This document summarizes Stuart Church's presentation on applying evolutionary thinking to design and behavior. It discusses how designs and behaviors are subject to natural selection processes like variation, replication, and selection. Successful designs and behaviors persist while unsuccessful ones fail. Experimental approaches that generate many variations, like Amazon's product experiments, can accelerate innovation. Most new innovations are small modifications of existing ideas and behaviors within the "adjacent possible." Cooperative behaviors can emerge from strategies like reciprocity in iterated games. Copying others' successful behaviors and designs is also an evolutionarily advantageous strategy.
Behavioural Meetup: Stuart Church on Darwin to Design
1.
2. From Darwin
to Design
Stuart Church
http://www.flickr.com/photos/tsevis/3288860652
@stuchurch
Applied Behavioural Science Meetup, Bristol
19 February 2015
3. A tale of two careers
Academic Research User Experience
(Other) Animal Behaviour (Human) Animal Behaviour
4. Todayā¦
1. What can we learn about design &
innovation from evolutionary systems?
2. What can we learn about behaviour
from evolutionary thinking?
5. Evolution & adaptation
Evolution is the change in the inherited characteristics of
(biological) populations over successive generations.
āThe survival of the fittestā
Variation
Replicators
Selection
8. What can we learn about
design & innovation from
evolutionary systems?
9. Designs are ideas that are
culturally inherited. Good designs
serve a purpose and persist. Poor
designs get forgotten.
http://www.ļ¬ickr.com/photos/dvanzuijlekom/7324829530
The unit of selection is the idea
rather than the design itself.
Designs as memes
10. Species #fail
99.9% of all species that have
ever existed are extinct
http://www.ļ¬ickr.com/photos/bugmonkey/2844115494
Product #failāØ
80-95% of new products fail in
the first year (Source: Acupoll)
Failure is the norm
14. Can design be replaced by
experiments?
āIf you double the number of
experiments you do per year
youāre going to double your
inventivenessā
Jeļ¬ Bezos, 2004
(Amazon runs 1000+
experiments per year on its
website)
By Steve Jurvetson (Flickr: Bezosā Iconic Laugh) [CC BY 2.0 (http://creativecommons.org/licenses/by/2.0)], via Wikimedia Commons
16. Most successful ideas & innovations
tend not be that diļ¬erent from what
already exists.
On a fitness landscape, feasible steps
are the closest steps in gene space or
meme space.
The āadjacent possibleā
17. http://www.ļ¬ickr.com/photos/33134305@N04/3090115697
1992 2007
Can ideas be too innovative?
āThere was no mental slot in peopleās heads that the
Newton could glide intoā¦.Consumers are willing to
overlook technical glitches if they have a firm grasp of what
a product is and what itās supposed to do. ā
āWhatās important is that, for the first time, so many
great ideas and processes have been assembled in one
device, iterated until they squeak, and made
accessible to normal human beings.Ā Thatās the genius
of Steve Jobs; thatās the genius of Appleā.
Bruce Tognazzoini
18.
19. What can we learn about
behaviour from
evolutionary thinking?
24. "Grooming monkeys PLW edit" by Muhammad Mahdi Karim (www.micro2macro.net) FacebookYoutube (original photograph), Papa Lima Whiskey (derivative edit) - Own work. Licensed under GFDL 1.2 via Wikimedia Commons - http://
commons.wikimedia.org/wiki/File:Grooming_monkeys_PLW_edit.jpg#mediaviewer/File:Grooming_monkeys_PLW_edit.jpg
25. The Prisonerās Dilemma
Played once
Only sensible strategy is to DEFECT
For Iterated Prisonerās Dilemma
Best strategy is TIT-FOR-TAT (be
nice, then follow opponent). i.e. the
best possible strategy in the long
term is a cooperative one.
A classic example of āGame Theoryā
ā¢ 2 suspects, arrested on suspicion of a crime
ā¢ Put in separate cells and asked to testify against the other
ā¢ So, can either testify (defect) or stay silent (cooperate)
26. āRaising the stakesā
What if we can vary the amount of we invest in altruistic acts? (e.g. amount
of time spent grooming).
Some possible strategies are:
ā¢ Give as good as you get
ā¢ Non-altruistic
ā¢ Raise the stakes
ā¢ Short changer
ā¢ Occasional short changer
ā¢ Occasional cheat
āRaise the stakesā is an ESS - start low and, if matched, invest more next time
Roberts & Sherratt (1998) Development of cooperative relationships through increasing investment. Nature 394: 175-179
27. Reciprocity is powerful & deeply embedded
Start with small, low risk acts of
kindnessā¦.
Implications
āGivingā (in whatever form) to people/customers etc elicits
a strong desire for others to respond in kind
30. āCopyingā is a successful strategy
Science. 2010 Apr 9;328(5975):208-13. doi: 10.1126/science.1184719.
Why copy others? Insights from the social learning strategies tournament.
Rendell L1, Boyd R, Cownden D, Enquist M, Eriksson K, Feldman MW, Fogarty L, Ghirlanda S, Lillicrap T, Laland KN.
Rendell et al (2010) set up a computer tournament to explore how successful
social learning strategies are in the long term.
The scenario was a ārestless multiarmed banditā - 100 behaviours, each
associated with a diļ¬erent payoļ¬. Aim is to maximise payoļ¬ over the long
run. 108 teams entered diļ¬erent strategies.
Strategy options are to: innovate, copy or exploit.
The winning strategy relied almost exclusively on social learning!
āIn an environment where the world is changing, the best strategy is a lot if
imitationā