, in fact you aren’t really playing me you’re playing the combined intelligence and flexibility of my 16 chess pieces. They can move whenever they want, they don’t have to wait for you to move.
Suddenly you say well wait a minute that’s not fair, but that’s the environment we’re in now, you’re not against one iconic decision maker you’re against this networked set of competitors. Maybe they’re in coordination intentionally, maybe they’re unintentionally coordinating. So, what I would argue is chess actually may reinforce a more mechanical structured game than the world allows right now.
Stan McChrystal speaking on The Tim Ferriss Show: General Stan McChrystal on Eating One Meal Per Day, Special Ops, and Mental Toughness, starting at about 1:38:00
What Stan’s saying here is that the power of the network is that each of his chess pieces is empowered to make choices. Each node (person) in the network is imbued with agency! That is, they have explicit permission to make choices. This is very cool because instead of mechanically capturing the bridge, if it’s pointless because the bridge isn’t useful any more they can pick a different course of action. I’m getting ahead of myself, let’s cover the basic stuff first about why networks are good, then get onto the cool stuff.
Why networks are cool
People don’t really like hierarchies because, unless you are the king, it feels a bit like the man is keeping you down. This is mainly an excuse to get a video into this; it’s not really about you, it’s about outcomes. If the outcomes are good, then there’s a pretty good chance it’ll be good for you too.
The big thing that networks give you automatically is robustness. (Some people call it self healing.) If one part of the network is broken then information can still flow around it. Information also diffuses in a network. That means that as soon as some information is made or discovered then it starts being useful to people. The number of people that it’s useful to just keeps growing as more people know. (If you are feeling nerdy, it’s a geometric growth, exponential if you must!) That makes it hard to control information if you are a secretive type, but being secretive is the enemy of being successful in the modern world!3
Networks don’t automatically get information to the right person straight away. Diffusion isn’t a directed phenomenon. When you let them be, humans are smart. They solve that problem by sending information straight to the person who needs it. This doesn’t work in strict hierarchies because of chain of command issues.
What purpose does for networks
Networks can get information from anywhere to anywhere through the shortest route. Often the shortest route is just to skip the network as it’s drawn here and just go straight to the end point. If you aren’t the end point, but the information is useful to you, then it’s that bit closer to you!
In huge networks that can get a bit hectic. It’s like being expected to keep up to date with everything going on in the world with nobody there to tell you. So big networks have developed strategies to deal with it. This feels so obvious that it’s going to sound stupid: journalists do this job for big civilian networks (towns, countries, etc.) The global intelligence community uses something called fusion centres4. The job of these fusion centres and news outlets is to make the information in the network accessible5.
This is starting to sound a bit chaotic! How are we supposed to know that everyone knows what’s going on? This is where purpose comes in. The point of the purpose is to give each of the nodes in the network a direction. It’s like a magnetic field that makes migrating animals go in the right direction. This is an important thing! If there’s no sense of purpose then as soon as the explicit commands break down everyone panics!
[N]o plan of operations extends with any certainty beyond the first contact with the main hostile force
Moltke on the Art of War, Helmuth von Moltke, 1859-70
Everyone has been giving General McChrystal a lot of credit for these sorts of ideas. He did an excellent job of implementing them, but as far as I can tell the credit needs to go to Helmuth von Moltke the elder.
Von Moltke was a big deal in the Franco Prussian War. His quote above has morphed into: “No plan survives first contact with the enemy” which is a bit snappier. In the intro to my version of Moltke on the Art of War the editor says:
New guidelines established two points that became the hallmarks of Moltke’s ideas on command. Commanders should order as little as possible (leaving details to subordinate commanders) and they should take care to limit their orders to what was practicable.
Daniel Hughes, Moltke on the Art of War
Auftragstaktik blended strategic coherence and decentralized decision making with a simple principle: commanders were to tell subordinates what their goal is but not how to achieve it.
Superforecasters, Philip Tetlock talking about von Moltke
What he’s saying is that the further from the action you are, the more abstract your directions should be. It’s a fancy way of saying don’t micro-manage people. Hughes goes on to say Moltke was willing to allow subordinate commanders to deviate from the details of his directives so long as their actions were consistent with his overall concept.
. I’d say that “consistent with his overall concept” is in line with what I see “consistent with the stated purpose” doing for any other organisation. It’s pushing agency into the network. Diffusing decision making to the people best equipped to make those choices. There is a much better treatment of this in Superforcasters by Philip Tetlock. He says that when military people consult for business the businesses struggle with their preconceptions. Getting them to abandon their command-and-control hierarchies is hard. Strange that the military has been doing for over a century now!
A modern example
The Agile Fluency Model is a way of classifying agile teams on how badass they are. I saw a nice talk about this at Agile Australia this year. A level 4 team isn’t just getting their work done, they are looking at the market and pre-empting what features to build next to capture the most value. You can’t micro manage a team like that, you just need to point them in the right direction and keep obstacles out of their way. If my memory serves me right they were saying that a lot of the teams at Spotify were level 4.
Conclusion
This is all pretty in depth, I’m amazed that you’ve made it this far! Perhaps I’ve convinced you that having a clear purpose is useful. Not just as a marketing tool, but as a force that aligns all the nodes in your network.
-
It put into words an idea that I’ve had for a while. I wonder if there’s a word for that? ↩
-
I’ve talked at length about this, but the headlines/sound-bytes are: “If you don’t trust the people you work with then you are working in the wrong place”, “That kind of behaviour went out with the Gestapo”, etc. I think that by making information more freely available within a company, better decisions will be made–no caveats. If you think that there is anything gained from hiding information then I’d love to hear the arguments, maybe I’ll change my mind! ↩
-
depending on how you are holding your mental tree ↩
-
Stan’s chief of staff, Chris Fussell wrote a thesis about it: What Makes Fusion Cells Effective? (PDF) ↩
-
In the case of fusion centres, they will also try to get it to the right person. Something that journalists aren’t yet able to do; I bet they will soon though! ↩