iiinova — where your digital self keeps evolving

M @mick.bsky.social · 09/08/2025

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.

110

M @mick.bsky.social · 12/06/2025

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.

020

M @mick.bsky.social · 24/05/2025

TIL the original Vi editor (long before Vim) was on track for multiple windows … until a tape drive broke web.archive.org/web/20060701...

020

M @mick.bsky.social · 09/05/2025

New Pope used social media when it was still social

000

M @mick.bsky.social · 08/05/2025

Everyone’s an Einstein now

010

M @mick.bsky.social · 15/04/2025

Something I haven’t understood with these goofy AI images is why companies with thousands of engineers insist on purely relying on multimodal models. Instead, they could “do things that won’t scale”, and give the model access to tools like ImageMagick so things render as users expect.

100

M @mick.bsky.social · 04/04/2025

“(T)hink of our timelines as being like 2070-2100, it’s just that the last 50-70 years of that happen in the years 2027-2028.” www.dwarkesh.com/p/scott-daniel

000

M @mick.bsky.social · 28/03/2025

ChatGPT’s autobiographical comics hit hard

110

M @mick.bsky.social · 12/03/2025

Wow, that YouTube report on Honey practices kicked off a Chrome policy change (if we’re reading between the lines!)

120

M @mick.bsky.social · 08/03/2025

Tech debt bubble on the horizon is a human-machine race. Companies saddle themselves with mountains of code slop while there’s an exponential improvement in the AI that can tame it.

020

M @mick.bsky.social · 03/03/2025

Just landed on my ChatGPT (iOS): a button to read out the response in audio and another button to re-run the response with a different engine

000

M @mick.bsky.social · 17/02/2025

Protip for politicians and business leaders who just discovered podcasts and aping Trump: you don’t need your own podcast interviewing random celebs for a few months. It takes years to build an audience and most people won’t care anyway. You go on other people’s podcasts. That’s how Trump won.

000

M @mick.bsky.social · 17/02/2025

LLMs can make you smarter, but they can also act as the ultimate confirmation bias machine in the hands of a user who wants to feel smart without having their beliefs challenged. As the user below demonstrates.

131

M @mick.bsky.social · 17/02/2025

Bashtag is buzzword of the day

130

M @mick.bsky.social · 10/02/2025

OpenAI Superbowl’s ad:

110

M @mick.bsky.social · 06/02/2025

After 3 decades, PDF madness finally tamed thanks to LLMs. Machine-readable and mobile-friendly. Straight to markdown with that bs at the cost of pennies. www.sergey.fyi/articles/gem...

040

M @mick.bsky.social · 03/02/2025

A blind experiment comparing an AI essay (by DeepSeek, what else!) to a human essay. 60% correctly guessed which one was AI, not much more than a coin toss.

100

M @mick.bsky.social · 25/01/2025

ABC make up ~30%! chatgpt.com/share/679506...

100

M @mick.bsky.social · 21/01/2025

This is how much ChatGPT API pricing dropped in a year. You pay ~ 10-15% of what you paid a year ago for a superior model and much bigger context window (max: 8K vs 128K)

100

M @mick.bsky.social · 21/01/2025

How to read more books in 2025: s/three-hour “conversation” podcasts/audiobooks/g 🫡

000

M @mick.bsky.social · 20/01/2025

Apple validating the “stochastic parrot” crowd, who would otherwise be laughed out of the room by now

010

M @mick.bsky.social · 17/01/2025

In a world where customers can get upfront pricing on SENDING A PAYLOAD TO SPACE, you can probably get by providing upfront pricing for your SaaS app. (As pointed out on HN.)

010

M @mick.bsky.social · 13/01/2025

ChatGPT’s model menu has to be the most hostile-to-normies microcopy in a successful software product. Compounding the sin of bizarre terminology by not explaining any of it. (By design or just low priority?) Gemini’s menu shows how much difference a few words of explanation makes.

