Shayon Pal

Played #Hades II on the #SteamDeck for the very first time today. It was absolutely amazing. 90 minutes well spent.

Going to be on a month-long trip soon. Great chance to get back to the #SteamDeck. Making sure all games are updated 😛

#Linux used to be my OS of choice for many years, before I moved to macOS some time around 2010. 15 years later, I am tempted to taste the grass on the other side once again. Getting really taken in by the Garuda distro.
garudalinux.org

Yet another week. Sigh! #ClaudeCode #ClaudeAI

Nice! First snow of the season #Toronto

It's a shame that Discord has now become an ad galore.

Well, in this case, you won't technically be coding. You'd be messaging with someone else who's coding.

The Comet browser from Perplexity has now become my discount code hunter whenever I am shopping online 🤣

Sheesh! Watching Loot and can't shake off the fact that Michaela Jaé Rodriguez is hot! 🌶️

I'd love to know a comparison study of % of people who commit suicide after going to a human psychologist v/s & of people who commit suicide after seeking therapy or therapy-adjacent advice from a LLM. I think it will be a very illustrative metric.

Definition of what's a "waste" is extremely subjective. You're welcome to your opinions. But it's also important to acknowledge that they are just that, opinions. Not facts.

it is nothing but a tool. Like any other. Learn how to use a knife, and one can make beautiful dishes. They don't and can land up in the ER in 5 minutes.

That's what I was hoping to understand. What kind of "ease of use"? What makes it better/different than tmux? 🙂

Holy shit! That sounds awesome. Well, barring the non-functional computer camera part 🤷🏻‍♂️
If you don't mind me asking, work-wise, what was your trade?

Ghostty + tmux is my jam. Sometimes, I've found zellij to lag a bit, especially when I would be mosh-ing into my dev machine from the iPad.

Nevertheless, what is it about zellij that identified with you more than tmux?

Also, (I think) zellij doesn't support saving sessions across restarts.

Hahaha! I tried my best to gauge how I could help 😛

Their goal is to build a GUI for the configs. It's just not prioritised as yet. Just about a months back, they introduced the `⌘ + ⇧ + P` command pallette.

Hopefully this will explain the world of "terminals" a bit better
www.perplexity.ai/page/ghostty...

JetBrainsMonoNerdFont on Ghostty terminal looks absolutely fire!

What a World Series this has been! Blue Jays have been legendary. Really rooting for them to win. Fingers crossed!

I agree. It absolutely screws up my Blink shell (on the iPad) with random artifacts.

Also, you're `assuming` that "facts" generated by AI will have a high probability of being wrong. No one is 100% correct, not even humans. It's all a game of probability, and it IS possible to increase the accuracy of "facts" as close to 100% as considered acceptable.

There are scenarios every where. I just took up an example, using the case that you shared something about Canva.

If I had some more context about you, I could've come up with scenarios where it could be helpful for you too 🙂

Happy - Claude Code Mobile Client Free, open-source mobile app for Claude Code. Control Claude AI from your phone with end-to-end encryption and seamless workflow.

Interesting. How is it different than claude.ai/code?
Also, my mobile tools of choice, at this point, are either `mosh` into my Mac Mini server using blink.sh or just launch and start typing in the happy.engineering iOS app.

Usually use Blink on the iPad, and Happy on the iPhone.

@daraddishman.bsky.social how I wish my browser's inherent spell check could easily warn me of these stupid typos mine 🤷🏻‍♂️

- Using pre-decided metric logic, discard poor performing ones
- Write new copies
- Publish and test again
- Keep iterating on this until desired success metric is reached

It can be really fun to discover what finally manages to achieve the desired success metric 🙂

Not just that, there is a concept of "AI Agents" that's orchestrable and can automatically do these and monitor results:
- Write 2-3 versions of copy
- Publish them on the website (just taking one use example use case here)
- Monitor performance

...contd

Also, to be able to use "robots" to give good output is also a skill in itself. Part of that skill is "Prompt Engineering".

Just a different skil than you (@djdynamic.ca) currently probably have. And that's perfectly okay too. Not everyone can or has to be great at everything.

Pardon my tardy typing. I meant "display ad", not "displad od". I wish Bluesky allowed post editing.

Let's say I am testing copy for a displad od on the web. I'd create a couple of variations of those and then run a multivariate test to check for:
- CTR
- CPM
- Revenue per 1M impressions
- Bounce Rate

That is what will decide which copy wins. I'd do the same for design ideas as well.