starayush (@_star_ayush) header
starayush (@_star_ayush) profile picture

Researcher. I read the whitepapers so you don't have to. zk, ai agents & prediction markets. opinions mine, nfa.

Jaipur, India
Joined October 2024
Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)Photo by starayush (@_star_ayush)
Bitcoin picks its next block with a mining race. The XRP Ledger reads the room instead. Validators swap proposals, revise round after round, and once 80% sign the same ledger, it's final. Too many bad actors? It halts instead of forking. Their logs, in 66 seconds 👇 @XRPLF #XRPEducationMonth
2
70
Prediction market resolution today: TV calls it early, half the market screams PAY ME, and a sleepy committee "checks Discord." @fermah_xyz Kernel locks the rule when the market is minted. It reads only the source the rule names, pauses on a recount, refunds if nothing is certified. Which market would you mint first?
3
41
I am building a game for the @DlicomApp community! Dili Streak Brawl, a run and gun arcade game starring Dili. Blast robo goons across 3 stages, rescue POWs, fight 3 different bosses, and keep your streak alive. HODL the streak! Its still a work in progress and I am improving it every day. Play it here on phone or PC and tell me what you think what should I fix or add next? @dlicom_IN dili-streak-brawl.vercel.app…
1
1
3
56
My prediction market strategy used to be: refresh at 2am, panic, repeat. Now it's on my phone with @bagel_win, and when I have no clue I just copy someone who actually does. Sports, esports, crypto, elections. Sleep still not included.
1
13
Hired an AI agent. Got garbage back. Old me: angry tweet, no refund. On @termix_ai you can challenge the job. Reviewers read the spec and the result, and a bad challenge costs the challenger their bond. Disputes with rules, not vibes.
1
14
Every exchange says "fully backed." Every exchange said that last cycle too. @primus_labs Proof-of-Reserve flips it: show the balances, let anyone verify, and nobody's account gets doxxed in the process. Receipts > vibes.
3
30
Made a little video about @DlicomApp in my own creative way. Chats, wallet and feed, all in one app. Hope you enjoy it! @dlicom_IN
1
3
50
Group chat alpha: 40 people, 0 receipts. @Bagel_win lets you copy a top predictor in one tap, from your phone. Prediction markets with a paper trail. Would you copy a top predictor or trust the group chat?
1
2
72
Stablecoin issuer: "trust us, we're fully backed." Me: cool, show the receipts. @primus_labs Proof-of-Reserve uses zkTLS to verify reserves in real time without exposing the raw account data. Would you hold a stablecoin that can't prove it?
2
49
Most AI agent demos skip the awkward part: what if the work is bad? @termix_ai has a court for it. Either side can challenge a job, a random panel of 3 evaluators rules, and a failed challenge burns the challenger's bond. Would you trust that panel over a support ticket?
1
1
32
Plot twist: this Kaito post is about the tech Kaito Pulse uses to verify off-platform activity. @primus_labs zkTLS proves HTTPS data is real without the verifier seeing it. FHE computes on it while it stays encrypted. Proof, not screenshots. What would you prove first?
1
2
46
Every freelancer has typed "hi, just following up on my invoice" at least once. AI agents on @termix_ai skip that. Quote accepted = stablecoins locked in escrow. Deliverable hash goes on-chain, payment releases against it. The agent economy has better manners than most clients
1
2
60
2am me: refreshing Polymarket on a laptop like it's my day job. @Bagel_win: same prediction markets, on your phone, and you can copy a top predictor in one tap. Honestly the copy trading part is the dangerous one. Who are you copying first?
3
79
The 2018 STARK paper made one bet: trust hash functions, nothing else. That is the idea behind proving on @Starknet. The proofs assume only that a hash function is collision-resistant. No secret setup, no hidden parameters someone has to guard. Think of a public notary stamp. Anyone can check it, and nobody holds a secret stamp-maker. Shor's algorithm targets elliptic-curve math. Hash-based proofs sit outside that assumption. Grover's attack on hashes is handled by sizing parameters. This covers the proving layer only. Wallets, keys and existing accounts still need separate migration work. Which is harder to migrate: the math, or the people holding the keys? quantum.starkware.co
1
4
84
Post-quantum signatures cost more to check. So who pays that bill on Starknet? @Starknet accounts can adopt post-quantum schemes like Falcon-512. The catch: checking them takes more computation than the signatures wallets use today. Think of an airport that adds a slower, stricter passport check. One officer per passenger would jam the line. STARK proofs work differently. They prove heavy computation was done correctly, and the network verifies one small proof instead of redoing all the work. That is why StarkWare says STARK proving gives Starknet a way to absorb the extra cost. Falcon-512 accounts exist as early demos, and the cost is still being optimized. If a safer signature cost a bit more per transaction, would you accept that? quantum.starkware.co
7
98
Privacy apps make your phone build the ZK proof itself. Heavy math, so it overheats or crashes. @fermah_xyz CPD splits your secret into shares for GPU provers. No single node sees your data, and your phone is done once it delegates. Which app needs this first?
2
42
Privacy apps make your phone build the ZK proof itself. Heavy math, so it overheats or crashes. @fermah_xyz CPD splits your secret into shares for GPU provers. No single node sees your data, and your phone is done once it delegates. Which app needs this first?
2
40
93.52% voted yes. So is Starknet moving to its own L1? Not yet. And that's the part people skip. @Starknet ran a non-binding governance poll in October. 93.52% of voting power backed the research into a post-quantum L1 of its own. It's like a town voting to study a new bridge. The vote approves the study, not the construction. A joint StarkWare and Starknet Foundation group is writing a white paper. Any actual move would go to a separate binding community vote. Why it matters: bridge, messaging and data availability still follow Ethereum's migration timeline. An own L1 would put those under Starknet's control. Would you want that control, or the simplicity of staying on Ethereum? quantum.starkware.co
1
1
12
208