Frankenstacks and Rhizomes
Like a lot of people who were lucky enough to snag some genesis ether during the Ethereum presale in 2014, I found myself scrambling last week to find my presale wallet and password, and figuring out the technology to move the ether to an exchange. If that sentence made no sense to you, you have some googling and crawling-out-from-under-a-rock to do.
There were some anxious moments. I'd forgotten that the wallet file was encrypted, so I had to dig through my email to find my password hint. Which turned out to be not enough of a hint, leading to a few minutes of frenzied guesswork. Then there were more anxious moments when my local install of Mist (the standard wallet) kept crashing when I tried to import my presale wallet -- after 12 hours of downloading the blockchain. Turned out to be a bug and I was able to use a different wallet to access my ether. Then the exchange crashed for a while. But finally I was able to make transactions and swim around in my precioussss liquid $200+ ether for a bit like Scrooge Gollum McDuck. It was a pretty modest windfall, and a quarter of the paper value has already evaporated today as I write this (I plan to hodl on to as much as I can afford to, for as long as I can), but it's been a crazy roller-coaster for the last couple of weeks. And I fully expect it to get crazier.
But the experience of suddenly being dumped into a messy, high-stakes emerging IT ecosystem got me reflecting on a theme I keep returning to lately: the sheer complexity of the IT stack we rely on today as individuals. Today, each of us lives within what I call a frankenstack. An assemblage of information technologies duct-taped together with a mess of protocols, and forming what philosophers call a rhizomatic structure.
My rhizomatic frankenstack
1/ Consider the difference between an onion and a piece of ginger. The ginger root is the motif for what philosophers call a rhizome. The onion for what they call an arborescence.
2/ With an onion, you can always tell which way is up, and can distinguish horizontal sections apart from vertical sections easily by visual inspection.
2/ With a piece of ginger, there is no clear absolute orientation around an up, and no clear distinction between horizontal and vertical.
3/ According to the linked Wikipedia entry (worth reading), a rhizome "allows for multiple, non-hierarchical entry and exit points in data representation and interpretation."
4/ If you tend to use the cliched "hierarchies versus networks" metaphor for talking about old versus new IT, you would do well to shift to the rhizomatic/arborescent distinction.
5/ Both onions and ginger roots show some signs of both hierarchical structure and network-like internal connections.
6/ The difference is that one has no default orientation, direction of change, or defining internal symmetries. Rhizomes are disorderly and messy in architectural terms.
7/ The diagram above shows a partial view of my personal frankenstack: a mess sprawling over wordpress, slack, mailchimp and dozens of other technology platforms.
8/ As a free agent solopreneur with a weird mix of activities, my frankenstack is probably more complex than most, but not as complex as some power-users I know.
9/ If you work in a large organization defined by an enterprise IT system, your frankenstack is likely more arborescent than mine. More onion-like.
10/ But this is not going to last much longer. Already, bleeding edge enterprise IT platform architecture is acquiring the rhizomatic characteristics of the consumer web.
11/ Why is the rhizome a better mental model for IT infrastructure than either hierarchies or networks? The answer has to do with the curse of dimensionality.
12/ All of us today live informationally high-dimensional lives. We manage many complex information stocks and flows that merge and mix in a labyrinthine permissions/security matrix.
13/ Hierarchies and networks are both clean, legible architectural patterns. Applying them to high dimensional situations is highly budrensome and largely useless.
14/ Consider an organization with a strictly hierarchical org chart. You could model it with a single variable: who reports to whom. "Level" is a dependent variable.
15/ Or consider an organization that's a strict network topology. You could model it with a graph: who is connected to whom. "Degrees of separation" is a dependent variable.
16/ Now start to throw in complicating factors. In orgs, you could have dotted-line relationships, floating assignments, full-time/part-time boundaries, staff vs. line etc.
17/ In a network, you could have different complicating factors: asymmetry vs symmetry in follows, algorithmically maintained feeds that drive interactions, and so forth.
18/ With each complicating factor, more new variables enter the picture. The dimensionality increases. However, not all dimensions are equally important.
19/ So you end up with a mess of organizing constructs: hierarchies, networks, group boundaries, permission levels, version histories, event histories, and so on.
20/ If you tried to extend a platonic concept like "hierarchy" or "network" you'd end up with impossibly high-dimensional structures that are empty for the most part.
21/ Instead, you switch to only modeling things that have information content, and doing so piecemeal. You end up with "crumpled" versions of high-dimensional structures in low-dimensional spaces.
