While CT spent the week writing threads about how SGP-3 kills complex apps, the guy behind it shipped a prop AMM to mainnet to settle the argument live. Under 3k CUs per swap. ~20k peak TPS of swaps and oracles hammering a SINGLE market, sequentially. Blocks didn't fill. Chain didn't blink. The only bottleneck was @cavemanloverboy's own machine sending transactions. Sit with that. The doomsday scenario got stress-tested on mainnet and the only thing that broke was the critics' argument. Here's what everyone's still missing: SGP-3 doesn't price compute. It prices contention. Requested CUs are what the scheduler reserves, and the real scarce resource on Solana is write locks. Request 400k, use 300k? You still locked capacity nobody else could touch. That's been free for years, and efficient apps have been paying your tab. The network over-requests ~20% on average. That's the invoice coming due. The MM complaint? By the author's own numbers, ONE pick-off loses a large prop AMM ~100x more than this proposal moves their annual costs. You're crying about parking meters during a car theft. And @toly already killed the repricing fear: the rate starts below the 50% priority burn Solana ran a year ago, and validators will never approve a price that eats their own revenue. SGP-3 doesn't punish complex apps. It punishes lazy ones. They were never the same thing. Some of you are about to find out which one you built.

On the Alpenglow bug bounty — and what "best in the biz" actually bought researchers A valid consensus-safety bug in Solana's new Alpenglow protocol. Confirmed by Anza. Patched by Anza. My reward: $0. Anza's CEO called this "the best bug bounty program in the biz." As someone who took part, I'll put that next to the receipts and let the Solana community judge. I reported a consensus-safety issue in the Alpenglow stack — startup rooting was selecting the root block off vote-state that discards the finalization certificate's block-id. Anza assessed it valid, and days later shipped the fix in agave #14418. Its own commit note states my thesis almost word for word: "Vote-state roots do not identify an Alpenglow block." My diagnosis. Their patch. It was then ruled a duplicate: another report of the same issue landed 3 minutes and 29 seconds after the two-week window opened, and under first-to-file (RULES §7) the earliest report takes the entire reward. Mine was valid too. It didn't matter. I accept first-to-file as a rule. What deserves scrutiny is the model around it: 1. The program missed the service level it published. RULES §9 promises "an initial severity determination within 7 days of submission." I got none until my advisory was closed on day 14 — twice the committed time. 2. First-to-file turns a two-week competition into a race decided in the opening minutes. The report that beat mine was on file eight hours before I even submitted; once a valid report exists, §7 disqualifies every later analysis of the same issue, however rigorous. A Consensus/Safety finding sits in a 3,125–12,500 SOL band (§5) — and the entire outcome turned on 209 seconds. 3. Participation was pay-to-enter: a non-refundable 0.5 SOL burn per submission (§1). Across 300+ submissions, that's 150+ SOL researchers spent simply to take part — with results disclosed only after the window closed. Anza confirmed the bug and shipped the fix — the outcome that matters for the network. My problem isn't any single researcher. It's a model where being 209 seconds faster outweighs every hour of analysis behind an identical, valid finding. The question is whether a program built to reward the people hardening Solana's consensus is actually designed to do that. As it stands, it rewards reaction time over research, misses the service levels it publishes, and charges admission to take part. For a network this important, it can and should be better. I'd welcome a response from @anza_xyz and @SolanaFndn — and I'd especially like to hear from the other 300+ who took part. How did it go for you? References: agave #14418 (github.com/anza-xyz/agave/pu…) · RULES.md (github.com/anza-xyz/alpenglo…) @anza_xyz · @bw_solana · @SolanaFndn · @solana

not enough people talking about this morning's solana close call a routing fault at one hosting provider knocked 28.83% of staked sol delinquent. finality halt line is 33.34%, so about 19.9M SOL of headroom left. recovered in ~30 min, nothing actually broke credit where due: the network held and validators bounced back fast. but it's a good reminder of why infra diversity matters silver lining is alpenglow. same scenario post-upgrade and the network keeps finalizing on the 60% quorum path. shipping later this year data from @MarinadeFinance #Solana #SOL #Alpenglow

🚨 @blockaid_ just flagged an ongoing exploit on @gardenfi HTLC contracts. ~$450k USDT already drained across: • Ethereum • Base • Arbitrum • BSC This is the same protocol that got hit for $11M+ last year. Cross-chain HTLCs keep getting targeted. If you’re interacting with any Garden contracts right now — stop and wait for official confirmation. Stay safe. #DeFi #Exploit #CryptoSecurity #GardenFi

🚨 $9.7M DRAINED. STILL LIVE. @TripleAHQ hot wallets just got cleaned across ETH, SOL, TRON, TON, Polygon & Arbitrum. Funds already swapped + bridged → 5,227 ETH sitting in one wallet. Deposits are STILL OPEN. New money is still getting vacuumed in real time. Team is completely silent. First flagged by @SpecterAnalyst + @PeckShieldAlert. Another brutal reminder: Hot wallets = public target practice. #Crypto #Hack #TripleA #Security Stay cold. Stay sharp.

another one. AFX perp dex just lost $24m on their arbitrum bridge. and the scary part is there was no bug. attackers got 5 compromised validator keys, so every withdrawal was signed correctly. valid quorum. the bridge did exactly what it was designed to do and handed over $24m. no audit catches this. the code was fine. the signatures were real. the money still left. "authorized" and "should be allowed" are not the same thing. if nothing caps what can physically leave the vault, a valid signature is all it takes to drain everything. $31m gone across two bridges in 7 hours. same lesson every time. source: cointelegraph.com/news/afx-p… #DeFi #Crypto

honestly impressed watching this. after everything that happened this year, most protocols just tweet "we've been audited" and hope. @FlashTrade actually sat down and capped their own worst case. payoff ceilings, OI limits, on-chain pyth checks, and an hourly outflow cap coming that watches the ledger live instead of just trusting the code. the amount of thought @XohebS and @allinbitcoin_ put into "where can money physically leave the vault" is not normal in this space. this is what taking user funds seriously actually looks like. #Solana #DeFi

2010: Iniesta, 116th minute. 2026: Ferran Torres, 106th minute. 16 years apart. Same story. One goal, one star. Messi leaves the stage. An 18-year-old owns it now. Spain 1-0 Argentina. Champions of the world. 🇪🇸🏆 @SEFutbol #FIFAWorldCup #ESPARG #VamosEspaña

The market knew. Spain: -156 to lift the trophy. +130 to win it in 90 minutes. Translation: champions — but only after extra time. Ferran Torres, 106th minute. First goalscorer at +700. Spain 1-0 Argentina 🇪🇸🏆 #FIFAWorldCup #ESPARG #VamosEspaña @SEFutbol @FIFAWorldCup

SpaceX stock crashed 40% in four weeks. Everyone saw that. What almost nobody noticed: the tokenized version on Solana traded through every single minute of it — weekends, 3am, while every brokerage on Earth showed "market closed." The crash wasn't the story. It was the stress test. On-chain equities passed. Full story ↓ xitter.cf/Dev_ABDR/status/207737…