joel

it's been a while since I leaked something

@romainhuet prob runs a lot like Chrome 😅

Time to take Codex for a spin on the MacBook Neo!

@Dexerto a bit rich

The YouTuber also took issue with the amount of AI slop on the site dexerto.com/youtube/linust…

Linus Tech Tips has called out YouTube for adding 30-second unskippable ads to its TV app "When will it be enough? Are we just going to go all the way back to cable TV with three-minute ad breaks? Can we not?"

@bbjess smdh

Some one should send this pretty girl a switch 2 js

@Dexerto nice article 👍

His last four days on stream added up to more than $84,000 dexerto.com/kick/clavicula…

Kick streamer Clavicular revealed that he made around $20,000 daily during his "Mog World Order" subathon

10x harder to get reach on a brand account vs personal

@FearedBuck wow that’s crazy… he’s using OBS

Clavicular ends and walks out of his Channel 5 interview with Andrew Callaghan after Andrew reveals he’s satisfied with how he looks and doesn’t need looksmaxxing

@TheMatsuXL bottom right is the best but i would make it punchier: “DO NOT BUY THIS” makes you want to watch to figure out which camera it’s referring to imo

Which would you click?

Good news! The Windows driver team at AMD was able to track down the bug using our tool. It should be fixed starting with the driver update that went out in January. If you were affected please update drivers and re-run and let me know if the leak is resolved.

I was in the middle of a large refactor in our fork of Chromium and this was a big regression. Codex started using gpt-5.4 mini for reviews and agent expert consultations. Giant difference in Chromium where reasoning about process and thread boundary subtleties are critical. Fast model is fine for code exploration, but not reviews. Luckily I was able to ask it to use gpt-5.4 xhigh for reviews and it sticks decently well. Other issue I see is it gets impatient waiting for agents and tries to rush them to completion which renders partial review results.

GPT-5.4 mini matters for subagents because it changes what feels worth handing off. The parent thread should hold the architecture, plan, and progress narrative. Fast subagents can explore the repo, check hypotheses, and preserve the parent thread’s limited attention.

@heshie i like the chromatic aberration on the overhead lighting

waiting’s for the subway on film - nyc march 13, 2026

there is a way out of this where you can ultimately start setting your own price and terms. if you understand how businesses operate & their incentives it’s a relatively simple playbook. i’ve only seen maybe 2 or 3 creators do it well.

@pushingorphans @InvolveZero @bbjess @StreamWithMeld you’re not gonna get a switch with that attitude

@InvolveZero @bbjess @StreamWithMeld I mean i feel like if @joelcompiles sent more people switch 2s, they would have sooo many more people trying out meld but until then seems like people dont have a reason to switch.

Okay hear me out. If @bbjess made the switch what are you waiting for. I can only imagine how many sources and scenes she has with like 15 different camera angles 🤣 so get on the switch to @StreamWithMeld and make it happen you won’t regret it ps not a paid ad just a really happy creator who made the seamless switch!!

lil sus in chat

@quillpetv @StreamWithMeld nice! what did you find challenging about switching over? anything we can make better or more clear?

@StreamWithMeld Thank youuuu! It went pretty well. Need to figure out a few different sound things but over all great!

Live soon! I’m switching to @StreamWithMeld after OBS crashed on me 500 times yesterday. Wish me luck? Twitch.tv/quillpe RT for free monthly giveaway entry!

@ZKabooze @iSageSen @StreamWithMeld What kind of content are you making with Meld?

Holy Cow! Day one streaming with @StreamWithMeld has made content creation so much easier. Import OBS assets in 1 click and you get to clip vertical videos for content, without the hassel of installing plugins or editing. Definetly will be continuing to use this.

@matteocollina @travisennis @nodejs this is going to be fantastic to have built in support for. great work!

.@nodejs has always been about I/O. Streams, buffers, sockets, files. But there's a gap that has bugged me for years: you can't virtualize the filesystem. You can't import a module that only exists in memory. You can't bundle assets into a Single Executable without patching half the standard library. That changes now 👇

@Dexerto your website is so stuffed with ads it’s not even possible to read this hard hitting journalism

He ended up pulling in a surprising amount from the stunt dexerto.com/twitch/twitch-…

Twitch streamer ERBY was banned after broadcasting himself in a dog cage for hours with donation incentives Viewers could pay to keep him inside longer, with $10 adding an extra hour to the stream

selfies look better in meld

@cubesol_greg i've seen it enough times now to feel confident: i don't like the logo (my 2 cents ofc) it's like this mismatch where the UI has an awesome skeuomorphic look and feel but the icon is generic and low contrast. I'd go skeuomorphic on the icon!

Hot take: The devs complaining about “AI code quality” are the same ones who never ask AI to write requirements or build a plan. Garbage in, garbage out applies to humans too. Here’s the new onboarding flow for my “AI sLoP” DJ software 👇

@cubesol_greg @staysaasy skill specifically

@staysaasy I’m plowing through very complex software with Claude Code in record time. What problems are you having exactly?

I’m on week five of trying to vibe code a replacement for some dumb saas that we use and it’s so incredibly frustrating that I’m slowly realizing it’s actually a quite complex and thoughtful piece of software.