banner
mick.bsky.social
M
@mick.bsky.social
tech, just tech

Ex-Google devrel, O’Reilly author, coiner of Developer Experience
Pinned
M @mick.bsky.social · Dec 1
Made a list for no-bs tech news bsky.app/profile/did:...
@pepchasdrdave.bsky.social since you introduced phrases into the contranym universe: “Blow up” can mean rapidly expanding (like a celebrity or cultural trend or a startup co) but can also mean the same kinds of thing dramatically shattering into oblivion. (“Explode” also used both ways.)
November 17, 2025 at 2:43 AM
The anti-glazing model
Qwen3-4B-Thinking (a 4GB download) just became the first model I've tested to directly push back against my pelican riding a bicycle test, calling it "oddly specific and completely unrealistic" and saying "this request violates physics and biology"
simonwillison.net/2025/Aug/10/...
August 11, 2025 at 7:19 AM
“One click away” turns out to give you about a decade of runway before you have to lift a finger
One thing we underestimated in the early days of the web was how far “just a click away” is. People thought Google & Amazon couldn’t dominate because competitors were just a click away.

Rampant misinformation exists online despite the ability to factcheck on Google or Wikipedia is just a click away
August 11, 2025 at 7:17 AM
That regression test suite must be wild when users are in a romantic relationship with your code base
August 11, 2025 at 7:15 AM
Commemorating the CSS wars of great !important. Wherein a lone user fought for their natural right to customize style against the imposition of the prevailing theme-smiths. Here the user prevailed and the links did shine.
August 9, 2025 at 1:12 AM
Reposted by M
also good-weird neal.fun
Neal.fun
Games, visualizations, interactives and other weird stuff.
neal.fun
June 30, 2025 at 9:58 PM
TIL Benford’s Law. Reminds me of Deutsch’s Beginning of Infinity.
The answer starts with a 1. As most large numbers do.
en.wikipedia.org/wiki/Benford...
June 30, 2025 at 12:16 PM
Google flexing AI up and down the stack. Just as ChatGPT eats into search, TPUs and Gemini should be a concern to AWS/Azure.
Source: OpenAI recently began renting Google's TPUs to power ChatGPT, marking its first significant use of non-Nvidia chips; Meta also considered using TPUs (The Information)

Main Link | Techmeme Permalink
June 28, 2025 at 3:09 AM
Should exist in a single device: Router, USB charger, Smart TV (toaster optional)
June 22, 2025 at 3:55 PM
In 1999, you could raise multimillion dollar VC, the one simple trick was don’t turn on revenue yet. 2025 says hold my beer – no product in sight.
Sources: Mira Murati's Thinking Machines Lab raised $2B led by a16z at a $10B valuation; Murati will hold board voting rights that outweigh all other directors (Financial Times)

Main Link | Techmeme Permalink
June 21, 2025 at 9:15 AM
Via his new “Software 3.0” talk, I learned of Andrei Karpathy’s menugen.app, which generates visuals from menu photos – should be a game changer for dealing with foreign-language menus.

A fitting vibe-coding demo from the man who tweet-coined vibe coding.
MenuGen - AI Menu Image Generator
Upload a menu and get AI-generated images for each dish
menugen.app
June 20, 2025 at 4:18 AM
Every since ChatGPT, Google’s product velocity and willingness to disrupt its core product 🔥🔥🔥
Google rolls out the ability to have back-and-forth voice conversations with AI Mode, available in the Google app by tapping the Live button (Aisha Malik/TechCrunch)

Main Link | Techmeme Permalink
June 18, 2025 at 11:55 PM
Reposted by M
I wasn't aware there were that many scientists working on AWS Billing
My wife walked past my screen and lost it at the (completely accidental! I swear!) positioning of layered windows on my screen.
June 18, 2025 at 4:29 PM
Share with anyone who’s not aware of deepfake risks. These samples are open source generated so forget about any rails.

Zero-shot (ie claimed to not be cherry picked) and this demo page includes the few seconds of original voice that’s being faked.
chatterbox_demopage
resemble-ai.github.io
June 13, 2025 at 6:11 AM
2025 software repair is weird.

A recent ChatGPT bug that makes it unable to read iOS screenshots. A Redditor posted this (theoretically) permanent fix, which is to post the following text. It immediately worked for me.
June 12, 2025 at 4:27 AM
Spare a thought for the best looking and finest sounding media personalities, for they are the first wave of humans whose existence will soon be doubted
June 11, 2025 at 2:43 PM
ChatGPT sycophantic behaviour has not gone away despite OpenAI's claim. Now I have to constantly request external references to justify its responses that tell me how wise o so incredibly insightful I am.

Feels like it's playing shrink and merely echoing what I say in different words. Eliza++
June 11, 2025 at 1:49 PM
“If you had a machine that could instantly recall or create any facet of information, do you need a link?”
A link is all you need - aifoc.us/a-link-is-al...

I get uncomfortable when people say "the web will always win" or "don't bet against the web". While there are things that make the medium unique and special but with LLMs the nature of the link might change
LinkedIn
This link will take you to a page that’s not on LinkedIn
lnkd.in
June 9, 2025 at 2:22 PM
Reposted by M
I have been logged out of my toothbrush.
May 29, 2025 at 11:42 AM
AI force-multiplies the paradox of learning with technology. The better the technology, the dumber it makes students in the absence of careful guidance.
June 7, 2025 at 5:44 AM
An authentic “WFH cafe sounds simulator” would ship with a boomer screaming on a video call, an influencer live streaming their keto bowl, and 360-degree audio of those TikTok comedy feeds with zany sound effects.
June 4, 2025 at 2:50 AM
Nice summary dissecting the ignorance and motivated reasoning surrounding AI coding. It’s your job as an engineer to stay current and this field is moving fast.
June 3, 2025 at 3:04 AM
Threatening to leave your chatbot is the new moving to Canada
June 1, 2025 at 11:29 AM
Low hanging fruit for OpenAI is an iOS keyboard. I’m typing all day long into ChatGPT with constant typos that a personalised keyboard would largely prevent.
May 31, 2025 at 5:05 AM
Reposted by M
A white hat researcher used o3 to find a 0day. The internet is not nearly secure enough for our new commodity reasoning powers.

sean.heelan.io/2025/05/22/h...
How I used o3 to find CVE-2025-37899, a remote zeroday vulnerability in the Linux kernel’s SMB implementation
In this post I’ll show you how I found a zeroday vulnerability in the Linux kernel using OpenAI’s o3 model. I found the vulnerability with nothing more complicated than the o3 API &#821…
sean.heelan.io
May 30, 2025 at 9:46 PM