
Anton
317 posts


@SolflareEmpire @KasparasF @TheRealEmko_ @LegendlarrySol Kek, you must have a low level of analysis if you hired an employee from Synternet
English


From the point of view of “legitimacy,” this is a smart move, but there is still no influx of funds. Even in such a situation, without the use of loyalty programs, pure organic growth will not help.
Api3@Api3DAO
We're happy to announce the first $EIGEN/USDC market on @Morpho on Ethereum mainnet. Believe in @eigencloud, use your $EIGEN and tap into liquidity. app.morpho.org/ethereum/marke…
English

Power the network with Energy Web ⚡
Liquid staking lets you secure the chain while being compensated in $EWT and $USDC 🟣
Bridge. Stake. Earn.
🌐 energywebx.com
English

@pravijn L2s could work too in some cases, e.g. in financial use cases it might work well, healthcare companies may not be willing to do that though for business optics reasons
English

Private, user-owned software powered by cryptographic primitives is more important than ever.
A few I'm very excited about:
- insurance markets onchain
- data marketplaces for agents
- private agents
- privacy-native L1s for specific verticals
26 ideas we want to invest in 👇
Electric Capital ⚡️@ElectricCapital
1/ We @ElectricCapital want to invest in 26 ideas that give users control, privacy, & access. Why now: •Trust in institutions is collapsing •AI centralizes power but lowers build costs •User-owned tech defends personal freedom data.electriccapital.com/posts/investin… 👇
English

@Deebs_DeFi @0xSammy In your opinion, will this algorithm change constantly or only with minor adjustments?
English

@0xSammy you should check this out about diversity score in particular
x.com/Deebs_DeFi/sta…
Deebs DeFi 🛰@Deebs_DeFi
English

X open sourced the latest algorithm
Here are 10 insights pulled directly from the GitHub:
1) Drive Multiple Engagement Types
The Phoenix transformer predicts 15+ actions including P(favorite), P(reply), P(repost), P(quote). Posts triggering multiple predictions get higher weighted scores in the final ranking formula (phoenix/ scoring)
2) Optimize for Video Watch Time
Video duration is hydrated separately in the candidate hydration stage and P(video_view) is a distinct prediction target. Completion signals quality to the ranking model (home-mixer/ hydration, phoenix/ predictions)
3) Prioritize Quote Tweets Over Retweets
P(quote) exists as a separate weighted prediction while basic reposts may score lower. Quotes add unique commentary the algorithm values (phoenix/ Weighted Scorer)
4) Post When Fresh Engagement is Likely
The AgeFilter removes posts older than threshold. Thunder’s in-memory store only retains recent posts for sub-millisecond lookups (home-mixer/ pre-scoring filters, thunder/ retention)
5) Maximize Dwell Time
P(dwell) is explicitly predicted by the transformer. Time-spent-viewing correlates with content quality and ranks accordingly (phoenix/ predictions)
6) Avoid All Negative Signals
P(block_author), P(mute_author), P(report), P(not_interested) have negative weights that push down your final score for similar users (phoenix/ Weighted Scorer)
7) Build Engaged Followers, Not Just Numbers
Phoenix’s two-tower retrieval model encodes user engagement history into embeddings. Your content surfaces to users with similar embedding vectors (phoenix/ User Tower & Candidate Tower)
8) Encourage Profile Clicks
P(profile_click) is a tracked positive signal in the engagement predictions. Bio optimization and compelling posts that spark curiosity boost this metric (phoenix/ predictions)
9) Diversify Authors in Threads
The AuthorDiversityScorer attenuates repeated author scores to prevent feed domination by one account (home-mixer/ scoring stage)
10) No Shortcuts Work
The system eliminated “every single hand-engineered feature” per the README. The Grok-based transformer learns patterns from raw engagement sequences only (phoenix/, ported from xai-org/grok-1)
Now bookmark this, click through to my profile, QT (not RT) and ensure you’re online at my peak engagement hours to optimize for velocity. Thank you.

