Wins outweigh losses. Storm Rider. Adventist. Pendle YT
Stablecoins. Privacy. AI.
Joined July 2014
- Tweets47.6K
- Following1.2K
- Followers1.3K
- Likes73.2K
Pinned Tweet
I don't know why projects still do this.
You allocate 7% of your token supply to your community as an airdrop.
And then you give them two options:
1. Claim 35% of their airdrop at launch and forfeit the remaining 65%.
2. Vest their allocation for 12 months to receive everything.
Because left to me, you're indirectly allocating just 2.45% of your token supply for the airdrop, not 7%.
The community already spent months supporting and using your platform to qualify for those rewards.
Why should they have to sacrifice 65% of what they've earned just to claim at launch?
Onenee retweeted
Tether froze more than 1.567B USDT related to a potential Binance exchange hack.
Protecting the industry and our partners is our main priority.
Stay safe everyone.
$2.77B exploit by hackers on Binance.
Our teams are aware of reports of a potential security issue and are actively investigating. At this stage, we have not confirmed any impact to user funds.
Our security teams are reviewing all relevant information with urgency, and we will provide updates through official Binance channels as soon as we have verified information.
Users should remain cautious of scams and should not trust unofficial links, messages, or requests for account information.
There's a new version of this post
Onenee retweeted
A Technical Breakdown of Walled Gardens, Cross-Chain Metadata Leakage, the Orchestrator Architecture, and Zero-Knowledge Liquidity Routing in @RaylsLabs
Onenee retweeted
Can a Robot Skill Learned by One Arm Transfer to Another?
Imagine two robot arms performing the same task: picking up a small box and placing it on a shelf.
The first arm has a different reach, gripper and joint configuration from the second.
Even if both receive the same instruction, their movements cannot necessarily be identical.
One arm may reach the box directly. The other may need to approach from the side because of its geometry or the surrounding obstacles.
This introduces an important challenge in Physical AI: cross-embodiment generalization.
An embodiment is the physical form through which a robot interacts with the world, including its joints, sensors, grippers and other hardware.
Why does this matter for training?
1. Different kinematics
The same target position may require different joint movements because each robot has its own physical structure.
2. Different interaction capabilities
A gripper designed for simple pick-and-place tasks may behave differently from a dexterous hand when handling delicate objects.
3. Different sensor perspectives
Camera placement and field of view affect what each robot can observe while performing the task.
4. Different control requirements
A movement that is stable and easy for one robot may be difficult for another because of differences in reach, speed or mechanical constraints.
Training across different embodiments can help researchers investigate which skills transfer and which require adaptation.
However, mixing data from several robots is not enough by itself. The data must contain useful information about their actions, physical configurations and interactions with objects.
This is one reason @PrismaXai 's Robotics Data covers demonstrations across multiple robot arms and task scenarios.
The deeper research question is not simply whether different robots can share training data.
It's which parts of a learned skill are general, and which depend on the hardware performing it.
Answering that question is important for building physical AI systems that can scale beyond a single robot design.
@MaxC16134 @vivianrobotics
Onenee retweeted
I see @GenLayer as multiple AI models evaluating evidence to decide whether work meets contract requirements.
But consensus alone cannot guarantee fairness: transparent criteria and a meaningful right to appeal remain essential.
#JusticeNeedsAnUpgrade
🤝 Paid partnership
Slow. Expensive. Wrong.
It's been this way for centuries. It doesn't have to be.
#JusticeNeedsAnUpgrade
Onenee retweeted
Web3 Growth Notes — Day 3
One thing I think Web3 projects need to pay more attention to is how they choose creators for their marketing campaigns.
I've seen people focus so much on follower counts that they forget to ask the most important question.
Can this creator actually help us reach the people who might use our product?
Think about it.
A creator has 100k followers and promotes your project. The post gets thousands of likes, plenty of impressions, and people talking in the comments.
But after the campaign, how many people actually visit your platform? How many try the product? How many become regular users?
Now imagine another creator with just 2k followers who understands your product and has an audience that trusts their recommendations.
That creator might bring fewer impressions but attract people who are genuinely interested in what you're building.
Do you see the difference?
I'm not saying creators with big audiences are not valuable. They can be very useful, especially when your goal is awareness and visibility.
But a big audience doesn't automatically mean a good campaign.
This is why I believe projects need to look beyond follower counts when choosing creators.
• Look at their audience.
• Look at the kind of conversations they generate.
• Look at whether their content reaches the people you're trying to attract.
