,
24 tweets,
14 min read
Read on Twitter
THREAD: What is complexity? Concepts as introduced by @EcoLabs + @Ecocene / @cecanexus, with notes especially from the perspective of behavioural medicine. Full poster here: cecan.ac.uk/index.php/news… 
1. Feedback
Note: There are many different terms for two types of feedback loops, e.g.
- Positive vs. negative
- Amplifying vs. stabilizing
- Reinforcing vs. balancing
Take a moment to consider how these play out in your theory of change.
Note: There are many different terms for two types of feedback loops, e.g.
- Positive vs. negative
- Amplifying vs. stabilizing
- Reinforcing vs. balancing
Take a moment to consider how these play out in your theory of change.
2. Emergence
Note: You can't understand a complex system by merely looking at one part.
According to Doyne Farmer, non-linearity is necessary for emergence. I've been trying to look for a piece that explains this simply, let me know if you have one in mind!
Note: You can't understand a complex system by merely looking at one part.
According to Doyne Farmer, non-linearity is necessary for emergence. I've been trying to look for a piece that explains this simply, let me know if you have one in mind!
3. Self-organisation
Note: See Schelling's Segregation Model for a simple example of how surprising trends can arise from tiny preferences.
Note: See Schelling's Segregation Model for a simple example of how surprising trends can arise from tiny preferences.
4. Levers and hubs
Note: For recent work in how hubs and structure affect behaviour, see this:
Note: For recent work in how hubs and structure affect behaviour, see this:
5. Non-linearity
Note: Here's an interesting piece on application in medicine: fooledbyrandomness.com/medicine.pdf. I'd like to internalise it better, so let me know if you want to discuss.
Note: Here's an interesting piece on application in medicine: fooledbyrandomness.com/medicine.pdf. I'd like to internalise it better, so let me know if you want to discuss.
6. Domains of stability
Note: Relatedly, the terrain is known as the fitness or attractor landscape (see wonderful tutorial by @ncasenmare: ncase.me/attractors/). Understanding how it can be single-peaked, rugged or dynamic can improve your thinking.
Note: Relatedly, the terrain is known as the fitness or attractor landscape (see wonderful tutorial by @ncasenmare: ncase.me/attractors/). Understanding how it can be single-peaked, rugged or dynamic can improve your thinking.
7. Adaptation
Note: In social/behavioural interventions, adaptation can lead to unforeseen (iatrogenic) consequences; hence you need to incorporate more than just one simple outcome in the evaluation.
Note: In social/behavioural interventions, adaptation can lead to unforeseen (iatrogenic) consequences; hence you need to incorporate more than just one simple outcome in the evaluation.
8. Path-dependency
Note: Time exists, as does history, and both have implications to what happens next in a system you intervene in. Which histories impede change, which facilitate it? What are the temporal dynamics? (journals.copmadrid.org/jwop/article.p…)
Note: Time exists, as does history, and both have implications to what happens next in a system you intervene in. Which histories impede change, which facilitate it? What are the temporal dynamics? (journals.copmadrid.org/jwop/article.p…)
9. Tipping points
Note: You might want to look into the Cusp Catastrophe model, if you want to start thinking about these in your own data or theories of change.
Note: You might want to look into the Cusp Catastrophe model, if you want to start thinking about these in your own data or theories of change.
10. Change over time
Note: Rather assume instability / non-equilibrium dynamics, and try to find evidence of the contrary – not the other way around.
Note: Rather assume instability / non-equilibrium dynamics, and try to find evidence of the contrary – not the other way around.
11. Open system
Note: Awareness of outside influences is crucial. Think in systems affecting systems, and try to create a systems map of these influences (see medium.com/disruptive-des… by @LeylaAcaroglu for an example).
Note: Awareness of outside influences is crucial. Think in systems affecting systems, and try to create a systems map of these influences (see medium.com/disruptive-des… by @LeylaAcaroglu for an example).
12. Unpredictability
Note: Here's a blog post on the topic: mattiheino.com/2016/12/27/det…. Also see @yaneerbaryam's lecture "The Science of Prediction" for a different take: necsi.edu/events/vidlib/
Note: Here's a blog post on the topic: mattiheino.com/2016/12/27/det…. Also see @yaneerbaryam's lecture "The Science of Prediction" for a different take: necsi.edu/events/vidlib/
13. Unknowns
Note: One way to deal with unknowns is to place yourself in positions where there are large gains and capped losses (edge.org/conversation/n…). I'm interested in considering implications for intervention design, if anyone wants to have that discussion.
Note: One way to deal with unknowns is to place yourself in positions where there are large gains and capped losses (edge.org/conversation/n…). I'm interested in considering implications for intervention design, if anyone wants to have that discussion.
14. Distributed control
Note: This is the rationale for #localism, why hierarchies (and centralised health interventions?) fail under complexity (necsi.edu/complexity-ris…), and @necsi's manifesto of teams as a solution (necsi.edu/teams-a-manife…).
Note: This is the rationale for #localism, why hierarchies (and centralised health interventions?) fail under complexity (necsi.edu/complexity-ris…), and @necsi's manifesto of teams as a solution (necsi.edu/teams-a-manife…).
15. Nested systems
Note: Systems are also nested in timescales (see e.g. researchers.mq.edu.au/en/publication…). What you do daily affects what you (can) do weekly, and the constraints or options which appear on slower (e.g. weekly) timescales affect what's possible on faster ones.
Note: Systems are also nested in timescales (see e.g. researchers.mq.edu.au/en/publication…). What you do daily affects what you (can) do weekly, and the constraints or options which appear on slower (e.g. weekly) timescales affect what's possible on faster ones.
16. Multiple scales and levels
You can't just analyse behaviour on one level. There are ways to deal with this, such as multiscale information theory (sciencetrends.com/describe-compl…), and @SimonDeDeo has a tutorial on @ComplexExplorer (complexityexplorer.org/courses/67-int…).
You can't just analyse behaviour on one level. There are ways to deal with this, such as multiscale information theory (sciencetrends.com/describe-compl…), and @SimonDeDeo has a tutorial on @ComplexExplorer (complexityexplorer.org/courses/67-int…).
The end.
Well, not really. There's no end in complexity, only new dynamics, more or less stable, replacing old ones. The system you intervene in doesn't stop when you manage to hit a change target. Have fun and be responsible!
Well, not really. There's no end in complexity, only new dynamics, more or less stable, replacing old ones. The system you intervene in doesn't stop when you manage to hit a change target. Have fun and be responsible!
Note, again, that pictures are not my work, but that of @EcoLabs, @Ecocene, @cecanexus, @DrAlexPenn, @marthabicket, @bapeterj, and @DioneH_TIHR.
Afterthought: Often the concepts get thrown around without much regard to their implications to practical quantitative modeling. I have an entry-level type of paper coming up (watch this space), but would love to hear of your favourite ones!
Find more distilled takes on #ComplexityScience concepts at #ComplexityExplained. Includes interactive explorables!
Complexity: "The essence of a thing is not in the stuff the thing is made of, but in the patterns of organisation that thing embodies"
- @normonics
- @normonics
> "All of the properties in the world we're interested in, all the phenomena, are not of some substance that thing is made from, but from how the various substances interact and self-organise into patterns, and those patterns are what the actual "thing" were interested in is... >
> And it's very difficult, because it's natural for us to think in substance and in terms of "what is this thing, what's it made of?"
But what we find more and more is, that what a thing is made of, is at least secondary to what is the organisation of that stuff it's made of."
But what we find more and more is, that what a thing is made of, is at least secondary to what is the organisation of that stuff it's made of."