Engineering@Engineering
We have open-sourced our new 𝕏 algorithm, powered by the same transformer architecture as xAI's Grok model. Check it out here: github.com/xai-org/x-algo…
English

@AIslop_ @0x007alpha Study the narrative of this profile. Jarvis must start applying AI slops, otherwise you won't get the coverage you need.
English

@AnisLaour @MaxxGrey101 @Api3DAO All altcoins are bad, don't fall in love with them.
I am interested in api3 from the point of view of transparency and the product. Nothing more.
English

@AFFKKT @MaxxGrey101 @Api3DAO The whole alt market is down ? You are Just a hâter i am not saying api3 is perfect i am saying you are actîng like it’s the only bad token
English

@AnisLaour @MaxxGrey101 @Api3DAO Show me the growth in the price of the api3 token over the last year for the actions that the project is taking.
I think you're an api3 fanatic. Good luck with your investments.
English

@AFFKKT @MaxxGrey101 @Api3DAO Most projects have nothing to do with their tokens lol 😂😂😂😂😂 even stocks work the same way are you serious ?
English

@AFFKKT @MaxxGrey101 @Api3DAO Lol you didn’t reply to the points i saif and pulled this clearly you know nothing about how crypto prices work
English

@AnisLaour @MaxxGrey101 @Api3DAO @AnisLaour your answer is here. We'll see each other when the token price is $0.2-0.3 and you will once again prove your sectarian point of view to me. As I wrote earlier, the project's actions do not reflect the token price.
English

@AnisLaour @MaxxGrey101 @Api3DAO You can continue to buy api3, but what they do has nothing to do with the token. As long as you keep buying, the huge inflation will continue to grow.The logic is incredibly simple, and where am I wrong?
DIA shows the best results for the year.
English

@MaxxGrey101 @Api3DAO Partly true, but DYOR. I don't think you are an active participant in Discord discussions. And if you knew what was going on in the project, you wouldn't judge so superficially.
Tokenomics is the main problem, not what they do🫡
English

It will be interesting to observe the evolution of projects, or the disregard of such sets of rules.
fairlaunch.org
English

@mattob @brian_armstrong @Revolut "hey want to hold dollars vs local currencies"
a fairly accurate observation
English

really interesting question from @brian_armstrong
personally, I would love to see the base app become “a better bank account”
right now, I use @Revolut
it’s a great product, but I earn 0% interest on the cash I have in my Revoult account
additionally, I get 0% interest on any of the stablecoins I hold (USDC / USDT) I have in my wallet
if the Base app made it easy for me to “earn in my sleep” by getting interest on my cash holdings, this would be amazing
for people in developing nations, they want to hold dollars vs local currencies, and IMO, the base app can also be their bank account
sure, it’s not as sexy as building a new social network, but IMO, this is a massive opportunity base is perfectly positioned to capture
they have a rock solid team around the world, a great leader in @jessepollak, the support of @coinbase and a desire to bring people onchain
good idea, bad idea?

English

@AmandaMashburn It is also effective to make notes about when you started reading:
Start date.
End date.
With the help of eidetic memory, it is easier to analyze the cycles of your life and the books you read at that time.
English

One major difference between physical and digital mediums – the size and format (book, magazine, newspaper, etc.) serves as a visual indicator of the time commitment. In digital, you need to replace that cue somehow. No one likes getting a few paragraphs in only to realize the article is actually a 20+ minute read. Conversely, if something takes 5 minutes to get through, I know I can go ahead and knock it off my list.
English

A request for 𝕏 articles:
Display estimated reading time (or word count) so folks can decide if they want to read the article then and there or save it for later.
@nikitabier
Amanda Mashburn 🇺🇸@AmandaMashburn
@venturetwins Reviewing articles with your (AI) editor then watching the likes and bookmarks roll in.
English








