Hacker News: kimjune01

New comment by kimjune01 in "They’re made out of weights"

kimjune01 — Thu, 04 Jun 2026 19:19:22 +0000

In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6.

"What are you doing?", asked Minsky.

"I am training a randomly wired neural net to play Tic-tac-toe", Sussman replied.

"Why is the net wired randomly?", asked Minsky.

"I do not want it to have any preconceptions of how to play", Sussman said.

Minsky then shut his eyes.

"Why do you close your eyes?" Sussman asked his teacher.

"So that the room will be empty."

At that moment, Sussman was enlightened.

New comment by kimjune01 in "Ian's Secure Shoelace Knot"

kimjune01 — Thu, 04 Jun 2026 18:26:19 +0000

oh this again, i found this on stumbleupon circa 2007

Auditing DeepSWE

kimjune01 — Sun, 31 May 2026 16:11:50 +0000

Article URL: https://june.kim/auditing-deepswe

Comments URL: https://news.ycombinator.com/item?id=48346878

Points: 3

# Comments: 0

Show HN: 97% on SWE-bench Verified with subscription-token agents

kimjune01 — Sun, 24 May 2026 18:03:28 +0000

Article URL: https://github.com/kimjune01/swebench-verified

Comments URL: https://news.ycombinator.com/item?id=48259575

Points: 2

# Comments: 0

New comment by kimjune01 in "AI is making me dumb"

kimjune01 — Thu, 14 May 2026 19:25:08 +0000

using AI to red-team your thoughts and assumptions is the fastest way to get smart since the dawn of time

New comment by kimjune01 in "The US is winning the AI race where it matters most: commercialization"

kimjune01 — Thu, 14 May 2026 19:21:41 +0000

coopetition

New comment by kimjune01 in "Saying Goodbye to one line of APL"

kimjune01 — Thu, 14 May 2026 19:18:57 +0000

numpy is the APL reincarnate

New comment by kimjune01 in "Claude Account Suspended Seconds After Purchase?"

kimjune01 — Thu, 14 May 2026 19:18:18 +0000

I appealed one time, took 3 weeks to get my 30 dollars back. will never deal with that again

New comment by kimjune01 in "56% merge rate on 316 cold OSS PRs in a week"

kimjune01 — Thu, 14 May 2026 12:57:32 +0000

I've been running a public experiment on how far I can take agentic coding harnesses. When I ran out of my own stuff to work on, I looked to open source.

So far, I got the harness to create 368 PRs, 110 of which were resolved. Of those, I got a 56% merge rate.

The harness: https://github.com/kimjune01/sweep

But as soon as I realized what I was doing to maintainers, I worked all night to build the same pipe but backwards for PR review.

The defense: https://github.com/kimjune01/immune

Which means: bots on the contributor side, bots on the maintainer side. The surprise? What's left is two humans on either end. I accidentally invented a H2H protocol.

I'm happy to share what I learned.

56% merge rate on 316 cold OSS PRs in a week

kimjune01 — Thu, 14 May 2026 12:57:32 +0000

Article URL: https://github.com/kimjune01

Comments URL: https://news.ycombinator.com/item?id=48134744

Points: 3

# Comments: 2

New comment by kimjune01 in "Arena AI Model ELO History"

kimjune01 — Thu, 14 May 2026 07:23:10 +0000

Although Arena is adversarial and resistant to goodharting, it's not immune. Models that train on Arena converge on helpfulness, not necessarily truthiness

New comment by kimjune01 in "Software Developers Say AI Is Rotting Their Brains"

kimjune01 — Wed, 13 May 2026 14:21:49 +0000

are merged PRs a measure of velocity? github.com/kimjune01/

New comment by kimjune01 in "Why senior developers fail to communicate their expertise"

kimjune01 — Wed, 13 May 2026 14:20:40 +0000

we lack the shared data structure for it

New comment by kimjune01 in "Show HN: I submitted 316 AI-generated PRs to open source"

kimjune01 — Wed, 13 May 2026 07:55:33 +0000

the garbage gets smarter with each iteration

Show HN: I submitted 316 AI-generated PRs to open source

kimjune01 — Tue, 12 May 2026 18:12:49 +0000

Article URL: https://june.kim/speedrunning-open-source

Comments URL: https://news.ycombinator.com/item?id=48112050

Points: 6

# Comments: 2

New comment by kimjune01 in "SWE-bench Verified no longer measures frontier coding capabilities"

kimjune01 — Mon, 27 Apr 2026 01:34:04 +0000

AI labs should compete on a bench that's adversarial, such as go or Starcraft

New comment by kimjune01 in "Does internet advertising work? (2020)"

kimjune01 — Sun, 26 Apr 2026 04:09:17 +0000

internet ads are largely a story of parasitic exchanges using information asymmetry to suck as much surplus out of the middle as possible.

https://june.kim/advertising-journey/

New comment by kimjune01 in "Open source memory layer so any AI agent can do what Claude.ai and ChatGPT do"

kimjune01 — Sat, 25 Apr 2026 12:20:00 +0000

I agree. silent agent doing agentic things async is what would be helpful, not requiring a modification to the main prompt

New comment by kimjune01 in "Open source memory layer so any AI agent can do what Claude.ai and ChatGPT do"

kimjune01 — Sat, 25 Apr 2026 12:18:53 +0000

this is a patch on top of the broken flat-compaction caching algorithm used by coding agents. Why not fix the cache algorithm directly? Union-find is a better impl june.kim/union-find-compaction

New comment by kimjune01 in "Tell HN: Claude 4.7 is ignoring stop hooks"

kimjune01 — Sat, 25 Apr 2026 00:09:48 +0000

hi Thariq, I dont know how else to reach someone at claude code, so here goes:

I solved context compaction by using a better caching algorithm. It's being implemented in gemini-cli with limited success.

https://june.kim/union-find-compaction https://github.com/google-gemini/gemini-cli/pull/24736