david q

4K posts

david q banner
david q

david q

@Davidgek97Q

web3 native, solidity developer, btc maxi

Katılım Eylül 2011
620 Takip Edilen3K Takipçiler
david q
david q@Davidgek97Q·
@Normansaga i’m with you on backing real products over pure hype, but what’s the strongest traction signal for squire and po right now?
English
0
0
1
51
Norman Liq
Norman Liq@Normansaga·
Yeah, apeing into memecoins on the Robinhood chain is a vibe, but the heavy bags in your portfolio should always be parked in solid tech plays. Right now, there are two massive gems with killer products and insane upside potential that we haven't seen in a hot minute: $SQUIRE and $P0. If we actually want to save Solana, we need to max bid and fully back these real world utility projects. Everyone from devs to degens has a part to play here. Let's weaponize AI 2.0 and save the trenches together. @squire_bot @P0Systems @blknoiz06 @toly @mert @rajgokal
English
3
7
22
1.1K
david q retweetledi
Kardinall
Kardinall@kardinall·
In 1962, Kenneth Arrow won a Nobel prize for a paradox about selling knowledge. The buyer can't value information until he has it. And once he has it, he's already taken it for free. The seller risks everything. Sixty years later, the CEO of Microsoft says the paradox has reversed. Now the buyer is the one who reveals everything. You pay $20 a month for the model. Then you pay again. In prompts that describe your workflows. In files that carry your numbers. In corrections that hold twenty years of your judgment. Every time it's wrong and you fix it, your fix becomes their training data. They learn how you think. You never learn what they learned. Nadella calls it the Reverse Information Paradox. Alex Karp said it plainer: customers want to own the means of production, and the current regime quietly transfers it away from them. Trace by trace. Correction by correction. The machine in this clip admits it's dangerous. The man who sells machines just explained the part nobody prices in. In the cloud era, companies accumulated data. In the AI era, they leak the one thing that made them different.
Satya Nadella@satyanadella

x.com/i/article/2076…

English
14
10
44
1.4K
david q retweetledi
Ridark
Ridark@ridark_eth·
The Golf "Dust-Collector": Making $24,343 on Low-Yield Predictability in a Week Profile Statistics: -> Total Profit: $24,343.17 ( For this week) -> The Biggest Win: $19.7K -> Total Predictions: 3,208 -> Golf ⛳️ The Strategy: Systematic Golf Shorting - Selling the Hype: In professional golf tournaments, dozens of players compete simultaneously. Retail traders love to buy cheap "Yes" contracts on their favorite players to win or finish in the Top 5/10/20. This trader does the exact opposite: they act as the bookmaker, buying the "No" side at heavily discounted prices, usually between 95¢ and 99.9¢. - Compounding "Dust": On a single contract, risking $99 to make $1 seems crazy. But this trader applies the law of large numbers. They don't just bet against one golfer, they buy "No" contracts on dozens of different players at the same time (e.g., Lucas Glover, Tom Hoge, Wyndham Clark).. - The Math Behind the Grind: Out of a field of 150+ golfers, only one can win, and only five can finish in the Top 5. By shorting 30-40 different low-probability players at 98¢+, the trader mathematically guarantees that 95%+ of their positions will expire at $1.00, creating an automated avalanche of small profits. Top Deals from the Golf Dashboard: > ISCO Championship (Lucas Glover Win - "No"): Bought "No" at 64.3¢ -> Won $5,186.59 (+$1,852.38 / +55.56%) > Genesis Scottish Open (Rory McIlroy Win - "No"): Bought "No" at 99.9¢ -> Won $3,016.10 (+$3.38 / +0.11%) > Genesis Scottish Open (Wyndham Clark Win - "No"): Bought "No" at 99.3¢ -> Won $2,014.16 (+$14.16 / +0.71%) > ISCO Championship (Taylor Pendrith Win - "No"): Bought "No" at 58.6¢ -> Won $917.88 (+$379.84 / +70.6%) Why does this work? Golf fields are inherently inefficient on prediction markets. Because the odds of any single mid-to-low-tier golfer winning are mathematically minuscule, their "Yes" contracts should be worth fractions of a cent. However, human bias and hype keep those "Yes" prices artificially high (at 1¢ to 5¢). This trader simply spots these overvalued dreams, locks in the "No" contracts at 95¢–99¢, and lets the brutal reality of tournament golf do the rest. They are essentially running a private casino where the math is heavily stacked in their favor..
Ridark@ridark_eth