020

M @mick.bsky.social · 08/01/2025

They trick Apple and Google to send a real notification to all devices by calling from a spoofed phone number belonging to the victim. So they are comfortable the call is genuine (wrong). Still on phone, the agent then gets them to visit a fake site, where they enter their credentials.

120

M @mick.bsky.social · 08/01/2025

Phishing can be a sophisticated multi-role operation. Krebs outlines the modus operandi of the team that tricked Mark Cuban into losing a crypto stash, a musician having their creations leaked, and countless other victims.

100

M @mick.bsky.social · 04/01/2025

The above compared Gemini 1.5 Flash to o-1 since they are about the same speed. Gemini 1.5 Pro at least understood the question but didn't answer it by listing any examples (below). Gemini 2 can't go online, so is not something I'd use when ChatGPT is available.

000

M @mick.bsky.social · 04/01/2025

Gemini is daft and obtuse. I try it every few weeks and nope out.

100

M @mick.bsky.social · 24/12/2024

When ChatGPT landed, one of my predictions for an in-demand occupation was lawyer due to the enormous tangle of contracts that were about to be produced. The same opening now applies to senior developers who can tame AI code bases.

210

M @mick.bsky.social · 16/12/2024

Learning Vim has high ROI for knowledge workers since it can use it everywhere – IDEs, editors, command-line – even if the “Vim mode” replicas are less than perfect. news.ycombinator.com/item?id=4242...

010

M @mick.bsky.social · 16/12/2024

Not really excited about locked-down versions of tech that's been available from any web browser since 2022. Making Siri functional should be priorities 1 to 10.

000

M @mick.bsky.social · 05/12/2024

Trying PocketTube. With this many YouTube subs, I may need professional help or at least a helpful app 😅

010

M @mick.bsky.social · 02/12/2024

LLMs with autocomplete: another nail in the coffin for trad search.

020

M @mick.bsky.social · 02/12/2024

I love the idea of a Markdown-based protocol for data structures such as timelines. And stoked to see this one already ships with an Obsidian plugin. markwhen.com

010

M @mick.bsky.social · 06/07/2023

Live look at Twitter HQ

020

M @mick.bsky.social · 28/06/2023

The engineer in me always wonders if they took into account the resulting bugs

100

M @mick.bsky.social · 20/06/2023

Mastodon is its own worst enemy

100

M @mick.bsky.social · 10/06/2023

Some people feel programming in Python is a guilty pleasure” Me, a Ruby programmer:

000

M @mick.bsky.social · 08/06/2023

Good app! I found about 15% of the people I follow on Twitter here. Also nice that we can have app-specific passwords (ie API tokens) for Bluesky now.

030

M @mick.bsky.social · 06/06/2023

“Feels magic … as if your mind is controlling the experience” – BrainOS foreshadowing at WWDC23

000

M @mick.bsky.social · 05/06/2023

On soft forks, lessons from Chromium/Webkit/Blink glazkov.com/2023/05/06/stages-of-pr…

000

M @mick.bsky.social · 05/06/2023

iOS app spend is 0.75% of *global* GDP

000

M @mick.bsky.social · 30/05/2023

Chess analogy to coding with an AI copilot (“when the rubber duck talks back”). Makes the empirical case that those who don’t take advantage of this will be left behind. blog.jonudell.net/2023/05/24/when-t…

000

M @mick.bsky.social · 23/05/2023

Fintech fantasies versus reality 🔥 🔥🔥 www.bitsaboutmoney.com/archive/depo…

000

M @mick.bsky.social · 22/05/2023

Apple’s playing smart by restricting v1 to solo at-home use cases. Wearers won’t go outside and cop last decade’s “Glass-hole” stigma.

000

M @mick.bsky.social · 14/05/2023

Don’t want to be a downer, but prompt injection is unsolved, which makes many AI products vulnerable to attacks that can leak sensitive user data. @simonwillison.net points out that even OpenAI’s first official attempt to guide developers on this is inherently flawed.

110