Tomas Petricek
tomasp.net | tomas@tomasp.net | @tomaspetricek
Convivial tools
Tool shouldn't just make some activity more convenient
Tools should preserve individual agency!
Avoiding 'bad' complexity
Don't paper over tool deficiencies with a second tool
Maintenance leads to compounding claims on time
Take the first tool out, think the problem anew
Mu is a stack designed from the ground up: to fit in a single brain and to not grow complex over time.
Pre-modern
Craftspeople crafting their own low-tech tools
Modern
Scientist designs optimal tools for us
Post-modern
Messy pluralist world that we need to navigate
Scientific forestry
Spruce forests are easy to plant and harvest
They get infected and are not resistant
Soviet collectivization
Standardized organization of farming
Not flexible enough to produce e.g. raspberries
Dark twins?
"As it executes instructions, the SubX emulator
monitors for labels that start with a $watch-
prefix."
Unexpected consequences?
What desirable systems would become hard to think/create using the Mu approach?
Using fewer abstractions - carefully designed to leak in just the right ways - can make the maintenance task more approachable to end users.
Mu approach - pre-modern?
Use with caution, make sure system 'fits in the brain'
Haskell approach - modern?
Optimal non-leaky abstractions designed by scientists
Alternative approach? - post-modern?
Cannot fit in the brain, but need to be able to cope..
Kevin Lynch (1960). The Image of the City
Mu is an impressive attempt to address
What is the approach and its dangers?
Tomas Petricek | tomas@tomasp.net | @tomaspetricek