{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreibob5exjdwrrpvumhv4aixcz3y37vgwdzc6tm6hgvuynagxjy7kna",
    "uri": "at://did:plc:kbsenndul3hp5zagrexczjhs/app.bsky.feed.post/3mmghkvb5zab2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreicyewwgyye4v3eardw7wjtxrotg3vtq6lbzscxvqzdiips6nhav6q"
    },
    "mimeType": "image/png",
    "size": 64305
  },
  "path": "/notes/multitasking/",
  "publishedAt": "2026-05-22T02:20:00.000Z",
  "site": "https://v5.chriskrycho.com",
  "tags": [
    "Assumed audience"
  ],
  "textContent": "_It’s still terrible for us all._\n\n**Assumed audience:** People who are open to hearing an experience report from someone “in the wild” of professional software development with “agentic” coding tools. I am not talking _here_ about the ethics or legality of their training, but that doesn’t mean I don’t have carefully-considered opinions about those considerations; that just isn’t that post.\n\n* * *\n\nMultitasking has always been truly terrible for our ability to think clearly and work effectively. LLM-based “agents” don’t change that; they just offer one more invitation to distraction and split attention. Ignore the hype that says you must be running a multi-agent swarm: do good work instead.\n\n* * *\n\n_Some notes from my journal this morning, lightly expanded, along these same lines, prompted by finding myself as deeply muddled and distracted-feeling as I have been in a long time yesterday —_\n\nContext switching and _ways_ of working with LLMs that exacerbate context switching are not helping me be effective at work. Focus, using “old-fashioned” techniques like closing other windows, using a pomodoro timer, and explicitly prioritizing goals helps.\n\nOn LLMs: it strikes me that their relatively slow response is part of what makes it so easy to get into context-switching brain-fried mode. That 30-second pause, that 3-minute wait: it makes it _so_ easy to open Slack or go start reading a document. Even if there are some messages that legitimately warrant a response, even if someone is waiting for me to read and comment on the document, that switch is costly. My focus on the problem at hand is gone, and the likelihood that I will just keep switching after that only increases.\n\n_And, after a morning of making some conscious choices to work the way I know works best:_\n\n  1. Closing all the other windows besides those expressly necessary for the task I am working on and explicitly single-tasking helps. Of course it does! We all know it does!\n\n  2. Multi-window support in Slack leads to too-many-chats-at-once, exacerbating the tendency to split attention.\n\n\n\n\nAnd: focusing and shipping things and using my brain instead of “delegating” to Claude Code or Codex etc. just plain _feels_ better… but it’s also more _effective_.\n\n* * *\n\nThanks for reading my feed! Thoughts, comments, or questions? Shoot me an email!\n  *[LLM]: large language model",
  "title": "[notes] Multitasking",
  "updatedAt": "2026-05-22T02:20:00.000Z"
}