Pavan Anand
LightNews LightNews
banner
paanvaannd.bsky.social
Pavan Anand
@paanvaannd.bsky.social
880 followers 850 following 310 posts
🥼 MD (Family Medicine) & 🧑🏾‍💻 programmer in Riverside, CA.

Interested in almost everything under the Sun… and within and beyond it as well.

Views: mine, left-leaning, optimistic tinged with realism; repost ≠ endorsement; posts ≠ medical advice.
Posts Replies Media Videos
paanvaannd.bsky.social
I mean, S03E08 is one of the best episodes of television ever.

The story, cinematography, acting, its dark humor, optimistic ending, the emotions it evokes… I’ve watched it 4-5 times so far and I can’t help but tear up during it. Words can’t describe its beauty.
Elliot Alderson and Angela Moss, characters from Mr. Robot, sit with their backs to each other, separated by a door. Both are distraught and severely depressed, but they’ve just had an uplifting conversation that ends with them realizing that they’ll each be alright.
October 20, 2025 at 5:16 AM
paanvaannd.bsky.social
Best "contact me" website section I've seen in the wild yet.
Text of a website with a section header reading "Contact Me", below which appears "If you want to contact me, find a way to reach me."
July 26, 2025 at 8:54 PM
paanvaannd.bsky.social
Well, this is interesting...

Used ChatGPT to help me bootstrap my #Zsh + Zinit config.

o4-mini kept hallucinating; o3-pro got it right.

I switched back to o4-mini in the chat to say thanks.

o4-mini is confused since it thinks I'm still criticizing it.

It seems to not "see" o3-pro's response.
A chain of reasoning output by ChatGPT's o4-mini-high model in a chat that notes. The entire text cannot be transcribed as there is too much to fit within the character limit, but the essence is that the o4-mini-high model keeps trying to reconcile my statement of "This is correct; thank you so much!" with the prior message to o3-pro stating "You have been incorrect so far. Please try again." This causes the model a lot of confusion as it seems to think that I thanked it prior to o3-pro's response instead of after it. It therefore seems to be stuck resolving a seeming paradox where none exists.
July 6, 2025 at 3:56 AM
paanvaannd.bsky.social
Might even be 4x the crowds by some estimates!

encyclopedia.ushmm.org/content/en/a...
A screenshot of a paragraph within the Holocaust Encyclopedia website that reads: “Organizers went to great lengths to discourage appreciation of the artworks. Despite this, public attendance exceeded all expectations. It is estimated that more than 2 million people passed through the cramped space in 1937. By contrast, the Great German Art Exhibition around the corner was heavily promoted and held in a spacious new building. Still, it attracted fewer than 500,000 visitors.”
March 30, 2025 at 6:17 PM
paanvaannd.bsky.social
P.S., in response to this statement at the end of the Vulture article linked above: rightly so.

An artist's work takes on a life of its own after release. The public decides what it means.
A screenshot of the end of a paragraph in the Vulture article about Kendrick Lamar mentioned in post #3 of this thread. The highlighted section reads "fans will continue to ascribe political profundity to Kendrick's music whether or not it actually exists." The rest of it reads: "Last summer, when young protestors in Kenya mobilized to prevent a proposed tax hike on essential goods, the movement's anthem was a reworking of "Not Like Us."
February 9, 2025 at 11:38 PM
paanvaannd.bsky.social
Dell’s marketing team was paid millions to copy Apple’s device naming scheme and make it worse…

Their original brand name scheme wasn’t great but this is hilariously bad.
A PC Mag screenshot showing a table equating current Dell laptop brands to the new Dell laptop brand naming scheme. I can’t explain much further at this point without confusing you or myself. It makes no sense. In essence, no more Latitude, Inspiron, or Precision, only Dell, Dell Pro, and Dell Pro Max with a combination of Plus and Premium suffixes to those brands for different configurations.
January 7, 2025 at 4:14 AM
paanvaannd.bsky.social
Almost done with the new set already! At least those in the Common-Ultra Rare rarities.

