Month: August 2008

It’s been a while since I fin­ished read­ing Steven Berlin John­son’s Emer­gence. I picked up the book because ever since I start­ed think­ing about what IxDs can learn from game design, the con­cept of emer­gence kept pop­ping up.

Johnson’s book is a pleas­ant read, an easy-going intro­duc­tion to the sub­ject. I start­ed and fin­ished it over the course of a week­end. There were a few pas­sages I marked as I went a long, and I’d like to quote them here and com­ment on them. In order, they are about:

1. Prin­ci­ples that are required for emer­gence to happen
2. How learn­ing can be unconscious
3. Unique skills of game players
4. Gar­den­ing as a metaphor for using (and mak­ing) emer­gent systems

A cheat sheet

Let’s start with the prin­ci­ples.¹

“If you’re build­ing a sys­tem designed to learn from the ground lev­el, a sys­tem where macroin­tel­li­gence and adapt­abil­i­ty derive from local knowl­edge, there are five fun­da­men­tal prin­ci­ples you need to follow.”

These prin­ci­ples togeth­er form a use­ful crib sheet for design­ers work­ing on social soft­ware, MMOGs, etc. I’ll sum­marise each of Johnson’s prin­ci­ples here.

“More is different.”

You need to have a size­able amount of low-lev­el ele­ments inter­act­ing to get pat­terns emerg­ing. Also, there is a dif­fer­ence between the behav­iour you will observe on the microlev­el, and on the macrolev­el. You need to be aware of both.

“Igno­rance is useful.”

The sim­ple ele­ments don’t have to be aware of the high­er-lev­el order. In fact, it’s best if they aren’t. Oth­er­wise nasty feed­back-loops might come into being.

“Encour­age ran­dom encounters.”

You need chance hap­pen­ings for the sys­tem to be able to learn and adapt.²

“Look for pat­terns in the signs.”

Sim­ply put, the basic ele­ments can have a sim­ple vocab­u­lary, but should be able to recog­nise pat­terns. So although you might be work­ing with only one sig­nal, things such as fre­quen­cy and inten­si­ty should be used to make a range of meanings.

“Pay atten­tion to your neighbours.”

There must be as much inter­ac­tion between the com­po­nents as pos­si­ble. They should be made con­stant­ly aware of each other.

Now with these prin­ci­ples in mind look at sys­tems that suc­cess­ful­ly lever­age col­lec­tive intel­li­gence. Look at Flickr for instance. They are all present.

Chicken pox

I liked the fol­low­ing pas­sage because it seems to offer a nice metaphor for what I think is the unique kind of learn­ing that hap­pens while play­ing. In a way, games and toys are like chick­en pox.³

“[…] learn­ing is not always con­tin­gent on con­scious­ness. […] Most of us have devel­oped immu­ni­ty to the vari­cel­la-zoster virus—also known as chick­en pox—based on our expo­sure to it ear­ly in child­hood. The immu­ni­ty is a learn­ing process: the anti­bod­ies of our immune sys­tem learn to neu­tral­ize the anti­gens of the virus, and they remem­ber those neu­tral­iza­tion strate­gies for the rest of our lives. […] Those anti­bod­ies func­tion as a “recog­ni­tion sys­tem,” in Ger­ald Edelman’s phrase, suc­cess­ful­ly attack­ing the virus and stor­ing the infor­ma­tion about it, then recall­ing that infor­ma­tion the next time the virus comes across the radar. […] the recog­ni­tion unfolds pure­ly on a cel­lu­lar lev­el: we are not aware of the vari­cel­la-zoster virus in any sense of the word, […] The body learns with­out con­scious­ness, and so do cities, because learn­ing is not just about being aware of infor­ma­tion; it’s also about stor­ing infor­ma­tion and know­ing where to find it. […] It’s about alter­ing a system’s behav­iour in response to those pat­terns in ways that make the sys­tem more suc­cess­ful at what­ev­er goal it’s pur­su­ing. The sys­tem need not be con­scious to be capa­ble of that kind of learn­ing.”

Empha­sis on the last sen­tence mine, by the way.

Patience

John­son writes about his impres­sion of chil­dren play­ing video games:⁴

“[…] they are more tol­er­ant of being out of con­trol, more tol­er­ant of that explorato­ry phase where the rules don’t all make sense, and where few goals have been clear­ly defined.”

This atti­tude is very valu­able in today’s increas­ing­ly com­plex world. It should be fos­tered and lever­aged in areas besides gam­ing too, IMHO. This point was at the core of my Play­ing With Com­plex­i­ty talk.

Gardening

“Inter­act­ing with emer­gent soft­ware is already more like grow­ing a gar­den than dri­ving a car or read­ing a book.”⁵

Yet, we still tend to approach the design of sys­tems like this from a tra­di­tion of mak­ing tools (cars) or media (books). I not only believe that the use of sys­tems like this is like gar­den­ing, but also their cre­ation. Per­haps they lie in each other’s exten­sion, are part of one nev­er-end­ing cycle? In any case, when design­ing com­plex sys­tems, you need to work with it “live”. Plant some seeds, observe, prune, weed, plant some more, etc. 

I am going to keep a gar­den (on my bal­cony). I’m pret­ty sure that will teach me more about inter­ac­tion design than build­ing cars or writ­ing books.

1. The fol­low­ing quotes are tak­en from pages 77–79. [↩]
2. This reminds me of Nas­sim Nicholas Taleb’s The Black Swan, where­in he writes about max­imis­ing your chance of hav­ing serendip­i­tous encoun­ters. [↩]
3. Tak­en from pages 103–104. [↩]
4. Page 177. [↩]
5. Page 207. [↩]