Samuel Thomas

Advisor @quotientvc

[Michelle @michellelsun]

.@nvidia picked one of the priciest dexterous hands on the market for GR00T, @SharpaRobotics's Wave, 22 DoF, $50,000.

The dexterous hands market is much wider and cheaper:
- Linkerbot L30: 22 DoF, $13,800, 500-skill dataset, claims ~80% share
- @AGIBOTofficial OmniHand: tiered SKUs from $1,350
- Wuji: mechanical transparency, $5,500

Four dimensions of what a robot hand should optimize for. Here's how they stack up:

Full teardown in this week’s newsletter.

[Samuel Thomas @Sthomas1100]

@bj0hn5on @hussam_9414 That’s a fact.

[Samuel Thomas @Sthomas1100]

@hussam_9414 @bj0hn5on @SemiAnalysis_ @roaner @AnushElangovan @LisaSu Winning Llama2 70B Interactive in one narrow configuration is real progress. Calling it beating B300 broadly is the mental acrobatics here. AMD chose the one test they could win.

[Samuel Thomas @Sthomas1100]

@hussam_9414 @bj0hn5on @SemiAnalysis_ @roaner @AnushElangovan @LisaSu AMD submitted selectively and skipped DeepSeek R1, Qwen3-VL, and disaggregated inference entirely. These are not niche tests. They are the workloads hyperscalers are deploying at scale right now.

[GitFish @gitfish]

GitFish V2 is now live on @Solana.

1 GitHub project -> 1 coin. Popular projects cost more points to launch.

Support open source. Find the next big thing.

P.S. We're airdropping bonus points to V1 users daily over the next five days. The first drop is live — enjoy!

[Arc Intelligence @IntelligenceArc]

1/ Introducing ATLAS: A Hybrid RL Reasoning System for Compounding Intelligence, the learning layer for agentic AI.
RL + adaptive optimization.
Enterprise-ready, fast to deploy, developer-friendly.

[Farm Capital (mFARM) @FarmCapitalXYZ]

We have achieved 21.3% APY right after the official launch of mFARM - our innovative, market-neutral, on-chain solution designed for stable, high-yield farming, in partnership with @MidasRWA.

mFARM offers tokenized exposure to strategies such as fixed-rate lending, early-stage liquidity provisioning, and arbitrage-based opportunities, with extra airdrop points as rewards.

Why mFARM?

1/ The current APY of 21.3% offers higher returns than other stablecoin savings and yield farming platforms, such as AAVE's USDC Supply Interest Rate of 4.5%, Ethena's sUSDe APY of 6.5%, and Centralized Exchange USDC Earn Programs with APYs of 3-10%. mFARM is currently the highest-APY product among the 12 products in the Midas ecosystem.

2/ mFARM is safe and governed with full transparency.
- Governed by Midas, one of the biggest asset tokenization platforms with over $900M TVL and a partnership with BlackRock.
- Adopts MPC Fodefi wallet infrastructure for governance to avoid malicious activities and minimize smart contract risks.
- Full transparency of mFARM wallets on DeBank, which depositors can access to view mFARM positions anytime.
- A diversified portfolio constructions with risk diversification

3/ Unique Strategy
- Through strategic collaborations with top-tier ecosystem projects, mFARM offers special TVL terms and points rewards from Falcon Finance, Huma Finance, Avant, Resolv, YieldFi, Hyperwave, etc., which users could not access publicly. It also saves users time on airdrop farming.
- mFARM will soon launch an Euler lending market vault where users can deposit mFARM to borrow against USDC. This means users can leverage and loop on top of mFARM to generate extra yield. At the same time, users can access decent liquidity by holding mFARM to enjoy high yields while borrowing against stablecoins.

4/ Why Deposit now?
- There are no management fees or performance fees in the first 1-3 months (or even less). Afterward, a 20% performance fee will be charged.
- By depositing $10,000 as one of the first 100 users, you will gain an NFT issued by us later on. We treasure our early supporters and would like to reward them as we make money together.

Deposit now: https://t.co/8KEdh2bWy7.

[steve @Chefsteve]

Before Kalshi, I worked at a well-known mainstream news company. My role centered around data-driven storytelling (translating data from source to screen).

I feel qualified to offer my take on how prediction markets democratize data collection by fixing the data lag problem.

Introduction
In my last role, I realized how traditional data-gathering comes with a fatal flaw: lag.

Datasets react slowly to fast change. Data is gathered in arbitrarily long periods of weeks and years.
1. It takes a month for a jobs or inflation report to resolve to a number.
2. It takes a season or a year to measure cumulative weather patterns.
3. It takes a day to assess how the market digests a piece of news.

