Department of Alchemy 4-273

For the rest of the summer, I’ll be in the office on Tuesdays, so I won’t be able to attend the Berkman luncheons in person. However, I tuned in today via live webcast (oh the wonderful innovative potential of technology) and took down notes. The discussion about borrowing and novelty in collaboration hit home a bit, from my very strange experiences in Calculus AB during junior year of high school. I won’t get into why my teacher limited the number of questions I could ask per class (maximum of three per day), but the two or three quizzes we had per week were collaborative efforts between two or three people to arrive at a shared grade. I still find it weird that my best group ended up during my pairing with one of the slackers of the class, while I performed near the top. A strange team, yet I’d say there was limited tension between the novelty and reuse of applying our skills to solving the few questions on the quiz sheet. I’d usually bring to class the necessary new material while my partner would go over my work, rework it in places, and sort of the small mistakes that I missed in review. The value of my original material and his reuse of my applied knowledge, I’d say, was fairly equal.

So, on to the notes…

—

The Dynamics of Collaborative Innovation: Exploring the tension between knowledge novelty and reuse

Karim Lakhani, Ned Gulley

Karim:

we think collaborative innovation as more modern: open-source/Wikipedia
most major innovations: highly collaborative in history

airplane development: not just Wright brothers, but creation with multiple people
pre-Wright brothers: network of 10 individuals
locus on innovation: moved over to Europe after Wright brothers

collaborative innovation: Meyer’s Analysis

dynamics of collaborative innovation: how people build off of others’ work

Ned:

contest at MathWorks: MATLAB programming contest
usually: smartest person gets the prize
but: not how ideas move/work in the world
contest: notion of borrowing/stealing ideas in contest: create page of code

Competitive Wikipedia
everyone: encouraged to edit articles
if article made worse: thrown away; if better: article edit it kept
would Wikipedia display article editing winner?

MATLAB week-long open collaborative competition for programmers
- entries automatically scored, ranked, displayed immediately
- code author score are visible at all times
- anyone can modify other’s code

leaders –> view entry: person makes new entry and becomes leader

first place: completely objective
good code: gets better optimization score from test lead

really about reputation and interaction with community

what we see in practice:
people: anxious to acknowledge people they took code from

types of changes:
- Big changes (leaps)
I know a much better way to do this, replaces previous code
- Small changes (tweaks)
minor optimization; tweakers don’t need to understand full optimization to improve code

code: improves over time
sometimes: people take best code at certain point in time & make it worse

by inserting new idea after previously solved problem: people swarm on it to work with and improve idea

tough question: how would you value tweakers over leapers
hard to say who really is making the important contributions

systematic variations: tweak bombs: take the entry in the lead, sniff around for secret number replacements to test
changes to the lead entry: fly off like sparks

social signals: sent through entry titles
- scrambled eggs
- rotten eggs
- I didn’t start the fire
- Don’t get obfuscated… follow the light
- You Call This Collaboration? Give Me a Break

motivation:
to participate: opportunity: for personal glory or collaboration?

behavior of successful code:
high rank, time on top, high status author, clarity, elegance, novelty, etc.

tension: not between any two coders
code: wants to propagate
coder: wants to block code propagation

a chicken is only an egg’s way of making another egg
a hacker is only code’s way of making more code

—

Karim:

collaborative innovation: implicit tension between collective and individual:

collective point of view: value contributions that get reused more often
individual view: value being the top amongst peers

social value of contribution (code) = # of times lines of code reused
relative novelty: helps you; too new: others don’t use it/know what to do with it
value of adding new things, after a while: gets too complicated
not much value in borrowing code, but if you use it in the right way it’s very valuable

leaders: borrow > novelty, in this setting