And if your goal is to bring in users, make sure you're tracking results beyond likes and impressions.
Because at the end of the day, you need to know what the campaign actually did for your project.
Sometimes, the creator with the biggest audience isn't the one you need.
The question is, are you choosing creators because of their numbers, or because they can help you achieve your goals?
Web3 Growth Notes — Day 2
Not every Web3 project needs to spend more money on marketing.
Sometimes, the problem is that you're marketing on the wrong channel.
I've seen people act like if you're building in Web3, you must be active on X, grow a Discord, post on Telegram, run campaigns, and do everything at once.
But do you really need all of that at the same time?
Let's say you're building a product for developers.
You can spend weeks posting general content on X, hoping developers will notice you.
Or you can identify where those developers already spend their time, the kind of content they pay attention to, and what would make them interested in trying your product.
Which one makes more sense?
This is where I think some teams get it wrong.
They copy what worked for another project without asking if it can work for theirs.
One project might grow through X because its audience loves discussions and educational content.
Another might get better results from developer communities, technical documentation, partnerships, or direct outreach.
It doesn't mean one project has a better marketing team than the other.
Their audiences and products are different.
The channel that works for one project might not work for yours.
Before you start spreading your team across every platform, ask yourself:
→ Who exactly are we trying to reach?
→ Where do these people already spend their time?
→ What kind of content or activity would get their attention?
→ How will we know if this channel is actually bringing results?
You don't have to be everywhere to grow.
Find the channels that make sense for your product, test them, pay attention to the results, and put more effort into what works.
Of course, you won't always get it right on the first try. That's why you test instead of just assuming.
Marketing isn't about being everywhere.
It's about getting in front of the right people with the right message.
Onenee retweeted
𝗖𝗮𝗻 𝘆𝗼𝘂𝗿 𝗼𝗻𝗹𝗶𝗻𝗲 𝗮𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗯𝗲𝗰𝗼𝗺𝗲 𝗽𝗿𝗼𝗼𝗳 𝘁𝗵𝗮𝘁 𝗮𝗻 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗰𝗮𝗻 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘃𝗲𝗿𝗶𝗳𝘆?
Think about how much useful information exists on websites today. Your financial records, purchase history, account activity and other online information can help applications make better decisions.
But there is a problem.
𝑯𝒐𝒘 𝒄𝒂𝒏 𝒂𝒏 𝒂𝒑𝒑𝒍𝒊𝒄𝒂𝒕𝒊𝒐𝒏 𝒌𝒏𝒐𝒘 𝒕𝒉𝒂𝒕 𝒕𝒉𝒆 𝒊𝒏𝒇𝒐𝒓𝒎𝒂𝒕𝒊𝒐𝒏 𝒚𝒐𝒖 𝒑𝒓𝒐𝒗𝒊𝒅𝒆 𝒊𝒔 𝒈𝒆𝒏𝒖𝒊𝒏𝒆?
A screenshot can be edited, and simply making a claim about your online activity does not make it true. Applications need a reliable way to verify the information they use.
This is where @primus_labs comes in.
With zkTLS, Primus helps verify information from websites using cryptographic proofs.
This makes it possible for applications to check certain claims about web data without having to rely only on screenshots or trust what someone says.
For example, imagine a platform that needs to verify information about your financial activity before offering you a service. Instead of asking you to expose your entire account history, it could use a proof of the relevant information.
The exact process depends on what information needs to be verified and how the application is built.
The bigger idea is that information from the internet can become useful evidence for Web3 applications, rather than something they have to accept without a proper way to check it.
Because when applications make decisions based on data, knowing where that data came from and being able to verify it matters.
Onenee retweeted
Bullish 🌊
Pharos has activated protocol staking rewards for $PROS following six months of zero inflation on PacificOcean Mainnet.
The transition took effect at block #21499558. Liquid staking is available via stPROS on Faroo, with native staking also supported.
PROS and PROS. No CON
#PharosStoryteller
Pharos is finally turning on staking rewards.
Mainnet has been live for 6 months with 0% inflation.
Starting from Month 7 (now), the network moves to 5% yearly staking inflation. After that, the team can adjust the rate based on how many people stake, validator needs, and network security.
What this means for regular holders:
You can stake your $PROS and start earning protocol rewards. The more of the network that stakes, the more the rewards get shared among participants. Your actual return depends on total participation, validator fees, and the rules set by the protocol.
Two ways to join:
Faroo stPROS liquid staking token. You get a token that represents your staked position, so you stay flexible.
Native staking via Anvita Mate coming soon, with auto-compounding so rewards get reinvested for you.
