Skip to content
View con-dog's full-sized avatar
✍️
Cooking
✍️
Cooking
15 followers · 0 following
  • Global

Block or report con-dog

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
con-dog/README.md

Me

I build 山乇丨尺ᗪ small fast projects.

Note

If it has already been done before, I'm (probably) not interested.

If you tell me it can't be done, I will try to find a way to do it.

I write in a Stream of consciousness style, so expect grammatical errors, mistakes, and other superfluous language issues throughout my projects that I won't fix anytime soon.

Stats for Nerds

Note

All benchmarks performed on a 2021 MacBook Pro with 16GB RAM

Warning

Some benchmarks and claims change as I continue to revisit and expand / compress projects

Compression
Metric Description Project
1 byte Snibble's custom event protocol size snibble / snibble-bench
14 bytes Maze-solving Neural Network solving 96.5% of unseen mazes maze-solver
20MB Total size for a 100×100×10 chunk ray-casted world with O(1) lookup performance chunked-z-level-raycaster
100 bytes Typical Snibble replay size for a 5 minute game (entire state) snibble / snibble-bench
Performance
Metric Description Project
2ms Snibble's average total frame time with 256 simultaneous players (logic AND rendering) snibble
30x Speed of Microlex JavaScript lexical classifier vs PrismJS under equivalent benchmark microlex
2048 Simultaneous agents performing complex spatial/lexical/adversarial tasks at 60FPS snibble / snibble-bench
Accuracy
Metric Description Project
97.3% Blink link navigation prediction accuracy with a 10 byte Neural Network blink
100% Exact replay accuracy for Snibble replays snibble / snibble-bench

How

  • Focusing on performance (highest speeds, tiniest sizes)
  • Removing conventional layers (servers, databases)
  • Under heavy self-inflicted constraints (e.g. "can this fit in a QR Code?")

Warning

I’ve used LLMs heavily since 2026 to speed up experimentation.

Old problem: My idea generation outpaced my execution.

New problem: My execution can now outpace understanding.

I'm learning to navigate this problem in newer projects.

Communication

Caution

Include "con-dog" in your email title else your email will be filtered straight to spam.

Reach me on [email protected] (email alias)

Pinned Loading

  1. snibble snibble Public

    Can a multiplayer word game become a deterministic simulation engine? Yes it can

    1

  2. tiny-neural-network tiny-neural-network Public

    Can a 14 Byte Neural Network Solve a Maze? Can it transfer across tasks?

  3. snibble-bench snibble-bench Public

    Can a word game become an LLM benchmark - testing spatial lexical and adversarial reasoning? Yes it can

  4. chunked-z-level-raycaster chunked-z-level-raycaster Public

    Can we approximate Minecraft scale worlds using a raycasting rendering approach? Not quite!

    C 7

  5. serverless-architecture serverless-architecture Public archive

    An experiment in quines and bookmarklets - can a bookmarklet be used to update its own state?

    JavaScript 34 1

  6. hack-emulator-c hack-emulator-c Public archive

    Partially emulated HACK PC from primitive logic gates (in C)

    C 31 5