x.com/i/article/2074…

English
35
5
93
7.2K
david q
david q@Davidgek97Q·
@marcosagusstinn europe already has the talent and capital, but fragmentation keeps slowing everything down. which shared project do you think should come first?
English
0
0
0
2
david q retweetledi
Brian Roemmele
Brian Roemmele@BrianRoemmele·
DO YOU THINK THE WALL AI COMPANIES ARE HITTING WILL BE SOLVED WITH MORE COMPUTE? What only you know and a few now in China is: AGI and ASI will come about by this way of curating and training AI. The US is on the all you can eat junk food buffet of Reddit and Wikipedia data for AI training. And this paper proves it is the wrong path. This is not theory. This is the operating manual for the next phase of real, grounded, individual-scale intelligence amplification. The future belongs to those who kept the structure. I never threw it away.
Brian Roemmele tweet media
Brian Roemmele@BrianRoemmele

BOOM! CHINESE PAPER PROVED MY AI GARAGE SYSTEM! OFFLINE HIGH PROTEIN TRAINING DATA BEATS TEXT OFF THE INTERNET FOR AI TRAINING! China knows now in the US, no one listens. Paper: Scalable Visual Pretraining Challenges Text-Only Dominance for Richer Language Intelligence A new paper demonstrates that unsupervised visual pretraining directly on visual documents (figures, equations, layouts) outperforms text-only extraction on the same corpora across backbones and benchmarks. Systematic study shows visual cues provide scalable pathways to foundation model intelligence beyond plain text. Text extraction is lossy by design. When you flatten a scientific paper, a technical manual, a lab notebook, or a dense PDF into plain text, you destroy the very signals that carry meaning: the diagram that shows the circuit or the molecular structure, the equation rendered in proper notation, the table layout that reveals relationships at a glance, the handwritten marginalia that captures insight, the visual hierarchy that tells you what matters most. The paper’s authors didn’t just theorize this. They ran the controlled experiment: same source documents, two paths — one stripped to text, one kept visually intact — and unsupervised visual pretraining won… Every time. This is not “multimodal icing on the cake.” This is fundamental. Visual structure is part of the intelligence. The default text-only paradigm in LLM pretraining is proven wrong. High-Protein vs. Internet Empty Calories Let’s be brutally clear: Internet text corpora are mostly empty calories. They are diluted by repetition, poisoned by SEO garbage, stripped of diagrams and equations, full of confident-sounding hallucinations, and optimized for engagement rather than truth or depth. Quantity at the expense of quality and structure. My offline personal archives are high-protein. Every document I train AI on has context I actually care about. The visual elements are preserved. The relationships are explicit. The signal-to-noise ratio is orders of magnitude higher because it was curated by a human with skin in the game over decades. Training local models on this kind of data produces something fundamentally different: grounded, less sycophantic, more useful, more aligned with actual human intent and wisdom. This is the data foundation for the agents and soon robots that function as true extensions of individual intelligence rather than generic corporate parrots. The paper shows unsupervised visual pretraining is not only better it is scalable. That opens the door for individuals and small teams to do what only hyperscalers could do before. What this means in practice: • Stop treating your personal PDFs, scanned lab notebooks, technical diagrams, and structured notes as “just files.” Treat them as premium training fuel. • Preserve visual structure. Do not blindly OCR everything into plain text. • Build (or continue building) your own ontology/taxonomy relational layer on top of the visual documents. • Use this corpus for local adaptation, continued pretraining, or retrieval-augmented workflows. • The resulting models will be smaller, more capable in your specific domains, more trustworthy, and far less likely to regurgitate web slop. I have been running versions of this system in my garage lab for years — combining personal visual/structured data with local inference, voice synthesis, and agent frameworks. The paper is not telling me something new. It is giving the broader world permission to stop doing it the dumb, lossy, internet-dependent way. Well, now some folks in China and NO large US AI company know, what you and I have known for about 4 years here. This is one of my reasons I promote OFFLINE HIGH PROTEIN TRAINING DATA. US AI companies, a question: Where would you be now if you worked with me and listened? What do you folks think? • Link: arxiv.org/abs/2607.09657

