Nearing feature-completeness on the recommendation prototype. I've held off on doing any extensive style work on this one longer than usual -- counting on that to-be-implemented style to tie things together at the end.

User interface design.

via https://twitter.com/ben_304/status/947118897781547008

I'm trying out Airtable by categorizing all of the clothes I own.

From: https://airtable.com/

Evidently in Ganbare Dunk Heroes each side had three baskets you could shoot at. An impressively weird variation on the game of basketball.

via https://www.giantbomb.com/articles/colin-spacetwinks-top-10-games-of-2017/1100-5670/

I refactored the Recommendation prototype and most of the views are working. I need to figure out some design themes to make it feel more solid as an app. It doesn't quite know what it is yet, and because of that it feels brittle to use.

I changed fish shell's welcome message to use quotes from Oblique Strategies.

via https://github.com/mattreduce/oblique-fortunes

This looks fun to make/use.

From: https://twitter.com/GalaxyKate/status/946483362264317952

Artificial life experiments by Loren Schmidt.

From: https://twitter.com/i/moments/943912765797912577

I'm super interested in what Dyanimcland is up to.

From: https://dynamicland.org/

The recommendations prototype has taken a turn.

Frank Chimero continues to make understated sites with lots of subtle character. The layout is interesting and different without being overly showy about it.

From: https://faculty.co/ via https://twitter.com/frank_chimero/status/940650000979255296

I switched it so instead of dogs peeing at random they pee when their bladders are full. (Water source coming soon, for now their bladders just fill automatically.)

The dogs are now more likely to pee where a dog has already peed.

NYC Subway Travel Time Map by Nate Parrott.

From: http://subway.nateparrott.com/ via https://twitter.com/_callil/status/939139835813683205

Mario Clock by Zach Gage.

From: http://www.marioclock.com/ via https://twitter.com/helvetica/status/939146552035237888

I got better Three.js circle results using a 256x256 image with no hard edges (no semi-transparent pixels). They are still a little blurry around the edges though.

My circle sprites are lumpy and I don't really understand why.

Interesting interface for animation. Used with a neural network here, but you could do the same thing with a procedural generation system. I'm still trying to figure out for myself how machine learning products can learn from procedural generation experiments.

From: https://vusd.github.io/toposketch/

After lots of data structure reworking the visualization and the sidebar are now in sync.

Implementing dog pee.