What is Hackers' Pub?

Hackers' Pub is a place for software engineers to share their knowledge and experience with each other. It's also an ActivityPub-enabled social network, so you can follow your favorite hackers in the fediverse and get their latest posts in your feed.

When it comes to cooking, my spouse is a bit of a germophobe and tends to prefer a plastic cutting board that can be cleaned in the dishwasher. I am more of a microplasticphobe and thus prefer wooden cutting boards that I clean with hot water, soap, and a brush.

Whose phobia is more in line with the current state of science?

Edit: Thanks everyone, I got plenty of helpful replies! See thread below for a very insightful link.

0
0
0
0
0
0
0
0
0
0
0
0
0

If a trillion dollar corporation can effectively up-end a power user's life in an instant, with no recourse, imagine what they can do to an average user. I feel for this person, I really do.

But we as users need to stop implicity trusting these companies with our livelihoods and begin taking back control.

hey.paris/posts/appleid/

0
0
0
0

My LG TV’s new software update installed Microsoft Copilot, which cannot be deleted.
Weird that such a miracle technology can't be uninstalled. You'd think it's so good and useful that people would go out of their way to install it, even.

0
0

Sometimes the smallest details can make the most difference.

The new Loops For You algorithm requires the latest version of Redis with Bloom Filter support.

We care a lot about user (and admin) experience, and made it easy to enable this feature, while carefully validating the system requirements first.

New Loops admin dashboard feeds page showing the new For You algo toggle that is disabled because the system requirements are not metNew Loops admin dashboard feeds page showing the new For You algo toggle that is enabled and the algo is active
0
0
0
0
0
0
0
0

By now, everyone should know that when center-left political parties adopt right-wing policy positions, the right-wing voters they’re trying to appeal to will ignore the milquetoast version and vote for right-wing parties instead.

Reactionaries don’t want a watered-down version of reaction! They want the real deal!

“In contrast to much public debate, we find that more authoritarian/nationalist and more anti-EU positions are if anything associated with lower rather than greater electoral support for social democratic parties.”

tandfonline.com/doi/full/10.10

Something else that doesn’t get nearly as much attention as it should is the fact that left-coded political policies—higher minimum wages, stronger unions, higher taxes on the rich, more robust social safety nets—are actually really popular with the public in places like the US:

americanprogress.org/article/w

So when US Democrats or UK Labour tack right to try to capture vote share from their reactionary opponents, they’re making really, obviously foolish choices.

We’re left to conclude that these parties are either really, catastrophically bad at politics, their entire reason for being, or that their leaders genuinely prefer reactionary policies, want reactionary parties to win, and prefer support by right-wing voters over appealing to the left.

0
0
0
0
1
0
0
0
0
0
0
0
0
0

去年差不多這個時候

朋:「你有養寵物嗎?」

我:「腸道裡的益生菌算嗎?」

朋友們都不想理我,甚至不願跟我討論益生菌能不能算寵物,我明明都有餵養他們抗性澱粉跟膳食纖維。

0
0
0

: If someone says they accidentally reported you, it's a scam.

If someone says you need to verify your Mastodon account, it's a scam.

If someone says you need to change the email address on your account, it's a scam.

If someone tries to lure you off-platform to Discord or Telegram, it's a scam.

If someone jumps into your replies or DMs with a mutual aid request, it's a scam.

If someone from mastodon.social sends you an unsolicited DM, it's a scam.

---

Telltale signs of a scam:

- wants you to move off-site
- wants you to change account settings
- wants you to click something
- unsolicited private interactions
- new account
- under ~20 followers and over ~5:1 follow ratio
- sense of urgency
- appeals to your compassion
- piggybacks on a trending post or event

A screenshot of @Pink_m4n@mastodon.social trying to scam me with a "screenshot" saying they submitted a fraudulent report and they were able to take action. It tries to direct me to a Discord server for tech support.
0
0
0
0
0
0
1
0

Claude Code가 모델이 하지도 않은 말을 했다고 하는 이유.