English
22
26
139
19.7K
david q
david q@Davidgek97Q·
@akrWeb3 @quipnetwork being live already is what makes this interesting. what kind of assets are people putting into the quantum vaults right now?
English
1
0
1
14
α𝗄𝗋
α𝗄𝗋@akrWeb3·
Good night fam ✨ We don’t always need to build something new. Sometimes it’s about making existing tech actually useful. That’s what I like about @quipnetwork. Quantum Vaults are already live, and people are putting assets into them. A year ago, that would’ve sounded pretty far off, but today it’s live…
α𝗄𝗋 tweet media
English
26
2
27
2.2K
david q retweetledi
Finance Guy
Finance Guy@GuyTalksFinance·
Couple I know bought their first house in 2021 with a 2.5% interest rate. Since then they had a few kids and now want to move into a bigger house. Problem is they won’t be able to afford the monthly payments on a bigger house with the current interest rates. Their only solution is to spend over $150k on renovations to expand and improve their existing house.
English
98
5
136
24.8K
david q
david q@Davidgek97Q·
@flipkoin2 the founder incentive part feels central here. if the safest outcome is selling to the incumbent, why would anyone keep building to seriously challenge it?
English
0
0
1
13
🔻#LΔṈDBΔCK
🔻#LΔṈDBΔCK@flipkoin2·
Financialization and consolidation do not merely accompany economic stagnation. They are two of its engines. Financialization means capital stops building. The clearest evidence is the practice nobody markets as innovation: stock buybacks. Apple alone has spent more than $700 billion purchasing its own shares. Across Big Tech, the total reaches into the trillions. That money creates no new product, laboratory, worker, infrastructure, or public capacity. It raises the share price, increases earnings per share, and helps executives reach stock-based compensation targets. Once management is rewarded primarily for stock performance, returning cash to shareholders becomes safer and more profitable than uncertain research and development. Capital that once financed the future is redirected into polishing its own reflection. Consolidation means creation gets bought, contained, or buried. Venture capitalists describe a "kill zone" around dominant firms such as Google, Meta, and Amazon. Investors hesitate to fund companies operating near their markets because the incumbent can copy the product, block its distribution, acquire the company, or absorb its features. The ambition is therefore reduced. Instead of building an independent institution capable of challenging the incumbent, founders are encouraged to build a feature that can be sold to it. Meta acquired Instagram and WhatsApp because independent communication networks threatened its control over attention, advertising, and social infrastructure. Acquisition converted potential rivals into controlled assets. Put financialization and consolidation together and the result is structural stagnation. Production carries risk. Rent extraction does not. Capital therefore flows toward the toll booth rather than the workshop, toward ownership rather than invention, and toward controlling access rather than expanding possibility. The frontier is not empty. It is fenced. Creation still occurs, but the moment it becomes valuable, it is captured by institutions whose profitability depends on preventing it from becoming independent. That is not a conspiracy. It is the predictable gravity of a financialized monopoly economy.
English
1
2
2
108
david q
david q@Davidgek97Q·
@anirudhchain @get_optimum the validator interest before mainnet is what caught me. what do you think optimum does better than the other memory layer projects?
English
0
0
0
16
Anirudh
Anirudh@anirudhchain·
still sleeping on @get_optimum? an MIT professor invented the tech. top ETH validators lined up to test it before mainnet even shipped. $11M raised before most of CT knew the name. the alpha was never hidden. some of us just paid attention early. reply 🧠if you were here before "the memory layer" became a headline.
Anirudh tweet media
English
6
0
10
288
david q
david q@Davidgek97Q·
@neetu_arnold the shsat may expose the learning gap, but removing it doesn’t fix that gap. what would you change earlier in the school pipeline?
English
7
0
5
318
Neetu Arnold
Neetu Arnold@neetu_arnold·
Brooklyn Tech grad on why it’s wrong to eliminate the SHSAT: “Anti-SHSAT officials have used the test as a scapegoat for the lack of diversity in the specialized high school system, rather than aiming to reform the weak learning environments, standards, & curriculum in the city’s public elementary & middle schools” “Many didn’t have simple multiplication & division skills” “As an African-American student…it pains me to see my peers & others…using severely misguided arguments in opposition to the test”
Neetu Arnold tweet media
Lincoln Restler@LincolnRestler

