gklitt

5 ms·

1.
by gklitt 6mo ago
https://lord.io/spreadsheets/ is one of my favorite technical blog posts of all time. Highly recommend everyone check it out!
2.
by gklitt 11mo ago
Yes, I cite Brooks (and Harlan Mills, seemingly the original source of the idea) in the post!
3.
by gklitt 1y ago
Author here. I thought about including TCAS in the article but it felt too tangential… You’re right that there’s a voice alert. But TCAS also has a map of nearby planes which is much more “HUD”! So it’s a combo of both approaches. (Interest
4.
by gklitt 1y ago
Author here. Yes, I think the original GitHub Copilot autocomplete UI is (ironically) a good example of a HUD! Tab autocomplete just becomes part of your mental flow. Recent coding interfaces are all trending towards chat agents though. It’
5.
by gklitt 1y ago
Neat —- Scrappy looks like a lovely prototype! As the creators say in their writeup, it fits nicely into the lineage of HyperCard-style “media with optional scripting” editors, which provide a gentle slope into programming. In the section o
6.
by gklitt 1y ago
Actually, Patchwork has surprisingly few features! Think of it more like an OS than a product. The goal is a small set of composable primitives that let you build many things - documents, tools, branching/diffs, plugins… To answer your
7.
by gklitt 1y ago
> I agree, I feel like the authors are underestimating the effect the new AI is already having on the concept of local software crafting Coauthor here -- did you catch our section on AI? [1] We emphatically agree with you that AI is alre
8.
by gklitt 1y ago
Yes, I think atproto is a great example of the “shared data” pattern for composable tools! Especially since it handles public social scale, which is not addressed by the other systems we mention. AFAIK, atproto is primarily designed to supp
9.
by gklitt 1y ago
I gotta say I don’t understand your point about cooking — billions of people who aren’t professional chefs cook meals every day! These meals may not live up to restaurant standards but they have different virtues — like making it taste jus
10.
by gklitt 1y ago
Coauthor here. You make a fair point! Ease of use matters. We all want premade experiences some of the time. The problem is that even in those (perhaps rare!) cases where we want to tweak something, even a tiny thing, we’re out of luck. An
11.
by gklitt 1y ago
We're happy to share the Embark prototype with anyone who wants to try it out - just email me at geoffrey@inkandswitch.com and I can share a link with you. A couple reasons we've decided not to share the demo widely: 1) it's
12.
by gklitt 1y ago
I want to but I’ve been having trouble getting o3 to work - lots of errors related to model selection.
13.
by gklitt 1y ago
I tried one task head-to-head with Codex o4-mini vs Claude Code: writing documentation for a tricky area of a medium-sized codebase. Claude Code did great and wrote pretty decent docs. Codex didn't do well. It hallucinated a bunch of s
14.
by gklitt 1y ago
I just tried making it public, sorry!
15.
by gklitt 1y ago
yeah i actually do handle inbound email! just forgot to include that code in the shared version. the telegram inbound handler shows the rough pattern.
16.
by gklitt 2y ago
(post author here) Yes, it's absolutely great to have a "low floor" -- common use cases should be easy to do, without needing to learn a ton up front. But, hiding the structure is not the only way to achieve that! For example
17.
by gklitt 2y ago
This is outside my area of expertise, but the post sounds like it’s asking for “choreographic programming”, where you can write an algorithm in a single function while reasoning explicitly about how it gets distributed: https://e
18.
by gklitt 2y ago
We totally agree! Stay tuned for our next posts on that topic :)
19.
by gklitt 2y ago
Code is here! Totally support any riffing on this concept. https://github.com/geoffreylitt/wildcard
20.
by gklitt 2y ago
Creator of Wildcard here! Neat to see this pop up on HN after several years. As others have noted, this project isn't under active development anymore. One reason: while I still think there's great potential in browser extensions,
21.
100M Token Context Windows(magic.dev)
94 points by gklitt 2y ago|22 comments
22.
by gklitt 2y ago
Post author here. Just wanted to link to a talk that I gave which expands on the ideas in this essay and shows a couple evocative demos to make it more concrete: https://youtu.be/bJ3i4K3hefI?si=BCrqvyZiFFur3W9p
23.
by gklitt 3y ago
Post author here! I wrote this post five years ago. Since then, my conviction in the value of customizable software has only grown, but I've also updated my thinking in a few ways: 1) AI AI is rapidly getting better at coding. Current
24.
by gklitt 3y ago
I bought one of these for my house recently and it’s been great. We wrote custom software for our own info display, straightforward since you just serve images over an HTTP endpoint. I’m sure there are cheaper ways to achieve this result bu
25.
Automerge-Repo: A "batteries-included" toolkit for local-first applications(automerge.org)
219 points by gklitt 3y ago|43 comments
26.
Malleable Software in the Age of LLMs(geoffreylitt.com)
2 points by gklitt 3y ago|0 comments
27.
Humans Are Not Turing Machines(joshmpollock.com)
1 points by gklitt 4y ago|0 comments
28.
Fuzzy API composition: querying NBA stats with GPT-3, Statmuse and Langchain(geoffreylitt.com)
3 points by gklitt 4y ago|0 comments
29.
by gklitt 4y ago
FWIW, I also shipped Twemex on Firefox, but the userbase is comparatively tiny—-about 20x more users on Chrome (or Chromium-based browsers using the Chrome store) compared to Firefox. I’m excited about the ongoing push towards cross-browser
30.
by gklitt 4y ago
Ink & Switch has been doing work for a while on applying these same ideas of gradual enrichment to ink and tablets. Recent talk describing that work: https://www.youtube.com/watch?v=ifYuvgXZ108

More ›