ncnvrgnt
banner
inconvergent.net
ncnvrgnt
@inconvergent.net
procedural & graphics artist
parallel programmer
tired

website | https://inconvergent.net
gallery | https://img.inconvergent.net/plot/
a game | https://boxtype.app
store | https://buy.inconvergent.net
Pinned
i published a first version of my browser puzzle/platform game (+ level editor). you can play it for free (w/a keyboard) at boxtype.app. levels are encoded in the url; if you make something you like you can share it with anyone

i'd love to hear what you think!
its not new that you should consider whether you trust a service before giving them your data. but doubly so now that almost anyone can make a service-shaped artefact on the web in days. complete with every imaginable security hole and rookie mistake
February 1, 2026 at 8:17 AM
at least we can agree that more meetings should have been emails
February 1, 2026 at 1:38 AM
Reposted by ncnvrgnt
you can stack or layer LLMs any way you like; you are not going to create conciousness. no matter how much you want it to be true
January 31, 2026 at 2:43 PM
Reposted by ncnvrgnt
computers talk to each other all the time, and have done for decades. having them use natural language is just a step back in terms of efficiency and security
January 31, 2026 at 6:14 AM
this API requires a GPU ...
January 31, 2026 at 6:04 PM
things, i gather, have gone out of control
January 31, 2026 at 5:49 PM
i liked it better when people didn't care that much about computers in general. or at least i preferred it over everyone being enthusiastically wrong about computers everywhere and all the time
January 31, 2026 at 3:22 PM
i think this agument (dev. speed) can be applied to software dev in general. and possibly also to explain why some are so eager to use coding agents: they have bad tooling
I think most game dev teams underestimate how damaging slow workflows are to the ultimate quality of their game.

If it takes 5 minutes to launch the game and see your changes, I think your game is going to be MUCH worse than the alternate timeline where it took 1 minute
January 31, 2026 at 2:55 PM
you can stack or layer LLMs any way you like; you are not going to create conciousness. no matter how much you want it to be true
January 31, 2026 at 2:43 PM
computers talk to each other all the time, and have done for decades. having them use natural language is just a step back in terms of efficiency and security
January 31, 2026 at 6:14 AM
Reposted by ncnvrgnt
i like the tiny rainbow in particular
cube monument rendering for cube monument rendering reasons
January 28, 2026 at 7:02 PM
Reposted by ncnvrgnt
similar to an earlier rendering, to confirm that the tweaks to the volume light is behaving as expected
January 29, 2026 at 11:04 PM
qmv is good for renaming multiple files in the terminal. it opens the files in two columns in an editor, and you can edit the file names in the second column
January 30, 2026 at 5:07 PM
Reposted by ncnvrgnt
here is the other side. also not ominous. no orbs.
January 24, 2026 at 2:27 PM
Reposted by ncnvrgnt
another rendering of the cube monument for [redacted] cube monument reasons
January 28, 2026 at 7:45 PM
similar to an earlier rendering, to confirm that the tweaks to the volume light is behaving as expected
January 29, 2026 at 11:04 PM
i found another reason to use an exponential decay function
January 29, 2026 at 10:40 PM
Reposted by ncnvrgnt
The Virgo, my concept for a crewed Venus exploration vessel in the 2100s, based on the astrological glyph for the planet. The whole structure is spun, and the tether increases the overall radius for the inhabited sections while saving mass.
January 29, 2026 at 10:10 PM
Reposted by ncnvrgnt
Recommending this account labeler from Cornell that makes it easy to spot suspicious accounts and ones that are follow-farming.

bsky.app/profile/stec...
January 29, 2026 at 2:03 AM
another rendering of the cube monument for [redacted] cube monument reasons
January 28, 2026 at 7:45 PM
i like the tiny rainbow in particular
cube monument rendering for cube monument rendering reasons
January 28, 2026 at 7:02 PM
Reposted by ncnvrgnt
This very well articulates a rough idea I’ve had for a while which is that intellectual property theft is the wrong framing to understand what’s happening (and courts will legitimize it for pennies on the dollar anyways, as we’ve seen), but mass dispossession / enclosure is much more apt
January 28, 2026 at 1:28 PM
Reposted by ncnvrgnt
Using PGA2D (2D Plane-based Geometric Algebra) to explore wallpaper groups. (Still making examples for this library 😆.)

en.wikipedia.org/wiki/Plane-b...

#geometer
January 28, 2026 at 12:55 PM
good post about how generative AI (search) services will degrade information on the internet further michiel.buddingh.eu/enclosure-fe...
The Enclosure feedback loop
michiel.buddingh.eu
January 28, 2026 at 11:59 AM
every new AI product is another siphon made to gather original training data for that company
January 28, 2026 at 9:53 AM