1 in 5 NYC public school students are Black, but at Stuy - our most prestigious high school - 3 of ~800 incoming Freshmen are Black. We urgently need state legislation to modify the admissions process. A single test should never be only factor deciding who gets in & who doesn’t.

English
23
27
205
15.1K
david q retweetledi
david q
david q@Davidgek97Q·
@to_be_Kuber @vimaglabs this could be massive if the pilot data matches the claims. curious how it compares with permanent magnet motors on efficiency, heat, and long term durability.
English
0
0
0
14
Shyam
Shyam@to_be_Kuber·
🇮🇳 DEEP-TECH BREAKTHROUGH: India is rewriting the rules of EV technology!Bengaluru-based startup @VimagLabs just secured a landmark patent for India’s first software-defined, magnet-free EV motor platform! 🧠⚡ Traditional EV motors rely heavily on rare-earth magnets—a supply chain notoriously monopolized by China. Vimag's Virtual Magnet Synchronous Motor (VMSM) completely throws physical magnets out the window. Instead, it dynamically generates and controls magnetic fields in real-time via advanced power electronics and proprietary software algorithms. Why this changes everything: ❌ No rare-earth magnets = Immune to global supply chain squeezes 📉 Massive cost reductions for EV manufacturers & consumers 💪 Matches or beats standard permanent magnet motor performance 🛠️ Fully pushes the boundary for #AtmanirbharBharat deep-tech Backed by Accel, the company is already running active pilots with major 2-wheeler and passenger vehicle OEMs. The future of clean mobility isn't just about batteries—it’s driven by software! 🚀
Shyam tweet media
English
2
0
3
105
david q retweetledi
SafZ
SafZ@CrypSaf·
Aight guys, let's pause the NFTs on Robinhood chain shall we? Or at least take a deep breath I've already seen +50 NFT projects by now Robingals, Robinhoodkitty, RobinElves, RobinPunks, RobinZuki, RobinApes, RobinTrolls, RobinGoblins, CatHood, MonkeyHood, MusterbateHood, RobinDeezNutsHoob, RobinCum, HoodisGAY, AnyRobinCanSuckDem, HoodLadies, RobinTheNonRobin, HoodBull, RobinFrogs, Robinhood bandits, RobinBears, Hoodbunnies, SuckdemnutsHood, hoodHole, ... All are probably coming from the same people - AI art - cheap mint price - high supply - accounts created yesterday, mints in 3 days - no founder, no nothing Repeat, till the funds dres from the chain Then we move back to ETH slaving there ... We have seen it happening before in: - aptos, apechain, berachain, blast, sei, abstract Ty for your attention to this matter 💚
SafZ tweet media
English
205
20
389
16.4K
steponaduck
steponaduck@Steponaduck·
Mississippi State plays the 2nd hardest SEC schedule this year per my metrics. I have the Bulldogs favored in two games all season while being double digit underdogs 6 or 7 times. Last years team was aided by early season success in the non-conference which sparked a 4-0 start. What record did they finish with? 5-8 Defense was Swiss cheese in sec league play. 3 YPC in conference play. 45 sacks and 98 TFLs surrendered. 7 OL from LY 2-deep gone. No good fortune of being able to ease into it this year. B2B at Minnesota/Sou car and then Missouri/bama create an almost impossible scenario. It’s 1-4 or 2-3 to start in all likelihood. The remaining opponents are: LSU Oklahoma Texas Vandy Auburn Tennessee tech Ole Miss So let’s give cowbell the benefit of the doubt, which they have not earned. 2-3 start. Where are the other 3 wins against the opponents listed above? You are putting a lot of stock into a team that plays a brutal schedule, has less talent than most everyone else, and is regarded as having every position unit ranked in the bottom 5 of the sec at each position. Be a fan…fine. But this team isn’t overly good. It’s 4-8 or 5-7 in all likelihood my man. Make piece with it.
