Andrew Lee

Something that’s been on my mind lately: While population growth slows, the base of the age “pyramid” gets thinner. Young peoples’ power to make change through voting is diluted, in a sense. Young people need to understand that shifting dynamic and vote in higher percentages to compensate.

Big fan of this for Effect TS effect.kitlangton.com

Lovely to see what properly configured game theory incentives do to politics. Ranked choice voting ftw!

If you like beautiful apps and animations, you’ll like this. I love software as an art medium www.spottedinprod.com/blog/any-dis...

I need an explanation for how animejs.com was made, this is magical

The replies on here may not be as racist as Twitter, but they damn sure are hateful. Talk AI: FU, AI sucks go away Talk Business: Go away Talk Healthcare: Crickets. Engagement went from great convos on many topics, to agree with me or you are a nazi fascist We are forcing posts to X

I redesigned my personal website AND wrote a new blog post! It’s been more than a year since my last one. This one’s on the software development lifecycle. ✨ candrewlee.com/blog/2025-06...

I’m all for innovation and progress, but our economic system doesn’t give a safety net for those in now obsoleted careers. If it did, I think it’d be easier to root for tech everywhere. I don’t like having to be nuanced about new tech that also will make it harder to put food on the table for many 😅

The better genAI video models get, the more I think about misinformation, propaganda, and the information diet of society. LLMs unleashed misinformation at scale, but writing untruths was always possible yourself. Video models are so much more visceral, and enable impossibly realistic fake clips.

So much negativity here. For example, the comments on @nytimes.com posts are mostly garbage, and basically never related to the post content Putting on my tinfoil hat, I think there’s a lot of bots injecting political negativity here. Not sure if the goal is propaganda or just to make this site bad

Now that AI slop runs the world, I want a private web - a single restricted login for a network of sites. You’d only be able to register after the net’s “moderator” video calls you to verify you’re a real person. A counter to the Dead Internet.

Cook

Gemini 2.5 Pro + Deep Research has probably been my first truly game-changing experience with AI. I enjoyed Claude but it still wasn’t reliable enough. Gemini has been unbelievably good.

What a #severance finale! AGGHHHHHHHH. How am I supposed to wait for next season? 😔

Pretty sure this tweet is about the Luka - Davis trade (wild). The replies are so bizarre. Like LLM political rage bizarre. Are these real people?

Introducing Slow Roads 2.0 - procedurally-generated, scenic roads for endless driving zen. Built with @threejs.org to run in your browser - no installs, no ads, just roads. Play free at slowroads.io #threejs #webgame #javascript #procedural #webgl

OpenAI right now

This was fantastic. Definitely makes me wanna ship something whimsical. In no other time in history has it been so doable to make something fun in a few days that hundreds of thousands of people enjoy currently. Lucky to be alive in 2025 for that :)

I cannot express how strongly I do not care that deepseek R1 may have used openAI’s O3 outputs as data for distillation. That’s exactly what openAI did with our news, blogs, social media posts, videos, and more. Now that *their* data gets used by someone else, they want rule changes. 🙄

Wiz: DeepSeek left one of its critical databases exposed, leaking more than 1M records including system logs, user prompt submissions, and users' API keys (Wired) Main Link | Techmeme Permalink

Feeling pretty good about that prediction, #Nvidia is down 18% today after Deepseek spooked investors. The ironic part is that if better models release for free, doesn’t that incentivize consumers to buy Nvidia chips? Seems like Nvidia should be winning here, especially compared to LLM companies.

Remember that White House press release about avoiding C/C++ and preferring memory safe languages? The original report was at this url…and now it’s gone. Seems like C is back in style (always has been)

I could plot out the number of LinkedIn DM requests for referrals to positions at my company over time, and I bet it’d be a pretty good inverse proxy for the economy. Right now I’m getting a lot…

Let’s say something specific triggers an AI bubble pop. What would that event be?

This Trump meme coin crypto scam is one of the most shameless, embarrassing schemes I’ve ever seen. Meanwhile, TikTok gets taken away and then mentions Trump being the one who got it reinstated… that isn’t how laws work here. It’s difficult to look away from this dumpster fire. Holding on to hope

I see a ton of AI pessimism on here, and its understandable to a degree considering a lot of it feels like solutions in search of problems IMO. So here’s a question for self-described AI skeptics: Is there a use case for AI that excites you, either available now or conceivably in the near future?

Soooo when does the BlueSky business model crystallize so we know this site/protocol can be kept up?

A bug in iTerm2 causes ssh input and output to be logged to /tmp/framer.txt. Yikes :(

Mitchell Hashimoto is living the dev early retirement dream: - build a tool that meets your own high standards just for fun (ghostty) - organize a community of contributors and testers - release it publicly for free - fly off into the sunset with your family, piloting your own plane come on, man 🥵

There is a popular part of software dev twitter that is just loud, obnoxious people spewing engagement bait. It’s so immature. Unfortunately, there’s also a subset of insightful posts on good papers, performance advice, data structures, etc. which I struggle to find here. 😓

“Fish of Theseus” is one of the best blog titles I’ve heard in a long time. And it reads very genuinely. Some good, some bad, lots of fun overall. Most importantly, it’s a rewrite-it-in- #rust success story. The key was doing it incrementally (thus the title metaphor). fishshell.com/blog/rustport/

This is such a fun idea. I love all these new wrapped features in things this year :) Side note: I once fixed a bug in atuin and @ellie.wtf sent me a lovely little sticker with the atuin logo. Little gestures like that go a long way <3

Mitchell Hashimoto (of Hashicorp fame) just released the public 1.0 of Ghostty, the terminal emulator. ghostty.org Cannot recommend it enough. I’ve had a stellar experience with it as part of the beta group for the past year. It’s fantastic software.

I think Google is still in the best position to succeed in AI: - robust internet crawler for data - basically all of the top internet videos (Youtube) - massive amount of cash to burn I think based on their recent successes in Veo 2 and Gemini Pro, it’s clear they’re not gonna miss this train.

I think AI coding helpers push the needle even further towards Tailwind CSS. Locality of behavior/relevance is useful in general, but it’s especially important when using current LLMs IMO. When CSS styling is adjacent to its target, it reduces the outside context necessary, both to read & write.

Via Google’s quantum chip announcement (blog.google/technology/r...): “It lends credence to the notion that quantum computation occurs in many parallel universes, in line with the idea that we live in a multiverse” wow

Coding is one of my favorite things. Us young folks have a LONG career ahead, and it seems plausible that AI could overtake devs in that time. A “John Henry” moment to look for: when people like Fabrice Bellard struggle to keep up with AI… Coding may look very different in a couple decades.

> Devin launches > wants to replace junior devs > exposes a public vscode live share URL with access to your code > Devin creators have to issue an emergency security release Mission failed successfully?