Hacker News: tetrisgm

New comment by tetrisgm in "Turn your regular JavaScript code into a neural net"

tetrisgm — Fri, 16 Oct 2026 02:37:56 +0000

I see people hyping this on Twitter and I have no idea why. I don’t understand its applications.

New comment by tetrisgm in "AnyPS5: Port PS5 binaries to PC without emulation (87% system libraries mapped)"

tetrisgm — Fri, 09 Oct 2026 05:50:40 +0000

That’s quite far away. There are some fake ones that approximate it by doing mods

New comment by tetrisgm in "AnyPS5: Port PS5 binaries to PC without emulation (87% system libraries mapped)"

tetrisgm — Wed, 07 Oct 2026 06:14:52 +0000

False information. There are several examples of games that are patched to run on lower firmware.

New comment by tetrisgm in "AnyPS5: Port PS5 binaries to PC without emulation (87% system libraries mapped)"

tetrisgm — Wed, 07 Oct 2026 06:13:54 +0000

We are basically doing that on https://retroportingtoolkit.com

New comment by tetrisgm in "Vibecoding isn't as fun as writing code by hand"

tetrisgm — Tue, 06 Oct 2026 16:44:32 +0000

Making food in the microwave isn’t as fun as cooking by hand.

The comparison misses the utilitarian aspect.

New comment by tetrisgm in "I quit OpenAI because its culture is broken"

tetrisgm — Sat, 03 Oct 2026 18:33:10 +0000

These companies are large enough that someone is going to quit and feel very validated about their world view and how they are not aligned. That’s what makes it worthy of leaving in the first place. However that doesn’t make their criticism more valid or more worthy of coverage.

New comment by tetrisgm in "Several vulnerabilities have been discovered in the Linux kernel"

tetrisgm — Fri, 02 Oct 2026 00:34:17 +0000

That’s probably a great thing. The initial friction of AI overwhelming projects certainly sucks, but once there are better processes to deal with them it’s going to strengthen the quality of so many projects!

New comment by tetrisgm in "Codex Pricing: Pro has unlimited 5.6 usage"

tetrisgm — Wed, 30 Sep 2026 21:37:16 +0000

This had the asterisk of “reasonable use”, so unlimited is actually practically bounded. That limit is not stated

New comment by tetrisgm in "OpenAI: Tomorrow we are re-opening the Pro $200 subscription"

tetrisgm — Tue, 29 Sep 2026 22:58:21 +0000

Earlier this week I felt that Codex was close to on par with Opus 5.5, and that the resets made it better. I considered a second subscription.

It’s very disappointing to get an email from OpenAI explaining that my $200 will go from 20x to 10x. That is not what I paid for.

I cannot trust that the $500 plan will reliably give me the results I wanted. I suspect it will keep changing, making my workflow random.

New comment by tetrisgm in "Drawgent: Coding agent on a live Excalidraw canvas"

tetrisgm — Sat, 26 Sep 2026 20:20:04 +0000

Right I've already got MCP integration but it seems quite clunky. Do you have an example of an MCP implementation that you like? and maybe an example of code you used that did the integration nicely? Would love some pointers

New comment by tetrisgm in "Drawgent: Coding agent on a live Excalidraw canvas"

tetrisgm — Sat, 26 Sep 2026 20:19:57 +0000

New comment by tetrisgm in "Drawgent: Coding agent on a live Excalidraw canvas"

tetrisgm — Sat, 26 Sep 2026 17:58:53 +0000

I’ve been trying to find the best practice to connect a person’s existing ChatGPT/Codex / Claude accounts so they can be used in an app for the agent in the app.

Any advice? Everything seems clunky, even the best MCP efforts

New comment by tetrisgm in "Plan mode is dead"

tetrisgm — Sat, 26 Sep 2026 08:43:48 +0000

I have noticed this in practice the past month. It’s interesting because goals and plans are obsolete. Skills too.

Much of what distinguished a knowledgeable AI dev a year ago is now an anti pattern.

New comment by tetrisgm in "OpenAI have no mathematicians capable of understanding what they put out"

tetrisgm — Thu, 10 Sep 2026 07:49:47 +0000

The problem is that eventually, there will be no humans who can follow the results AI will give. That’s the endgame for this tech: to produce knowledge at speeds and quality beyond what we can

New comment by tetrisgm in "“Next-token predictor” is the wrong mental model for LLMs"

tetrisgm — Sat, 05 Sep 2026 01:15:27 +0000

That's perfectly put. The value of words and what we assign to them, let alone how we assign meaning, or the interpretation of what the other person says, is completely arbitrary. (in practice).

New comment by tetrisgm in "Portal by Spotify cut my Claude Code token usage by 90%"

tetrisgm — Sat, 05 Sep 2026 00:46:39 +0000

This is just offshoring but for models

New comment by tetrisgm in "Sony makes bold claim about game ownership"

tetrisgm — Thu, 03 Sep 2026 16:39:10 +0000

This is going to backfire spectacularly.

Between this claim, ramping down disc production, and the possibility of a $999 home console,they are going to have a difficult time.

There are still so few PS5 exclusives. Even fewer with PS5 Pro support. It’s going to take a lot to gain trust and justify a PS6. No clear path to victory here with a hardware update.

New comment by tetrisgm in "Is MCP Good Yet?"

tetrisgm — Tue, 01 Sep 2026 06:07:05 +0000

I have implemented MCP in a few projects. It’s gotten some traction with people who tinker with agents.

CLI or API plus docs does everything it can, with more coverage. But the thing is that so many vibe coders are excited to build a stack of tools, it’s still worthwhile to meet their expectations.

New comment by tetrisgm in "Fastpotify"

tetrisgm — Tue, 01 Sep 2026 05:21:18 +0000

and?

New comment by tetrisgm in "IPFS Maintainers Winding Down"

tetrisgm — Mon, 24 Aug 2026 19:34:32 +0000

Agreed. I was deep in the ecosystem, did the confs, shipped some apps. It’s hard to think of a use case that would be reliable now. Maybe some kinda file sync that is torrent like but http