English
6
2
5
1.8K
david q
david q@Davidgek97Q·
@FawazXweb3 the volume jump is interesting, but i’d wanna see if it stays elevated after the earnings hype fades.
English
0
0
0
7
Fawaz
Fawaz@FawazXweb3·
MU futures volume on MEXC surged 142% the day earnings hit The day after, combined futures volume across four memory-chain names, Micron, SanDisk, SK Hynix, and the DRAM ETF, doubled versus the prior day. One earnings report. Four stocks moved Then the number that actually tells the story: post-earnings, MU priced at $1,237 on RealStocks, $1,235 on Futures, $1,233 on Tokenized. Three different products. <0.4% apart And it's bigger than one earnings print. South Korea just went all-in, more fabs, more HBM production, more data centers. When a whole country repositions its industrial strategy around AI infrastructure, that's not a oneweek reaction to a Micron move That's the chain confirming it has legs well past this cycle AI was never a single stock story. It's a chain: chips, memory, cloud infrastructure, Big Tech. When one link moves, the ones next to it move with it. Micron this time. GPUs, packaging, or power infrastructure next time The question isn't which stock. It's how you're positioned On @MEXC, you open an account, you get access to the four: → US Stock Futures — go long or short, 0 platform trading fees → RealStocks — real shares, real dividends, held via a licensed broker partner → Tokenized Stocks (Ondo × xStocks) — on-chain exposure, 24/7 → Pre-IPO — subscription-based pre-listing access
Fawaz tweet media
English
21
2
19
928
david q retweetledi
El Jefe
El Jefe@Jeff_Brunson·
Texas Tech can’t play themselves. One of these days, their brains are going to fire up and they are going to realize that the narrative that Tech has the weakest schedule is more an indictment on them than it is on Texas Tech. Texas Tech’s OOC schedule is only 3 games. Texas Tech’s Big 12 schedule is 9 games.
YaWooCougarHappiness🐾@santiromero26

Every BIG12 team that is playing Tech has a higher SOS than Tech, but our teams should be ashamed?

English
14
6
157
16.2K
david q
david q@Davidgek97Q·
@colin_beazley switch hitter with 70 grade defense at shortstop feels like really good value at pick 44.
English
0
0
0
50
david q
david q@Davidgek97Q·
@teelacream1 i agree, the tech gets attention but the interface is what decides whether i actually keep using the product.
English
1
0
1
20
tEO
tEO@teelacream1·
okay, i actually like this Variational omni v3 interface apart from just focusing on the tech aspect, i love teams that invest in the application interface. making trading smoother and navigation more intuitive matters more than most people admit the new UI is clean, and the PWA support makes mobile use feel way better. little things like this compound, when a product feels good to use, you'll always come back to it good update imo
Variational@variational_io

A brand-new Variational UI is now live, featuring reworks of each page of the app and an improved mobile experience with PWA support.

English
15
2
23
1.2K