Instances of lag create pockets of inefficiency, ultimately causing loss to downstream actors:
1. A shock inflation report destroys your 401K.
2. A bean counter in a Midwest state ratchets up your home insurance premiums because you’re now in a flood zone.
3. Asymmetric information on a drug trial tanks a stock.
(I can list 50 more, but you get the idea)

How Prediction Markets Address Lag
Prediction markets turn jagged, slow-to-arrive datasets into a living probability stream.

Instead of waiting for the next step on a choppy staircase, you get a smooth curve that updates in real time. Such a source of truth makes actuarial, financial, and social models better by compressing the loss function of a staggered data medium that takes an arbitrary amount of time to net out.

But, Don’t You Need Data for PMs to Work Efficiently?
Yes! Of course. Without backing data, Prediction Markets are just Magic Eight Balls. Data collection will always exist – but prediction markets implicitly absorb that data in a dollar-backed manner.
1. Prediction markets are data aggregators, and will champion the views of accurate, unbiased, and current sources.
2. They account for hard data and soft signals in real time.
3. The efficient market rewards participants for correcting odds that are incorrect.

Does participant profile create bias?
Somewhat, but not completely. The profile of who participates does matter, but the design of prediction markets tends to self-correct bias in ways that make them unusually robust compared to most data-gathering systems.
1. Outcome tethering reduces drift. All markets ultimately resolve, and participants are rewarded for forgoing biases. Over time, participants seek to normalize their bias to maximize personal outcome.
2. Bias gets arbed. If a biased cohort of participants causes priced odds to drift, another cohort will be incentivized to forcibly reprice the market.
3. An efficient market consists of many transactions from heterogeneous participants. Like a stock market, retail, experts, hedge funds, etc. all offer their own edge that is assembled into the right-pricing of a market.

Net effect: Bias exists at the individual level but is continually priced out at the system level. Prediction markets transform the noise of individual conviction into a smoother, more accurate probability curve that outperforms lagging, one-dimensional datasets.

Who cares?
There is a marriage between data and prediction markets. One cannot exist without the other.

Through various right-pricing mechanisms, prediction markets establish a continuous curve of probabilities that cuts noise, compresses loss from laggy data, and yields a better "final number" for modeling. Outcomes in the continuous real world can now be rooted in more effective metrics.

In a real world scenario, this means referencing prediction market outcomes directly in models, news articles, analyses, and more. And this is just starting with adoption from market makers, news companies, and data platforms.

TLDR
Data usually comes in arbitrary intervals. Prediction markets have different ways of compressing loss between these intervals. You can use this right-pricing mechanism for useful real-world applications.

You're still early.

[XMTP @xmtp_]

As of today, XMTP is now a fully quantum-resistant decentralized messaging protocol.

This means, any developer, anywhere in the world can leverage XMTP to provide their users with private, decentralized, & quantum-resistant messaging.

[Packy McCormick @packyM]

Everything is technology.

https://t.co/7ssQ3RSO5c

[Nemil Dalal @nemild]

Today, @coinbasedev is announcing x402: an open standard/protocol to bring onchain payments to the web. Join us in killing the API key, and making the agentic and payable web a reality. All in one line of code.

Imagine a future where crypto is the default to pay for *anything*.

[Noise @noise_xyz]

People trade what they understand: attention.

The purest measure of attention is mindshare.

We’re partnering with @KaitoAI to turn mindshare into tradable trends.

[Samuel Thomas @Sthomas1100]

@EmanFT_ @shieldpay_co @SiliconVlyBank @Founders1st @isaiahudotong @LuissCarchi Congratulations!!

[eman.eth @EmanFT_]

We won 1st place! 🏆🎉

@shieldpay_co is proud to take 1st in @SiliconVlyBank + @Founders1st BHM Pitch Competition!

Huge thanks to SVB, StartupOS and judges for recognizing our vision to revolutionize banking and payments for export businesses 🚀

[Analog 🕓⛓️ @OneAnalog]

Creating history 🧑🏻‍💻 - Witness how Analog is rewriting the rules of blockchain #Interoperability, one block at a time!😎🦾 #timechain

Watch the #BTS of how Analog sprung to life, survived, and thrived through adverse circumstances ⏳#proofoftime 💜

@VictorYoungMe @ericwang1215 @sanchalr @AnalogIntern

[mert @mert]

big news

two former SEC chief economists have released a paper on Solana and the SOL market — check it out

happy to say @heliuslabs helped support the research alongside other dope Solana teams

link below

[Samuel Thomas @Sthomas1100]

@satsurance @btcstartuplab @quotientvc That's awesome. Okay, I understand. Congratulations! Of course, and I'll message you!