… yes, I probably (definitely) spent too much Poké Gold on this already.
A collection of cards, in rows of 5 cards each, displaying Pokémon cards from the Mythical Island card set in Pokémon TCG Pocket. A highlighted portion of the image shows the text “62/68,” indicating the poster of this image owns cards representing 62 of the 68 total Pokémon cards in this set of rarity Common to Ultra Rare. A full-art Celebi ex card from the Pokémon TCG Pocket mobile game.
December 17, 2024 at 7:07 AM
paanvaannd.bsky.social
Lolwut my prayers were answered as soon as I posted this. Wasn’t working for nearly half an hour beforehand.

Thank Arceus!
A carousel of floating Mythical Island Pokémon TCG packs in the Pokémon TCG Pocket mobile game.
December 17, 2024 at 6:30 AM
paanvaannd.bsky.social
Pls #PokémonTCGPocket gods, just let me open some of the new packs 😭
Pokémon TCG Pocket screen showing an error, preventing me from opening one of the new booster packs just released a few minutes ago. The error dialog box is titled “Error.” The text in the box reads: “Error code: 102-170-014 An error has occurred.” 2 buttons are located side-by-side below this message. The left one says “Retry” and has a white background. The right one says “To Title Screen” on a turquoise color.
December 17, 2024 at 6:26 AM
paanvaannd.bsky.social
#AdventOfCode Day 3

Why use plain old descriptive function, variable, & type alias names when you can use fun ones like "gobbledygook" instead?

#Python
A screenshot of Python code from my Advent of Code 2024, Day 3, Puzzle 1 solution. The code snippet makes use of fun names: "make_sense_of" for the function name, "gobbledygook" as the parameter name, & "sanity" as the function output's type alias. The full text contained in the screenshot is as follows (including whitespaces for code indentation): def make_sense_of(gobbledygook: str) -> sanity: return tuple(re.findall(r"mul\(\d+,\d+\)", gobbledygook))
December 13, 2024 at 3:44 AM
paanvaannd.bsky.social
Algorithmic choice is chic.

There is Bluesky, of course. But I've noticed that @anthropic.com's #Claude allows you to select the style of response you'll receive instead of having to become an AI whisperer to coax specific response styles for each prompt.
A screenshot of Anthropic's Claude desktop application showing a dialog allowing the user to select the style of Claude's responses. The default style options are "Normal," "Concise," "Explanatory," & "Formal," all displayed on the left side of the dialog box under a label that asks "How should Claude write responses?" The "Concise" option is selected in this image, with a blue checkmark on the right side of the selected option. To the right of the default styles, there is a description of the selected response style, reading "Shorter responses & more messages." Beneath the 4 default style options sits a button stating "Create & Edit Styles." Beneath it, and underneath the style response popup dialog, lies the prompt textbox of the application that specifies that the model currently selected is "Claude Haiku" and a blue dropdown button next to the model name accompanied by a feather pen icon that reads "Concise," indicating that this is the current response style selected.
December 7, 2024 at 8:40 PM
paanvaannd.bsky.social
My opponent blundered his Queen… or did he? #chess
Screenshot of message from the Chess.com app. Friend (username blurred for privacy): “Deadly mistake” Me (username paanvaannd on the Chess.com app): “^ this is either a lament of an unforced error or a threat of impending destruction and I can’t decide which…”
November 28, 2024 at 6:17 AM
paanvaannd.bsky.social
Several months of work & 666 lines of code later, my TMS research project's #bash script is finally in MVP status 🎉

(Apt line count, given how much of a struggle it's been trying to get this right)

This would've been *much* easier in #Python

github.com/grakesh-lab/...
Screenshot of a terminal screen showing a command that sums the number of lines of code in a project. The output of the command is "666." The Bash code in the image is as follows: find . -name "*.sh" -type f -print \ | xargs -I {} wc -l {} \ | awk '{$1=$1};1' \ | cut -d " " -f 1 \ | awk '{sum += $1} END {print sum}'
November 21, 2024 at 4:54 AM