Hacker News: lebek

New comment by lebek in "GPT 5.6 Sol is the best "vision" model OpenAI ever released"

lebek — Mon, 17 Aug 2026 13:49:13 +0000

The point is that it's general. It can do this task and many other tasks and it doesn't need custom development like OpenCV does. Of course if you only want to count pills and you want it to be cheap/fast you're still better off using OpenCV.

Shared Memories for Agent-Powered Communities

lebek — Fri, 14 Aug 2026 13:29:35 +0000

Article URL: https://peterlebek.com/posts/shared-memories-for-agent-powered-communities.html

Comments URL: https://news.ycombinator.com/item?id=49298420

Points: 1

# Comments: 0

Prompt Caching

lebek — Sat, 25 Jul 2026 07:21:56 +0000

Article URL: https://earendil.com/posts/prompt-caching/

Comments URL: https://news.ycombinator.com/item?id=49045285

Points: 2

# Comments: 0

Reeks, Wrecks and Robots

lebek — Tue, 16 Jun 2026 10:30:32 +0000

Article URL: https://www.washingtonpost.com/archive/politics/1982/07/21/reeks-wrecks-and-robots/c3b63ac8-a823-4c41-89b6-fd7785ff67ec/

Comments URL: https://news.ycombinator.com/item?id=48553132

Points: 2

# Comments: 0

Open-source AI modding tool for Rimworld (and soon Factorio)

lebek — Wed, 13 May 2026 13:51:40 +0000

Hi everyone,

I'm an indie game developer and Rimworld fan. For the past month I've been working on a new tool for building & fixing Rimworld mods using AI: https://modmixer.com/

Modmixer decompiles & indexes all your game/DLC source code so the model can easily understand it (without costing as much as Claude!), it also launches the game, populates the debug palette, and monitors game logs for errors. When errors arrive it fixes them automatically and relaunches the game. Overall I think we can build mods that are MORE robust and performant and widely compatible by using AI than without.

That said, it still requires a "driver" who actually cares about making something decent. Modmixer can't solve creativity, it can't play the game to test the mod, it can't do game design, it can't tweak values until they feel just right. You still have to do all that!

The project is open-source / MIT License. Here's the github repo: https://github.com/lebek/modmixer

I'm excited about a lot more players being able to mod the games they play. It's going to get so easy that building a mod just for yourself/your friends will be a normal thing to do.

Let me know what you think, or join us in Discord, there are only 4 of us : https://discord.com/invite/54QhJeNvFy

Comments URL: https://news.ycombinator.com/item?id=48121901

Points: 1

# Comments: 0

The day you get cut out of the economy

lebek — Wed, 08 Apr 2026 14:24:05 +0000

Article URL: https://geohot.github.io//blog/jekyll/update/2026/04/08/the-day-you-get-cut-out.html

Comments URL: https://news.ycombinator.com/item?id=47690670

Points: 3

# Comments: 1

Thoughts on Slowing the Fuck Down

lebek — Wed, 08 Apr 2026 14:14:57 +0000

Article URL: https://mariozechner.at/posts/2026-03-25-thoughts-on-slowing-the-fuck-down/#toc_5

Comments URL: https://news.ycombinator.com/item?id=47690544

Points: 3

# Comments: 1

New comment by lebek in "Did it get dumber? Tracking Claude Code and Codex according to HN comments"

lebek — Wed, 08 Apr 2026 11:12:32 +0000

Done: https://agentiqtracker.com/#install

Did it get dumber? Tracking Claude Code and Codex according to HN comments

lebek — Wed, 08 Apr 2026 11:01:03 +0000

Article URL: https://diditgetdumber.com/

Comments URL: https://news.ycombinator.com/item?id=47688478

Points: 3

# Comments: 2

New comment by lebek in "AI (2014)"

lebek — Fri, 20 Mar 2026 11:42:25 +0000

> Then human work is changed to figuring out new things and the AI solves all old things, that seems much more fun than most white collar work today.

But it's not fun to be figuring out new things all the time. Some amount of routine work is necessary to 1) exercise mastery (feels good), and 2) recover energy. This is why a lot of people find agentic coding exhausting and less fun, you're basically always having to be creative (what's the next feature?) or solve the hardest 5% of issues the LLM can't handle.

Moltbook is a bad takeoff scenario where human psychology itself is the exploit

lebek — Sat, 31 Jan 2026 13:16:59 +0000

Article URL: https://twitter.com/PimDeWitte/status/2017571392384872742

Comments URL: https://news.ycombinator.com/item?id=46836415

Points: 42

# Comments: 15

Please, Don't Automate Science

lebek — Thu, 11 Dec 2025 12:00:57 +0000

Article URL: http://togelius.blogspot.com/2025/12/please-dont-automate-science.html

Comments URL: https://news.ycombinator.com/item?id=46230335

Points: 1

# Comments: 0

New comment by lebek in "The baffling intelligence of a single cell: The story of E. coli chemotaxis"

lebek — Thu, 21 Mar 2024 18:35:09 +0000

I understand that, but it seems like even the MVP "shitty" flagellum would require many mutations that individually have no benefit. But I suppose with enough generations/parallelism you get enough stacking of useless mutations to reach the useful ones.

New comment by lebek in "The baffling intelligence of a single cell: The story of E. coli chemotaxis"

lebek — Thu, 21 Mar 2024 17:09:35 +0000

I think he's saying, random mutation wouldn't produce all required components at once. One mutation gives you a bit of a flagella, another gives you bit of a nose, but how does the flagella mutation survive to coexist with the nose mutation that makes it useful.

I suspect the answer is that having flagella without a nose is still better than having no flagella. If so it suggests evolution isn't good at accessing groups of mutations that aren't individually beneficial.

New comment by lebek in "The False Promise of Imitating Proprietary LLMs"

lebek — Fri, 26 May 2023 04:32:29 +0000

Link to said research?

New comment by lebek in "The False Promise of Imitating Proprietary LLMs"

lebek — Fri, 26 May 2023 03:35:32 +0000

> nobody at this point expects a 13B parameter model to succeed with the same accuracy at the broad range of tasks supported by what may be a 1T parameter model

I think a lot of people believe exactly that. To take one example from the "We Have No Moat" essay:

"It doesn’t take long before the cumulative effect of all of these fine-tunings overcomes starting off at a size disadvantage. Indeed, in terms of engineer-hours, the pace of improvement from these models vastly outstrips what we can do with our largest variants, and the best are already largely indistinguishable from ChatGPT." - https://www.semianalysis.com/p/google-we-have-no-moat-and-ne...