(First post, so: I'm the author of the Chromatron games and Lost in the Static. I'm also a "ringer", as I worked on some commercial games in the latter half of the 90s.)
This was a simple gameplay idea that I tested and couldn't make go anywhere: you slide the rows or columns horizontally or vertically, but you can't slide one if it's got a multi-row "brick" crossing to another row--or it forces both rows to move together (which then may cascade through other bricks). I couldn't find any good gameplay out of the idea.
I wrote Chromatron 1 two months later.
This was not a Chromatron sequel, but was intended to "continue the line" of those games -- it was another puzzle game with a color theme and name. You build these "abstract paintings" and then a "critic" "rates" them; it's actually basically just Zendo
; the biggest problem was how to prove that you knew a judge's rule. Multiple testers suggested a big collection of combo boxes where you spell out your guess, but I didn't like the idea of giving people a hint about what the range of things it worked with were. Instead, when you thought you had "solved" a critic, you could take a test, in which you were shown ten "paintings" and had to predict which way you thought he'd choose. If you got all of them right, the game assumed you solved it. If not, it showed you the right answer to one of the paintings you tested wrong on, corresponding to the Zendo 'construct a counterexample' rule. And when you retook the test you'd get a different set of paintings, and more of them (i.e. 12 the 2nd time, 14 the 3rd time, etc., to discourage brute force guessing).
The intent was that there'd be 30-40 critics, and then some "boss levels" where you had to make paintings that would make multiple critics happy simultaneously.
Besides dissatisfaction with the testing mechanic, another thing that made me unhappy about this is that you would tend to create really simple, trivial things for the "experiment" paintings (as seen above), despite the fact that the system would let you make much nicer-looking, more sophisticated things -- I made a ton of these for the tests, but there was really no way to make the user make them more interesting (except maybe by adding some sort of budget and a demand that you spend a certain minimum or something, but it seemed awkward).
But also after stopping, I later never went back to it because I decided intellectual puzzle games just wasn't a very satisfying niche to target.
This was a puzzle game sort of inspired by Four Crystals of Traziere. Here you would collect cards and then create spells with them by combining various aspects. For example in the screenshot the player has built a four-way destruction spell and is about to cast it to destroy the four diagonal pillbox barriers.
I wasn't able to make the puzzly aspect of this interesting enough; I couldn't get any emergent gameplay like in Chromatron, because I had to construct everything very carefully to avoid you getting too stuck. And the mechanic really cries out for emergent insanity, not carefully constructed puzzles. Kinda obvious in hindsight.
was a technology test for doing a 2D scrolling game with 3D hardware, mainly testing getting gamma-correct antialiasing for everything (without pixel shaders). The stars are parallaxing point sprites, not one big texture.
was a UI test for the UI for a 4X galaxy game.
This was a technology test for an idea for doing a 3D Descent game with random levels (i.e. a 3D roguelike). This is a "random level" made with CSG (with an arbitrary texture thrown on it and random lights). Because of my experience with CSG numeric-stability issues in a real shipped game, since this would generate random levels and designers couldn't tweak to fix it, it would only use axis-aligned surfaces.
This was a game I worked on after Lost in the Static, starting with the same engine, but now I was doing weird colored animation. I never found any compelling gameplay for it (you could switch your guy between red and green, and invert gravity, and there was room for some interesting platforming using those ideas, but it didn't have any oomph, no real hook that would make people want to play). Yes, that's what the avatar in LitS really looks like.
This is a bullet-hell shooter where you control the bullets (indirectly) in a Desktop-Tower-Defense-like way, and the bullet-dodgers are the enemies. I wrote very awesome bullet-dodging AI for the enemies, and the graphics, and a few turret types, enough to test the AI. The gameplay doesn't have the maze-construction aspect of DTD games, so there'd need to be something else to fill in for it--I was leaning towards having "support" systems you could build, which could e.g. synchronize all adjacent turrets, or make them fire faster, or etc. But despite being excited by watching the bullet dodging (they're much more sophisticated than human players--the collision "box" is the actual ship, and you can throw crazy patterns at them that humans could never possibly navigate, like mixtures of very different speeds--which means it's very different than it would be to have a computer player playing a regular human-playable bullet-hell game), I've never found the will to actually try building out the rest of the game, much less figured out if there's anything fun there. (No, it never actually reached 'beta', I have no idea why I added that so prematurely.)
I didn't actually intend the art to be DTD-ish, I just drew turret placeholder art that happened to look a little that way.