Keep on to blur preview images; turn off to show them clearly
progress will not be slowed down but why does this still live in a traditional IDE ?
building workers observability @cloudflaredev, prev founder @baselimehq (acquired by cloudflare), prev aerodynamicist
In South Korea, Trump gets a gold crown and highest state honor
The pulse of the nation in the palm of your hand.
Why x402? The web was built for humans. It was obvious at inception of the internet & browser that browsers should embed direct payment. This was the reason for baking in HTTP 402 from the get-go. But the technology for instant micropayments did not exist at the time. (Read the Stracherty Original Sin post for better context) Instead, the free an open internet was subsidized by ads. Ads meant that there was an incentive to add information to the web, for free, and monetize through distracting the human explorer. Ironically, ads created the 5 trillion token dataset of the internet which birthed LLMs through a rich human-generated text corpus. LLMs are the downfall of the advertising business model for the open internet. When a human asks an LLM for information, or asks an LLM-based agent to search, retrieve information, and execute on human intent, the LLM cannot be distracted and its partial attention cannot be monetized. x402 falls back to a model of the internet where each request is paid directly. It imagines an internet with a different and more human-aligned business model.
CEO @merit_systems -- monetization tools for developers. Retired zkVM guy @a16z
BTS leader RM addresses the APEC CEO Summit, advocating for cultural diversity and creative collaboration amid regional trade concerns
Top and breaking news, pictures and videos from Reuters. For breaking business news, follow @ReutersBiz. Our daily podcast is here: https://t.co/KO0QFy0d3a
Tbc, I think we fix these incentives by making it easier to build here, not bullying investors to change how they value companies. Let the markets figure it out. Supply-side reform, not new valuation engineering.
investing @a16z // curating https://t.co/ssslqn6eo7
new work together with @Jsevillamol and @EpochAIResearch how decentralized can large pretraining runs get, and how much would it cost? we model a hypothetical 10GW decentralized cluster, and find that the additional capex of networking is less than 1% of the total budget
@Jsevillamol @EpochAIResearch why do this? a single monolithic cluster is usually preferable, not least because of the added engineering complexity of decentralized training. but at 10GW, electric power constraints bite even more. networking geographically distributed sites can help!