22/ In my rhizome picture above for example, there are a dozen dimensions (the words in green). But I have a loose, gingery representation in 3d space.
23/ Things are related as they are pragmatically required to be, and interpenetrate and relatively orient as is most convenient, rather than to conform to a platonic pattern.
24/ The archetypal action in a rhizomatic information architecture is cut-and-paste. The spreadsheet is the archetypal integration tool: a sort of generalized clipboard.
25/ There is a relationship here to the idea that the medium is the message, and Conway's law (product structure mirrors org structure).
26/ Our information environments are becoming rhizomatic because our informational lives are becoming rhizomatic, and vice versa, in a chicken-and-egg loop.
27/ Tools with rhizomatic dispositions include, besides cut-and-paste and spreadsheets, things like IFTTT, Zapier, and at the enterprise level, things like microservices.
28/ Rhizomes aren't just about information. They are also about computational capabilities, distribution capabilities, relationships, trust, and permission architectures.
29/ Two of the biggest technologies evolving today -- the blockchain and machine learning -- are fundamentally rhizomatic in their DNA. You have to tread gingerly around them.
30/ Both have the predisposition to follow the contours of information content rather than idealized organization patterns. Neither is architecturally well-behaved and disciplined.
31/ Rhizomes are information dense, topologically complex (things connecting in weird ways), highly heterogeneous and variegated, structurally compact, and with strong form-content coupling.
32/ Unlike pure-paradigm architectures, rhizomes mix and match multiple architectural paradigms along with emergent structures to create high-dissonance information environments.
33/ What is it like to live in a rhizome? Well for starters, there are no default entry or exit points, no "onboarding manual" that teaches you how to survive in one, and no "up."
34/ There is perhaps a distinction between a n00b and an expert, but it is highly localized around specific corners of the rhizome. You can go from n00b to expert and back to n00b in 2 steps.
35/ In a traditional org, you can count the floors between the executive suite and say the shop floor where blue-collar workers build products on assembly lines. Authority falls as the elevator descends.
36/ n00b/expert relationships change slowly and predictably in space as you move. Expertise and authority turfs are simply connected and simply bounded.
37/ In a rhizome, in a move from point A to point B, relative knowledge and expertise might swing wildly. And the value of actions might swing wildly while you're moving.
38/ A rhizome is also a high-friction space. Movement through a rhizome involves an unpredictable stream of transaction costs. Every journey is an obstacle course.
39/ Sometimes there's a good FAQ page, other times a tweet makes a difference between a minute and a week. Discovered structure, rather than inferences from maps, dictates the cost of action.
40/ Sometimes having a programmer friend whom you can quickly email can make the impossible possible. Sometimes an online forum saves hours, sometimes it wastes hours.
41/ Sometimes a single click moves mountains. Other times, you need to move mountains to do one tiny thing. Effort-outcome relationships get out of whack.
42/ Sometimes really important things are trivially easy if you happen to know somebody who knows one weird trick. Other times, common-sense things turn out to be impossible.
43/ In a rhizomatic world, if your expectations and work habits are built around architectural cleanliness, you will get deeply frustrated and be perennially frozen.
44/ If you can only navigate well-paved paths and clean, well-lit spaces, you'll likely spend a lot of time in low-value, or even futile, ritualized behaviors while getting nothing done.
45/ You must be willing to adopt an opportunistic approach to navigating complexity, and switch from ugly hack to elegant beauty, from amateurish fumble to expert flourish, in an instant.
46/ You've heard of analysis paralysis, right? I have a similar concept I call aesthetic paralysis: the desire for elegance in behavior limiting your agency. Superficial beauty is expensive in a rhizome.
47/ This is not to say rhizomes are ugly. Once you accept their inevitability, the act of navigating a rhizome can start to acquire its own beauty.
48/ You start to be less attached to received ideas of importance, order, and logic, and learn to interact with the natural logic of the environment.
49/ You learn to adjust your aesthetic sensibilities to what you're experiencing, so you can actually see what's going on, rather than being bound by aesthetic expectations.
50/ You start to gain what a friend of mine just described as "infrastructure fluency." This is in a way the opposite of architectural taste: an ability to experience the artificial world in its natural language.
51/ Unlike in physical architecture, where there is such thing as an architect's view of reality, the only guide to a true rhizome is a burglar's guide. So stop worrying, and learn to love your frankenstack.