Latest posts

Mike Jackson wrote "Doing Serious Systems Thinking Means Getting Rid of Elephants, Icebergs and Leverage Points"¹ on LinkedIn, saying that the three metaphors most people learn systems thinking from should be retired, because each of them teaches the wrong lesson about complexity.

Jackson is the originator of critical systems thinking and has been making the case for methodological pluralism for forty years, so this is not a hot take. It is a settled position, stated with some heat, and I think it's excellent, with a caveat on my part².

The spine of the piece comes from C. West Churchman in 1968, talking about an old parable, where blind men describe an elephant in terms of what they can tell about it with their hands. One feels a tusk and describes a spear; another finds the trunk and describes a snake; another man feels the tail and says "this is a rope," none of them describing the elephant as a whole.

This is usually told as a lesson about partial views. Churchman noticed that the interesting character is the narrator, who has quietly awarded himself the only seat from which the whole elephant is visible, and then used it to find the blind men ridiculous.

The parable isn't about blindness. It's about the arrogance of the person telling it. Jackson takes that observation and walks it through the other two metaphors, of the iceberg and the leverage point.

The iceberg puts the observer around the waterline, able to see both above and below the surface, to see the deep structure and act on the true cause.

The leverage point describes someone using a lever to magnify effort to create a response. In this case, it requires someone who stands outside the system, understands its architecture, and knows exactly where to push or lift.

In every case the metaphor supplies a seat outside the situation, and complexity doesn't have one. The person describing the system is in it; it'd be like someone in real space describing the universe as if they were extra-natural, talking about space as an uninvolved observer³.

He has specific complaints beyond that, and they're also good.

The iceberg smuggles in a hierarchy of causes with mental models at the bottom, which implies every problem is cognitive, when causation in complex situations is circular and small events reshape structure.

The elephant is a stable object that can be trained and predicted, which is nothing like a situation whose parts and whole are changing each other as you engage with them.

And his critique of Meadows is the most careful part of the piece: she asks us to transcend paradigms, and her leverage points only make sense inside one, system dynamics, which her bibliography confirms.

His response is a list of where four other traditions would place leverage. That list is also worth the read - don't take my poor summaries as equivalents or replacements for original work!

But with that said, here is what I'd add, and it's an addition, not a correction.

Jackson's argument is that nobody gets the outside seat. I'd go further: there is no outside seat, because there's no outside. The narrator is the same kind of thing as the elephant and the blind men and the zoo the elephant lives in. Martin Buber would say we're all Its to each other, and I'd take that literally. Everything in the situation is engaging with everything else on equal footing, including the part of the situation that's doing the describing. The blind men aren't failed observers of a stable object. They're participants, and so is the elephant, which is touching them back - and to the man describing a tusk, it might as well be a spear even if the totality dwarfs the description.

Once participation is symmetrical, emergence stops being a property some systems have and others don't. Jackson, following Snowden, keeps a line between the complicated, where experts can trace causes, and the complex, where they can't.

I think that line runs through the observer's resolution, not through the world. The world is emergent all the way down, and the places where it looks merely complicated are the places where we happen to be able to see the structure at the scale we're standing in.

What that changes is small in method and large in posture. Jackson's prescription - engage, test alternatives, fail safe, expect emergence - is correct. But it's correct everywhere, not just past the cliff.

For people who draw systems for a living, this pays off directly.

The architect is not outside the system. The architect's model of it is one more component, running on one more participant, with the same partial inputs and the same latency as everything else in the diagram. The diagram is a participant too, because people act on it. And the service you drew is already being used by callers you didn't draw, which means that in a real sense it's using you.

I argued in "Emergent Design, Outside the Codebase" that uses of a system emerge from a designed substrate. The thing I'd add here is that the designer is inside that emergence, not above it, and the only honest diagram is one that has the person drawing it somewhere on the page.

flowchart TB
  subgraph YOU["you are here, in a box of your own"]
    subgraph OBS["the observer"]
      T["Rodin's Thinker"]
    end
    subgraph SYS["the system"]
      A["process"]
    end
    T -. "sees the whole" .-> SYS
  end

Jackson treats the regress as a flaw in other people's thinking. His closing list of four traditions is the parable one floor up, four approaches each holding a lever, with critical systems thinking as the narrator who can see what they're each holding.

I don't say that against him. I say it because his method survives the regress and the metaphors don't, which is the best argument he has, and it's one he didn't make. The regress isn't a mistake. It's where the argument lives. There is no irrelephant.