Neighbour Computing

  • even though cellular automata and waveform collapse do completely different things, they somehow produce visually similar results
  • neurons in a brain are in a sense neighbour-computing agents, learning to fire at the right time, based on the surrounding neurons (physical distance) and the environment
  • the techniques for querying for neighbours (up/down/left/right) keep popping up in various places - from implementing basic image filters, to GP-GPU simulation techniques, to cellular automata
  • in the Homoiconic Spreadsheets Eli Parra gives an example of building up a "streak counter" using a calculation based on specific condition on the canvas (of the cell and its neighbours):
    [prev-streak] [= if (is-checked) then (prev-streak + 1) else 0]
    • this could possibly mix very well with the ideas of "spatial queries" explored in the Programmable Ink (and Dynamicland), where a pattern like the one above wouldn't be copy-pasted around, but instead declared once (as a "match" on a specific configuration of the cells on the canvas)
  • taking this even further, is using the spatiality for semantic - executing code spatially - Neighbour Computing note says more about this