A Game Developer’s Wish List for Researchers

A Game Developer’s Wish List

for Researche

rsChris Heckerhttp://chrishecker.com

This talk assumes you care

about being relevant to

games.

Disclaimer

Also, I sometimes curse...a lot.

Games & Research Through The Ages

In days of yore, we were

passive consumers of (usually old)

research.These days,

there is active feedback

between the two

disciplines.

How to do a talk on research?

What is our

highest technology

priority?

Know Your Audience

performance?

Know Your Audience

1. robustness2. simplicity3. performance

Game Technology Priorities

Technology is a means, not an end.

Robustness

edge cases?failure modes?

parameters

simply connecte

d?

downsides?

negative results?

Interactivity prioritizes robustness!

Robustness

Simplicity

me

game

you

We are always about to fail!

Simplicity“...even crudeness, if two sticks and a rock

will do it, great!”

few parametersintuiti

ve outpu

t

pipeline

integration

explainable to artistsdepend

enciescode

preprocessing

markuporder

compatibility

art directable

Performancethe

constant

mattersms,

not fps!

real compariso

nsalgorithms

implementations

inputsscenes

working sets

not just embarrassingly parallel

worst casevs.

average case?

Real Data

Real Data

Real Data

Source Code: Yes!

source code is more rigorous

the cost to verify a paper actually works is prohibitively high

small fraction of papers are relevant;

small fraction of those work as advertised

What to research?Avoid solutions looking for problems.

Talk to game developers.

Graphics – must integrate well.AI – is game design.

Animation – interactivity is king.Perceptual Models

and Metrics – what is important?

Miscellaneous StuffPatents – don’t!

Patents – if you must be a douche, disclose your

douchiness in the abstract.Put your paper online, not

behind paywalls.Publish negative results.Answer emails.

Play games.

Thank you.

chrishecker.com