Trying desperately to put myself down for a power nap. Read this title. Read the story.. and there it was- another reminder that even when our older generations invented supernatural villains, the woman got the witch job.
Bye bye power nap 😒
AI doom predictions feel oddly linear: AI takes jobs-consumption falls-capitalism breaks. But economics also taught us that what gets sidelined as the “error term” can have outsized impact. Human irrationality for one.
Are we again underestimating variables we can’t model?
The election commission is striking off nearly one in five voters off the Karnataka's voter-list.
If you are one of them and don't act now, you risk losing your right to vote!
Takes only a few seconds to find out if your name appears in the revised voter-list or the deletion list. Check @ https://t.co/52uGCcnX7b
#SIR @DKShivakumar@PriyankKharge
As a consequence of the ongoing SIR exercise, has your name been deleted from the voter list?
You would think that would be fairy easy to figure it out.
You would be very wrong.
You would have to log on CEO Karnataka website, then find the Google drive link to your district, then go to your assembly folder. And then from a list if numbered PDFs of each booth, try and locate the PDF for your booth. Open the PDF and try and locate your name in it.
To make all our lives easier, I created this site to search much more intuitively using using your name or voter ID. Do share your feedback.
Note: This data is as on 6th August 2026
Disclaimer: Use at your own discretion. To ensure accuracy, check the CEO Karnataka ASDDO page and the ECI electoral search. Your ERO can also confirm the current official status.
https://t.co/6wRlQSow1o
"He" implies all, we were told.
Does it though?
Sitting through the education bill debate in the parliament, where the youth of this country have apparently been assigned one gender by grammar alone.
Peaceful protest mein mangvaye gaye arms, tear gas, lathi aur eent,
Ahinsa ka jawab jab ho jaye hinsa, chup rehna nahi hai theek.
Modi-media ki godd se utho, Instagram khol kar bhi dekh lo yaar,
Chhod kar blind-faith, poocho sadkon par kya chal raha hai is baar!
Sawal GenZ ka, jawab lathi ka
Jo samajh na aayein tumhein, woh nasamajh thehre,
Jo tumhare reet pe na chalein, woh bhatki hui lehre.
Jo sambhalta nahi tumse, un par bitha diye pehre,
Gumrah yeh Gen Z nahi, gumrah toh tum ho pyarey.
🧵
Rajasthan rally mein bola har leak pe maango istifa, sarkar hai weak,
Ab Centre mein hi ghapla, toh kyun ho gaye Vishwaguru dheent?
Das saal mein paper leak ko saat guna badne diya,
Dharam ke narrative se desh ka literacy rate bhi gira diya.
🧵
Ab ki baar, tumhare puraane formulas bilkul hain fail,
Yeh na jaat ka mudda hai, na dharam ka khel.
Na Hindu-Muslim ka card chalega, na 'Pakistani' bol paoge,
Padhayi ke sawaal par, kab tak 'Anti-National' ka thappa lagaoge?
🧵
I wrote it up as a visual essay in six panels:
https://t.co/Kf6WKwd5QA
Covers the methodology, both outputs side-by-side, three evaluation layers (ChatGPT, Claude
self-eval, human researcher), and what separates coverage from insight.
Codifying how you think is harder than thinking. I ran a controlled experiment to find out what happens when you encode human judgment into an AI prompt vs. leaving it out.
The outputs looked structurally identical. The rigour wasn't even close. ↓
The source material was @lennysan's podcast with Jenny Wen (Head of Design at @AnthropicAI's
Claude).
The experiment was run on Claude Opus. Both conditions analyzed the same conversation about why traditional design process is "basically dead."
The full visual essay ↓
Here is what I didn't expect:
The prompt-only version produced more creative synthesis. The judgment-encoded version produced more rigorous analysis.
Coverage ≠ insight. Structure ≠ rigour.
They are different tools. Most people only build the first one.
The results:
Prompt-only: 67% signal ratio, 12 clusters, better original synthesis
Prompt + judgment layer: 88% signal ratio, 0 hallucination, tensions
preserved instead of dissolved
The gap between two AI outputs isn't the model. It's the judgment you encoded - or didn't.