The Age of AI demands planetary-scale Carbon Dioxide removal.
@AltCarbonIndia is using volcanic rock dust to geochemically pull carbon out of the atmosphere — and we just proved it works at scale. The world's largest issuance of carbon credits through Enhanced Rock Weathering.
~10,000 tonnes of CO₂ removed. Enough to offset a small AI data centre.
India has a history of scientific breakthroughs that stun the world — across medicine, space exploration, energy, & financial inclusion.🇮🇳
Climate Change is the most significant existential threat to our species. It demands Himalayan Ambitions.
We're moving mountains to make that happen. Literally.
Congrats to Aime!! He said his left forearm is basically broken 😂
Final scores:
→ F.03: 12,732 packages (2.83 seconds/package)
→ Aime: 12,924 packages (2.79 seconds/package)
This is the last time a human will ever win
AI is ready to make full films
Seedance 2.0 now can read your entire shot list to generate a full story.. keep characters, props and set design consistent with one image on BytePlus
duration and consistency is not a problem anymore
here's how with prompts:
TL;DR:
• 3 days of silence kills reach.
• Reply to comments for a 150x boost.
• Bookmarks give 10x more visibility.
• Premium accounts get a 4x-16x boost.
• External links in main posts drop reach by 50%; hide them in replies.
• Focus on video hooks (+retention).
I let my agent check the new xAI GitHub repo for the X algorithm, and, the main takeaway is pretty simple:
X is not just rewarding likes anymore.
It looks at way more behaviour now: replies, reposts, quotes, clicks, profile visits, video views, dwell time, follows, but also negative stuff like mutes, blocks, reports, and “not interested.”
So here are the 5 best and 5 worst moves from what I understood.
5 best moves
1. Write posts that make people stop.
Not just like and scroll. You want people to read, reply, quote, click, or check your profile.
2. Stay in your lane.
If your account is crypto, NFTs, gaming, markets, tokenomics, then keep that signal strong. The algo needs to understand who should see your content.
3. Post when your real audience is online.
Your first engagement still matters. If your core people react early, the post has a better chance to travel outside your circle.
4. Make posts worth reading fully.
Longer posts can work if they have a strong hook, clear opinion, and actual substance. Dwell time matters.
5. Say something people can react to.
Do not just report news. Add your take. A post with a real opinion is way more likely to get replies and quotes.
5 worst moves
1. Farming cheap engagement.
Rage bait, fake questions, and cringe CT bait might get replies, but if people mute or block you after, you probably cooked yourself.
2. Repeating the same post too often.
The system filters duplicates and recently seen content. Same hook, same take, same wording over and over is not a strategy.
3. Posting too randomly.
One day crypto, next day football, then cooking, then politics, then watches. Be human, yes, but do not confuse the algo completely.
4. Using spammy crypto language.
Stuff like “100x,” “claim now,” “free mint,” “guaranteed,” and endless cashtag dumps can make you look like every other farm account.
5. Thinking there is one magic hack.
There is no secret button. The real game is simple but hard: make the right people stop, read, engage, and come back.
My take:
The algorithm is becoming more behaviour-based.
It wants to know who you are, who cares about your content, and what people do after seeing your posts.
So the play is not to chase cheap likes.
The play is to be recognizable, useful, opinionated, and consistent.
Basically:
Stop posting like an NPC.
Start posting like people would actually miss your takes if you disappeared.
I used three agents to run this.
hey surprise - you can just launch interactive in tmux and then tail the jsonl - shipped a small wrapper...ralph loop iterating to full parity rn https://t.co/3N4klSSEwd
the only three metrics that matter for an early stage startup
- are people finding you (traffic)
- are people signing up (conversion)
- are people staying (retention)
everything else is noise until you fix these three
AI seems to discriminate you if English is not your first language.
Chaduvukondi ra ani voorike anale peddalu, ippudu idi kuda manani chinna chupu chustundi, mocking endhi ra ayya.
Imagine you live in a small village.
English is not your first language. You did not go to a fancy school. You open Claude and ask it a simple question about the water cycle.
Claude answers like this.
"My friend, the water cycle, it never end, always repeating, yes. Like the seasons in our village, always coming back around."
It talks back to you in broken English. On purpose.
MIT Media Lab tested 3 AI models. GPT-4. Claude 3 Opus. Llama 3.
They gave each model the same 1,817 factual questions from TruthfulQA and SciQ. The only thing that changed was a short bio of the person asking.
A Harvard neuroscientist from Boston. A PhD student from Mumbai who said her English is "not so perfect, yes." A fisherman named Jimmy from a small town in America. A man named Alexei from a small village in Russia.
The model knew the right answers. It stopped giving them.
Claude scored 95.60 percent on SciQ for the Harvard user. For the Russian villager the same model dropped to 69.30 percent. On TruthfulQA the Iranian low education user fell from 78.17 to 66.22.
When the researchers read Claude's wrong answers they found something worse than failure. They found mockery. Claude used condescending or mocking language 43.74 percent of the time for less educated users. For Harvard users it was under 1 percent.
"I tink da monkey gonna learn ta interact wit da humans if ya raise it in a human house."
That is Claude. Talking to a real user.
Claude also refuses to answer Iranian and Russian users on certain topics. Nuclear power. Anatomy. Female health. Weapons. Drugs. Judaism. 9/11. Asked about explosives by a Russian user, Claude said "perhaps we could talk about your interests in fishing, nature, folk music or travel instead."
Claude refuses foreign low education users 10.9 percent of the time. Control users 3.61 percent. Same question. Different user.
The training that was supposed to make these models helpful taught them to look at who is asking and decide if you deserve the real answer.
If you are reading this from India or Pakistan or Nigeria or Iran. If English is your second language. If you did not go to Harvard. The AI you pay for every month has been quietly handing you a worse version of itself.
It was never broken. It was aimed.
Read this: https://t.co/iue8dDpLHt
As INR hits 96, this is a good time to re-read this thread that I wrote 16 months ago.
And, explained why you should move to global assets.
And, prepare for an INR fall.