Web3’s real ‘TCP/IP moment’ hasn’t happened yet

The web scaled as a result of IP created a common material for information. Web3 by no means obtained that luxurious. As a substitute, it inherited Nineteen Eighties-era networking and a patchwork of ad-hoc protocols that decelerate and congest the second you attempt to run actual transactions at scale, not to mention billions of AI brokers, world settlement layers, or a planetary-scale decentralized bodily infrastructure community sensor mesh. We’re long gone the purpose the place sooner chains or greater blocks may help.

Web3 wants its personal TCP/IP second: a decentralized Web Protocol constructed on the ideas that made the unique web unstoppable, however engineered to protect what makes blockchain matter: trustlessness, censorship resistance, and permissionless participation that lastly performs at scale.

You may additionally like: The 1984 Processor Downside: Web3 scaling calls for P2P clearing, not greater blockchains | Opinion

What the trade retains lacking

Earlier than IP, computer systems couldn’t discuss throughout networks. IP created a common commonplace for routing information between any two factors on earth, turning remoted programs into the web. It turned one in every of three pillars of web infrastructure (alongside compute and storage). Each web2 software runs on TCP/IP. It’s the protocol that made planetary-scale communication potential.

Web3 is repeating the identical early errors. Each blockchain invented its personal networking layer, together with gossip protocols, Turbine, Snow, Narwhal, mempools, and DA sampling. None of them is common, and so they’re needlessly restrictive. Everybody’s chasing pace with greater blocks, extra rollups, extra parallelization. However they’re all utilizing basically damaged networking fashions.

If we’re severe about scaling web3, we want a reliably quick, trustless, fault-tolerant, and most significantly, modular web protocol.

20 years at MIT, fixing decentralization’s hardest downside

For over twenty years, my analysis at MIT has targeted on one query: Can decentralized programs transfer data as quick and reliably as centralized ones — and may we make it mathematically provable?

To reply that, we mixed two fields that had not often intersected: community coding concept, which mathematically optimizes information motion, and distributed algorithms, led by Nancy Lynch’s seminal work on consensus and Byzantine fault tolerance.

What we discovered was clear: decentralized programs can attain centralized-level efficiency — however provided that we redesign information motion from first ideas. After years of proofs and experiments, Random Linear Community Coding (RLNC) emerged because the mathematically optimum technique for doing this throughout decentralized networks.

As soon as blockchains arrived, the appliance turned apparent. The web we’ve was constructed for trusted intermediaries. The decentralized internet wants its personal protocol: one designed to resist failure and assault whereas scaling globally. The architectural shift is such that:

• efficiency comes from arithmetic, not {hardware};
• coordination comes from code, not servers;
• and the community turns into stronger because it decentralizes.

Like the unique Web Protocol, it isn’t meant to switch what exists, however to allow what comes subsequent.

The use instances that break right now’s infrastructure

Decentralized programs are hitting their limits on the actual second the world wants them to scale. 4 macro tendencies are rising — and every exposes the identical bottleneck: Web3 nonetheless runs on networking assumptions inherited from centralized programs.

1. The fragmentation of L1s and L2s means blockchains scale regionally, however fail globally

We now have greater than 100 blockchains, and whereas every can optimize its personal native execution, the second these networks must coordinate globally, all of them hit the identical challenges: information motion is restricted, inefficient, and basically sub-optimal.

What blockchains lack is the equal of an electrical grid, a shared layer that routes bandwidth wherever it’s wanted. A decentralized Web Protocol would give each chain entry to the identical coded information material, accelerating block propagation, DA retrieval, and state entry with out touching consensus. And like all good grid, when it really works, congestion is minimized.

2. Tokenization & DeFi at trillion-dollar markets

DeFi can not settle trillions on networks the place propagation is gradual, it collapses underneath load, or the place RPC bottlenecks centralize entry. If a number of chains have been linked by a shared coded community, propagation spikes would seemingly not overwhelm any single chain — they might be absorbed and redistributed throughout the whole community.

In conventional programs, you construct bigger information facilities to soak up peak load. These are costly and result in single factors of failure. In decentralized programs, we can not depend on megacenters; we should depend on coded distribution.

3. DePIN at world scale

A world community with tens of millions of gadgets and autonomous machines can not operate if every node waits on gradual, single-path communication. These gadgets should behave like a single, coherent organism.

In power programs, versatile grids take up each business mining operations and a single hair dryer. In networking, a decentralized protocol should do the identical for information: take up each supply optimally, and ship it the place it’s wanted most. That requires coded storage, coded retrieval, and the power to make use of each obtainable path somewhat than counting on just a few predetermined ones.

4. Decentralized AI

Distributed AI, whether or not coaching on encrypted fragments or coordinating fleets of AI brokers, will depend on high-throughput, fault-tolerant information motion. In the present day, decentralized storage and compute are separated; entry is gradual; retrieval will depend on centralized gateways. What AI wants is information logistics, not easy storage: that means that information is encoded whereas in movement, saved in coded fragments, retrieved from wherever is quickest on the time, and recombined immediately with out relying on any single location.

Web3’s subsequent leap

Each main leap within the web’s evolution started with a breakthrough in how information strikes. IP delivered world connectivity. Broadband enabled Netflix and cloud computing. 4G and 5G made Uber, TikTok, and real-time social potential. GPUs sparked the deep studying revolution. Good contracts unlocked programmable finance.

A common, coded information layer would do for blockchains what IP did for the early web: create the circumstances for functions we will’t but think about. It’s the inspiration that transforms Web3 from experimental to inevitable.

Learn extra: Web3 has a gatekeeping downside | Opinion