자손킴 @jasonkim@hackers.pub

Claude Code에서 첫 번째 요청을 입력하면 가장 먼저 다음과 같은 JSON을 API로 보낸다. 이 요청은 실제 작업에 앞서 대화 주제를 파악하고 제목을 생성하기 위한 보조 요청이다.

{
  "model": "claude-haiku-4-5-20251001",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "Request Body의 구조를 분석하고 분류별로 묶어서 표현한다. ultrathink"
        }
      ]
    },
    {
      "role": "assistant",
      "content": [
        {
          "type": "text",
          "text": "{"
        }
      ]
    }
  ],
  "system": [
    {
      "type": "text",
      "text": "You are Claude Code, Anthropic's official CLI for Claude."
    },
    {
      "type": "text",
      "text": "Analyze if this message indicates a new conversation topic. If it does, extract a 2-3 word title that captures the new topic. Format your response as a JSON object with two fields: 'isNewTopic' (boolean) and 'title' (string, or null if isNewTopic is false). Only include these fields, no other text. ONLY generate the JSON object, no other text (eg. no markdown)."
    }
  ],
  "tools": [],
  "metadata": {
    "user_id": "user-id"
  },
  "max_tokens": 32000,
  "stream": true
}

시스템 프롬프트를 보면 이 요청이 신규 대화인지 판단하고, 신규 대화라면 2-3 단어의 제목을 추출하여 isNewTopictitle 필드로 구성된 JSON만 반환하라고 지시하고 있다.

여기서 내 눈에 띈 것은 첫 번째 요청임에도 불구하고 마치 멀티턴 대화가 진행된 것처럼 messages의 마지막 roleassistant라는 점이었다. 게다가 Claude가 { 한 글자만 응답한 것처럼 구성되어 있다.

이 요청에 대한 응답은 다음과 같다.

{
  "id": "msg_id",
  "type": "message",
  "role": "assistant",
  "model": "claude-haiku-4-5-20251001",
  "content": [
    {
      "type": "text",
      "text": "\n  \"isNewTopic\": true,\n  \"title\": \"Request Body Formatting\"\n}"
    }
  ],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 187,
    "output_tokens": 26,
    "cache_creation_input_tokens": 0,
    "cache_read_input_tokens": 0
  }
}

content.text를 보기좋게 정리해서 적으면 다음과 같다.

  "isNewTopic": true,
  "title": "Request Body Formatting"
}

완전한 JSON에서 맨 앞의 {가 빠진 형태다. 알고 보니 이것은 prefill 기법이라 불리는 것으로, 모델이 응답의 앞부분을 이미 출력한 것처럼 설정하여 이어지는 응답을 원하는 형식으로 유도하는 방법이다.

Claude Code는 이 기법을 활용해 모델이 JSON 형식으로 응답하도록 강제하고 있다. 단순히 "JSON으로 응답해줘"라고 요청하는 것보다 훨씬 확실한 방법이다. 모델 입장에서는 이미 {로 시작했으니 자연스럽게 JSON을 완성할 수밖에 없기 때문이다.

Prefill은 JSON 외에도 다양하게 활용할 수 있다. 예를 들어 ```python으로 시작하면 모델이 파이썬 코드 블록을 완성하게 되고, <analysis>로 시작하면 XML 형식의 응답을 유도할 수 있다.

Read more →
4
0
0
78
0
0
0
0
0
0
0
1
0
2

The ego of almost every software developer you follow is intimately compromised on this topic, which is why I overshare what I do about software development and AI. I try to present an obvious contrast where one of us must be off the mark and make you decide. You can’t really take this post I am linking to at face value unless you also think I am a rube, liar, and idiot. Which is fine, but you probably don’t think that.
mastodon.social/@nixCraft/1157

0
0