{"blyg":"0.4","id":"3fmhazg162snwcayffpp18ex2h","kind":"thread","origin":"https://blyg.aneeshsathe.com/","page":"t/3fmhazg162snwcayffpp18ex2h/","author":{"name":"Aneesh Sathe","url":"https://blyg.aneeshsathe.com/"},"created":"2026-11-15T22:34:40Z","updated":"2026-11-15T22:34:41Z","version":1,"content_md":"\nThe catchup functionality is now there on Burrow\n\n\n\n![](https://blyg.aneeshsathe.com/media/9hfa7f3m/dfa0f5639e83a30a21ca99fc7eea4ba28c63caa55f66b3b61a2f324c569c6fc4.png)\n\n\n![[76ya3cbedkm6h8zpbszg6xdn45 | Will see how Burrow can support this. Some nice affordances in the desktop mode:]]\n\n","content_html":"<p>The catchup functionality is now there on Burrow</p>\n<p><img src=\"https://blyg.aneeshsathe.com/media/9hfa7f3m/dfa0f5639e83a30a21ca99fc7eea4ba28c63caa55f66b3b61a2f324c569c6fc4.png\" alt=\"\"></p>\n\n<blockquote class=\"blyg-transclusion\" data-blyg-id=\"76ya3cbedkm6h8zpbszg6xdn45\" data-blyg-version=\"2\" data-blyg-label=\"Will see how Burrow can support this. Some nice affordances in the desktop mode:\">\n<p>Will see how Burrow can support this. Some nice affordances in the desktop mode:</p>\n<ul>\n<li>you can use your local AI login instead of api costs</li>\n<li>access to your notes in case there are items that relate to what you have worked on</li>\n<li>the integrated terminal would allow you to directly chat with your AI to continue the chat</li>\n</ul>\n<p>will first just build the catch up function, the rest remain as available surfaces based on use and ergonomics</p>\n\n<blockquote class=\"blyg-transclusion\" data-blyg-id=\"7evenwcss6s63p5tvcq9rzk50y\" data-blyg-version=\"1\" data-blyg-origin=\"https://venkateshrao.com/blyg/\">\n<blockquote class=\"blyg-transclusion\" data-blyg-id=\"6bd3qqywdyzz41k8ax4eb93tmv\" data-blyg-version=\"1\" data-blyg-origin=\"https://blyg.florianlohse.com/\">\n<blockquote class=\"blyg-transclusion\" data-blyg-id=\"6s84fnfkzexwv4h056xeqayac7\" data-blyg-version=\"1\" data-blyg-origin=\"https://blyg.artlu.xyz/\" data-blyg-label=\"artlu99&#39;s blyg: If you stay too large the concept/projects are the same size as you and it’s too mu…\">\n<blockquote class=\"blyg-transclusion\" data-blyg-id=\"014x8872vkcpr5nt1kkqp7h3e9\" data-blyg-version=\"2\" data-blyg-origin=\"https://blyg.florianlohse.com/\">\n<blockquote class=\"blyg-transclusion blyg-partial\" data-blyg-id=\"4kwy37c6qqxkdtvyaqca2vavfe\" data-blyg-version=\"1\" data-blyg-origin=\"https://blyg.aneeshsathe.com/\">\n<p>If you stay too large the concept/projects are the same size as you and it’s too much to deal with. Better to be much smaller and become enchanted by the little particular thing.</p>\n</blockquote>\n<p>This is important. It’s tempting to drop all caution commonly associated with building complex things, because &quot;AI will handle the complexity for me&quot;. But this will just get you stuck in deeper puddles of mud than the ones you frequently got stuck in before Claude holding your hand.</p>\n<p>Finding the right size for a project or a workpackage, aka scoping is essential, maybe the most important human contribution next to contributing taste, opinions and irrationality.</p>\n\n</blockquote>\n<p>I also think this important, and maybe the question of our time.</p>\n<p>How to do this (i.e., pursue elegance, beauty, craft) yet not get caught by the tyranny of small ambitions. It might be more costly to think too small, than to think too big.</p>\n<p>Like, how do you ski a black diamond run when there's a big open flat space ahead of you?</p>\n\n</blockquote>\n<p>Yeah, Scoping is essential, but the constraints and metrics around it have changed. Making sure delivery arrives in small testable and shippable increments used to be hard, bc many devs if left alone for too long would build idiosyncratic spaghetti code rabbit holes around them and then everyone got stuck in trench wars over conflicting details. Now Coding Agents gravitate more towards incrementally structured approach by default (at least in my experience) so it feels as if a more meaningful human contribution is pushing towards more risk and reward at the right moments - occasionally going for the black diamond route because it is worth it.</p>\n\n</blockquote>\n<p>UX not to self. I read this by scanning to get a vague sense of the conversation, then clicking into innermost quote, out again, then second-innermost, then back out again, and finally the third layer.</p>\n<p>While the full 4-layer post was intelligible in the sense that you might get a sense of what a group of people are talking about as you walk past in a party (which may be all you want), to actually join the conversation you have to kinda backtrack and untangle in the right order. But thus feels like an AI primitive. A “catch-up” button that reads and summarizes in the right order.</p>\n<p>Unlike live conversation where transcription bots deal with text that’s already serialized, written entangled conversation comes at you arbitrarily fragmented and out of order, since partial out-of-order quotes are possible. So the jigsaw puzzle has to be assembled.</p>\n<p>This is a kind of “conversation graph microscope” thing.</p>\n\n</blockquote>\n\n</blockquote>\n","content_hash":"sha256:c859d1ad1ac844c1d4e512e80ce2506d202ecc7cb7d522293b70f8168c874996","media":[],"transclusions":[{"id":"76ya3cbedkm6h8zpbszg6xdn45","version":2}],"changelog":[{"version":1,"at":"2026-11-15T22:34:41Z","note":null}]}