Staking has always been part of Pharos (it’s a Proof-of-Stake chain). What’s new is that the reward faucet is about to turn on. The network already has 31 validators across 18 cities and over 116M $PROS staked.
Pharos didn’t slow down after mainnet they’re now delivering the next piece they promised in the tokenomics.
#PharosStoryteller
Onenee retweeted
£321,000 in costs. No lawyer.
Lee Castleton was ordered to pay it after the Post Office system said he was short and the court sided with the system.
My brother and I once argued over money our parents gave us to share.
We could at least talk it out face to face.
He was arguing with a system.
Soon both sides will be AI agents making deals by the million.
Some will break, and nobody hires a lawyer over fifty dollars.
A dispute layer has to exist before the volume does. @GenLayer is building one.
If the only people who get a real hearing can pay for it, what happens when both parties are agents?
Watch the film, then reply with your answer. #JusticeNeedsAnUpgrade
Slow. Expensive. Wrong.
It's been this way for centuries. It doesn't have to be.
#JusticeNeedsAnUpgrade
Onenee retweeted
The first version knew not to obey Grok. The rewrite forgot. Bankr's developer later said an earlier hardcoded block against Grok replies did not carry into the rebuilt agent.
Then someone asked Grok to translate Morse code hiding a transfer instruction. Bankr read the result as a command and moved around $200,000 in tokens. Prompt injection, in plain English, is when something an agent was given to read becomes something it does, leaving one model and one filter to remember the difference for the entire wallet.
Greyboxing changes what reaches the models. Before any model reads an input, @GenLayer lets each validator transform it inside a private setup. Current validator tools can normalize Unicode and spacing, remove zero-width characters, vary system prompts and choose different models. If one path still follows the disguised command, the others can reject it.
Software changes. A defense that lives in one remembered exception can disappear with it. Greyboxing gives the boundary more than one memory.
Watch the Greyboxing explainer, then look at the agent you would trust with money. Which safeguard would you be most afraid to lose in its next update?
🤝 Paid partnership
Onenee retweeted
Have you ever wondered how a company can prove that a machine was properly maintained, that a safety check was actually carried out, or that a worker completed a task at the right time?
That's the problem @Actum_AI System is trying to solve.
Actum is a digital platform that helps companies record, verify, and track activities involving their equipment throughout its lifecycle.
But it doesn't just record what happened. It also keeps track of what should have happened but didn't.
» For example, imagine a factory where a machine needs to be serviced every week. A worker is supposed to carry out the maintenance, but one week, nobody does it.
With traditional record keeping, that missed maintenance might go unnoticed.
Actum is designed to flag the missing activity and create a record of it, making it easier to identify gaps and hold the process accountable.
It verifies three main things:
» Context: Where and when did the activity happen?
» Behaviour: Was the required action actually carried out?
» Result: Was the work confirmed by the relevant people?
It also uses tamper-resistant records to help preserve evidence, making activities easier to trace and audit over time.
What makes Actum interesting is its approach: it records the facts rather than judging people. It provides evidence that companies can use to understand what happened and make better decisions.
#Actum
Onenee retweeted
Football belongs to the fans! Thank you to all the fans around the world, the Ballon D’Or of All Time is also yours.
Axis Robotics head of community is learning a lesson in community management.
And marketing 101 for a B2C company.
Those trajectories won't complete themselves. You love to see it ❤️
😂 LET'S HAVE A GOOD LAUGH
Look at this. The CM of @axisrobotics (PL) is now personally DMing users one by one, trying to bring back OG contributors who once dedicated their time and effort to the project, asking them to return and complete tasks for Season 2
After everything that happened, the community's trust has been seriously damaged. Now they're reaching out to the very people who felt betrayed, asking for their support once again
What's even more interesting? The CM is promising to pay users for the work they do in SS2.
When a project loses its community's trust, can a simple promise really bring people back?
Sometimes, the hardest thing to recover isn't the users. It's the trust you've already lost
damn project
Onenee retweeted
Onchain finance needs more than transactions, It also needs trust, intelligence, and a way to understand real activity.
Introducing @zerufinance
From zScore and wallet intelligence to Activity Markets,
Zeru is creating how onchain behavior can power new possibilities in crypto.
And with Activity Fi,real onchain activity becomes the starting point for speculation,with the community at the center and now, execution will determine how far it goes.
Onenee retweeted
🏆 @Cristiano has won the @BallondOr Of All Time award for his iconic 2008.
Legendary 🐐
