Gravitational Dipole Self-Induced Motion ✍️
What if you could create a machine that moves forever without any fuel, engine, or outside push? Imagine two objects connected in a way that they launch themselves through space and never stop accelerating. This may sound like science fiction, but it is actually a serious theoretical physics idea with real mathematical support. It starts with one of the most puzzling concepts in science the possibility that negative mass exists.
To grasp negative mass, set aside everything your intuition tells you, because negative mass contradicts every lesson you've learned about how matter behaves. Normal mass attracts other masses through gravity and moves in the direction you push it. Negative mass does the exact opposite. It repels positive mass rather than attracting it. When you push it to the right, it accelerates to the left. If you push harder, it moves away faster in the wrong direction. Negative mass has negative inertia its resistance to motion works entirely backward. This idea goes against our everyday experiences and feels so strange that it seems the universe wouldn’t allow it. Yet, the mathematics of Einstein's general relativity handles negative mass without contradictions. The equations function correctly. In quantum physics, negative energy closely related to negative mass has been shown in laboratories. So, while negative mass hasn’t been observed directly, it hasn’t been ruled out by physics either.
Now things get really exciting. Take a positive mass and a negative mass, connect them at a fixed distance like two balls on a rod, and see what happens when they interact gravitationally. The positive mass gets pushed away from the negative mass it is repelled because the negative mass has reversed gravitational charge. The positive mass accelerates away to the right, while the negative mass is attracted toward the positive mass, also moving to the right. However, because the negative mass has negative inertia, pulling it right causes it to go right as well. Analyzing the physics shows that both masses end up accelerating in the same direction together. The negative mass chases the positive mass, which is simultaneously pushed away by the negative mass, propelling the whole system through space like a runaway train without a track or engine. The puzzling part for physicists is that when you sum all the inertial forces acting on the system, they balance out to zero. The system has no net inertial force. Yet it accelerates. Something is moving it, but it isn’t a direct force between the masses. It’s something much subtler.
The key lies in a part of gravity that most people don’t learn about. Einstein's general relativity shows that gravity isn’t just a simple attractive force between masses. Moving masses create what is called a gravitomagnetic field a gravitational analogue of magnetism, produced by mass in motion, just like magnetism is created by electric charges in motion. This isn’t just fringe theory; it is a confirmed prediction of general relativity. NASA launched a satellite named Gravity Probe B specifically to measure the gravitomagnetic field created by the rotating Earth, and the results confirmed Einstein's predictions with amazing precision. So, gravitomagnetism is real. When the positive-negative mass pair rotates and spins like a propeller through space, it creates a gravitomagnetic field around it, filling the surrounding space with a structured pattern of flowing field lines, resembling an invisible magnetic bubble wrapping around the spinning pair.
Inside this gravitomagnetic field, there is a property called the torsion field, which measures how much the system's gravitational influence is twisting and rotating through space at any point think of it as measuring the spin of a gravitational whirlpool around the system. As the dipole moves, this torsion field changes over time, continuously pulsing. Now we come to one of the most beautiful ideas in physics.
We built real-time tracing directly into Firedancer. 📊
See shreds, votes, and every single transaction at 10ns accuracy: https://t.co/yepReQoQ2j IMO the coolest feature this release, and wild tech to enable it.
Your first $250k need not come from a16z, Sequoia, or YC...
10 pre-seed investors writing $100k -$500k cheques:
(Thank me later if you close your round 😌)
1. Sarah Smith Fund - solo GP, $16M Fund I, typical cheque $250k . Backs "irrationally intense" founder CEOs (Menlo Park)
2. Script Capital (@evantana) - $250k - $500k, leads pre-seed rounds under $1M for technical founders (SF)
3. Redbud VC (@redbudvc) - $250k - $500k into "outsider" founders. Closed a $25M Fund II in Mar, 5x its first fund
4. Incisive Ventures (@MartinGTobias) - $100k - $500K into B2B software from a $10.5M debut fund (Seattle)
5. Two Ravens - $20M fund, $250k - $500k into founders from non-obvious backgrounds (Seattle)
I promise you won't regret turning notifications on for @arorabhavyam if you like content around founder programs, investors and AI! 😌
6. Argon Ventures - $250k -$500k first cheques, leads pre-seed. Run by Brightcove's ex-CEO and co-founder (Cambridge, MA)
7. Homebrew (@hunterwalk) - $100k - $500k, now funded entirely by the two GPs' own money. No LPs to please
8. Graham & Walker (@LeslieFeinzaig) - $100k -$400k into underestimated founders, $15M Fund II (Seattle)
9. Darkmode Ventures (@ashevat) - $150k -$250k into AI + dev platforms, from Twitter's former developer platform head
10. Marketplace Capital - $100k - $250k, marketplaces only, from the founder who sold GetMeIn to Ticketmaster
10 more (smaller cheques you can stack) + a sheet with all 20 coming in the quote tweet. Keep an eye if you want! 😉
ALL19 APPS ARE STILL OPEN FOR 2 MORE DAYS
• $400K in funding
• Build in Soho, NYC
• Pitch tier-1 VCs at Demo Day
• A lifelong network of Alliance founders
Apply by November 18: https://t.co/AlbmnHKI34
Bayes’ Rule is the ultimate mathematical cheat code for upgrading your opinions with actual evidence instead of just stubbornly doubling down!😊
https://t.co/BnqFrQjZrE
someone just walked out of @Lighter_xyz with $15.5m
no smart contract bug. no flash loan. just 2 leaked api keys
> deposit $10k from a fresh wallet
> open a 0% fee pool from the victim's account
> pour their usdc in, $1m at a time
> redeem, withdraw 10,393,056 USDC, swap to ETH
two market makers got drained in under an hour. $LIT -16%
if you run a bot on lighter, rotate your api keys before you finish reading this
Meet @Moshaikh
The guy behind Aptos, a $20B Scam.
He Co-founded Aptos, raised $400M from major VCs, watched $APT holders lose over -99% of their value, and stepped down as CEO right after his token compensation vested.
His explanation? A much-needed “mental health break.”
Fast forward to 2025: launches his own $50M crypto VC fund, then casually tweets:
“I have a nice $SUI bag now ;-)”
Yes, SUI. Aptos’ direct competitor, built using the same Diem technology.
So the guy who raised hundreds of millions building Aptos, collected his compensation, walked away, started a VC fund, and bought into the competition while $APT holders were left holding the bag.
1/ Solana transactions that write the same account must run serially, so block replay waits on that chain while other threads sit idle.
Much of that waiting is behind transactions that end up failing...
With FRODO, an idea we're testing in Mithril, we cut a lot of that wait.
The 1847 Equation That Trains Every Neural Network
θ ← θ − η∇L(θ), written by Augustin-Louis Cauchy to solve the equations of astronomy.
L is the loss, a landscape of how wrong the model is, and θ is where the model stands on it.
The gradient ∇L points uphill, toward the steepest rise, so the minus sign turns the model around and sends it downhill (red arrow).
η, the learning rate, is the length of each step: too small and the ball crawls (left), too large and it overshoots the valley and flies off (right).
A real network walks this landscape in billions of dimensions at once, using a small random batch of data to estimate the slope at every step.
Language models, AlphaFold and image generators all learned by walking downhill on a 179-year-old rule, one tiny step at a time.