Serendip is an independent site partnering with faculty at multiple colleges and universities around the world. Happy exploring!
Week 10 (emergence)
What makes a good emergence model? Think about ones you've made so far, ones we've looked at together, and look at Serendip's collection. Guides you want to follow in your own work?
Remote Ready Biology Learning Activities
Remote Ready Biology Learning Activities has 50 remote-ready activities, which work for either your classroom or remote teaching.
Narrative is determined not by a desire to narrate but by a desire to exchange. (Roland Barthes, S/Z)
What's New? Subscribe to Serendip Studio
A Random Walk
New Topics
-
1 week 15 hours ago
-
1 week 3 days ago
-
1 week 3 days ago
-
1 week 4 days ago
-
1 week 4 days ago
A good model shows something.
I feel that a good model is very much like an essay - it must have a thesis statement, and then prove that thesis statement.
Most of the netlogo modelling that we have done, I feel, has been less about making good models and more about just learning how to make models.
The model that Moser and I worked on for instance, started with an idea for a model. We approached it like a programming problem - "It would be cool to have the turtle do this" and "Oh, and then we could have the patches do this!"
A good model is the opposite of this - "I wonder what would happen if you put a wolf amidst sheep." Then, from there you program the sheep and you program the wolf, using the model to "predict" how they interact.
I guess, to summarize it, a good model should use separate observations we have made to help us form conclusions about something we previously did not understand. It should tell us something.
Personally, I really like
Personally, I really like the "social organizations without a director" model. It is simple and displays something interesting and surprising. It was surprising to me, because I really presumed that there is a director.
So, a good model to me would be a similar one - something that produces unexpected outcomes, because emergence has a lot to do with unexpectedness in my view.
a cyclical development process
Every good model should provide a diagram or illustration of an idea about how something in the physical world functions. While the thing illustrated can be somewhat intangible or even abstract, such as ideas about the emergence of life or randomness, the designer should pick a single, coherent and concise phenomenon from the physical world. The purpose of the model is then to provide a diagram of a mechanism that could be creating the phenomenon.
For example, a designer who is interested in the phenomenon of traffic jams might design a model to illustrate a possible mechanism of why jams emerge. The model would not replace emperical evidence, which would necessitate looking at cars and drivers in the real world, but could serve to test certain theories and assumptions derived from real world observation to see what would happen if they were true. For example you could use an emergent model to examine how focusing on the distance before the next car might affect a driver’s speed. Essentially, this kind of model (which starts with a theory and forms an instance) works in the opposite direction as scientific observation (which starts with an instance in the physical world and forms a theory) and enables a more effective, cyclical development process.
It’s important to remember that the model doesn’t prove the cause of the phenomenon as different causes may have the same effect. In other words, knowing that A could cause B does not prove that A does cause B because B could be caused by C independently of A. However, we never really know what’s going on in nature; rather, we take observations and tell the best story that we can. Emergent models offer a way to test how good that story is.
I think a good model must be
I think a good model must be relatively simple, (something you could explain the logic of in a few sentences) based on a real world example or phenomenon and should accurately reproduce the phenomenon. Langton's ant strikes me as an extremely interesting model, but ultimately one that has little other value. Other models we have looked at, like the traffic model and the flock of birds, have more potential use in explaining real world phenomena. Certainly langton's ant, the game of life, and CA are interesting and potentially very useful, but it is harder to see how these models can be used as more than a sort of "proof of concept" showing that yes, emergence does exist. I think if emergence is to gain acceptance as an important way of looking at and understanding the world we should be modeling things that actually occur and looking for the simple rules that can produce them.
Yes I agree
I agree that it is important for a model to have some application to real world phenomena. But as it has been mentioned during our class discussion, the best way for models to be used in explaining emergent phenomena is to show that very simple models can explain real world phenomena on a surprisingly accurate level, because what underlies a lot of the very complex things we see are actually quite simple. At least in my mind, it's when a model shows us that, that I get that "Wow!" feeling.
Some clarifications
Just to clarify, I didn't mean to say that I think that everything in the world is actually really simple. I just wanted to say that I think good models are those that can provide a different way of thinking about the complex phenomena we see. Like the ant colony model. I really like that one because it demostate the idea that there may be some organization, that can actaully be modelled by a relatively simple model, that underlies what appears to be randomness.
the element of surprise
While I agree with Natsu and Jess that simpler models tend to be clearer (and easier to implement!), I think it goes beyond that. Regardless of the complexity of the functions that go into making it, a "good" model is one that can mislead you into thinking that you know what is going to happen...and then throw you for a loop. The element of surprise is key because this is what makes a viewer want to go back and explore the model (and its idiosyncrasies) further. Langston's Ant works so well because both viewers' predictions for the model AND their self-realizations of their own knowledge are proved to be faulty. Now that's "wow" factor.
good models = clear
I'v really enjoyed looking at the multiple models in the net logo model library so far. Even after seeing so many, I still think my favorite is langton's ant. I think that the reason for why this is becasue it is so simple and clear.Still, after a while it makes you amazed and inspired. It doesn't require a lot of explanation, the model kind of explains itself.
Variables everywhere!
I agree. The models that have a dozen variables, while potentially interesting, tend to scare me off. If I knew exactly what they were trying to express, I might enjoy them more, but as it is I tend to be afraid of touching any of the variables. Also, if a model has too many variables, it's hard to see which one is making which changes. So while I, personally, might not go as simple as langton's ant, I do think that the variables and such should be kept to a minimum.